Remove duplicate contents in gcc/config/rs6000/t-xilinx
[official-gcc.git] / gcc / ChangeLog
blobbd4715e2c421f00e0d87699afb855a666e8ead25
1 2013-11-08  Tom de Vries  <tom@codesourcery.com>
3         * config/rs6000/t-xilinx: Remove duplicate contents.
5 2013-11-07  Andrew MacLeod  <amacleod@redhat.com>
6             Joseph Myers  <joseph@codesourcery.com>
8         * tree-core.h (enum cv_qualifier): Add TYPE_QUAL_ATOMIC.
9         (enum tree_index): Add TI_ATOMICQI_TYPE, TI_ATOMICHI_TYPE,
10         TI_ATOMICSI_TYPE, TI_ATOMICDI_TYPE and TI_ATOMICTI_TYPE.
11         (struct tree_base): Add atomic_flag field.
12         * tree.h (TYPE_ATOMIC): New accessor macro.
13         (TYPE_QUALS, TYPE_QUALS_NO_ADDR_SPACE): Add TYPE_QUAL_ATOMIC.
14         (TYPE_QUALS_NO_ADDR_SPACE_NO_ATOMIC): New macro.
15         (atomicQI_type_node, atomicHI_type_node, atomicSI_type_node)
16         (atomicDI_type_node, atomicTI_type_node): New macros for type
17         nodes.
18         * tree.c (set_type_quals): Set TYPE_ATOMIC.
19         (find_atomic_core_type): New function.
20         (build_qualified_type): Adjust alignment for qualified types.
21         (build_atomic_base): New function
22         (build_common_tree_nodes): Build atomicQI_type_node,
23         atomicHI_type_node, atomicSI_type_node, atomicDI_type_node and
24         atomicTI_type_node.
25         * print-tree.c (print_node): Print atomic qualifier.
26         * tree-pretty-print.c (dump_generic_node): Print atomic type
27         attribute.
28         * target.def (atomic_assign_expand_fenv): New hook.
29         * doc/tm.texi.in (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): New @hook.
30         * doc/tm.texi: Regenerate.
31         * targhooks.c (default_atomic_assign_expand_fenv): New function.
32         * targhooks.h (default_atomic_assign_expand_fenv): Declare.
33         * sync-builtins.def (__atomic_feraiseexcept): New built-in
34         function.
35         * config/i386/i386-builtin-types.def (VOID_FTYPE_PUSHORT): New
36         function type.
37         * config/i386/i386.c (enum ix86_builtins): Add
38         IX86_BUILTIN_FNSTENV, IX86_BUILTIN_FLDENV, IX86_BUILTIN_FNSTSW and
39         IX86_BUILTIN_FNCLEX.
40         (bdesc_special_args): Add __builtin_ia32_fnstenv,
41         __builtin_ia32_fldenv, __builtin_ia32_fnstsw and
42         __builtin_ia32_fnclex.
43         (ix86_expand_builtin): Handle the new built-in functions.
44         (ix86_atomic_assign_expand_fenv): New function.
45         (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): New macro.
46         * config/i386/i386.md (UNSPECV_FNSTENV, UNSPECV_FLDENV)
47         (UNSPECV_FNSTSW, UNSPECV_FNCLEX): New unspecs.
48         (fnstenv, fldenv, fnstsw, fnclex): New insns.
50 2013-11-07  Steve Ellcey  <sellcey@mips.com>
52         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add fp64 directory.
53         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add -mfp64 flag.
54         (MULTILIB_DIRNAMES): Add fp64 directory.
55         (MULTILIB_EXCEPTIONS): Add new exclusions.
57 2013-11-07  Aldy Hernandez  <aldyh@redhat.com>
59         * gimplify.c (gimple_regimplify_operands): Do not set
60         SSA_NAME_DEF_STMT.
61         * graphite-sese-to-poly.c (remove_simple_copy_phi): Same.
62         (rewrite_close_phi_out_of_ssa): Same.
63         (rewrite_phi_out_of_ssa): Same.
64         (rewrite_degenerate_phi): Same.
65         (handle_scalar_deps_crossing_scop_limits): Same.
66         * tree-if-conv.c (predicate_scalar_phi): Same.
67         * tree-parloops.c (create_loads_for_reductions): Same.
68         (create_final_loads_for_reduction): Same.
69         (create_loads_and_stores_for_name): Same.
70         (transform_to_exit_first_loop): Same.
71         (create_parallel_loop): Same.
72         * tree-ssa-loop-im.c
73         (move_computations_dom_walker::before_dom_children): Same.
74         * tree-ssa-loop-manip.c (rewrite_phi_with_iv): Same.
75         * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
76         Same.
77         * tree-ssa-propagate.c (substitute_and_fold): Same.
78         * tree-vect-loop.c (vect_finalize_reduction): Same.
79         * tree-vect-stmts.c (vectorizable_call): Same.
81 2013-11-07  Mike Stump  <mikestump@comcast.net>
83         * config/pdp11/pdp11.c: Include dbxout.h.
84         * config/picochip/picochip.c: Likewise.
86 2013-11-07  Cong Hou  <congh@google.com>
88         PR tree-optimization/56764
89         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
90         Combine alias checks if it is possible to amortize the runtime
91         overhead.  Return the number of alias checks after merging.
92         * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
93         Use the function vect_create_cond_for_alias_checks () to check
94         the number of alias checks.
96 2013-11-07  Jeff Law  <law@redhat.com>
98         * varpool.c (ctor_for_folding): Fix typo in comment.
100 2013-11-07  Joern Rennecke  <joern.rennecke@embecosm.com>
102         * config/arc/arc.c (arc_ifcvt): Use commutativity, e.g.:
103         reg_a := reg_b + reg_a  ==>  reg_a := reg_a + reg_b
105 2013-11-07  Jeff Law  <law@redhat.com>
107         * doc/invoke.texi (-fisolate-erroneous-paths): Document.
109         * gimple-ssa-isolate-paths.c (gate_isolate_erroneous_paths):
110         No longer check if we have __builtin_trap, assume it's
111         available.
113 2013-11-07  Thomas Schwinge  <thomas@codesourcery.com>
115         * doc/sourcebuild.texi (Top Level) <lto-plugin>: GNU ld can use
116         linker plugins, too.
118         * config/arc/arc.h (LINK_COMMAND_SPEC): For
119         -ftree-parallelize-loops=*, link to libgomp and its dependencies.
120         * config/ia64/hpux.h (LIB_SPEC): Likewise.
121         * config/pa/pa-hpux11.h (LIB_SPEC): Likewise.
122         * config/pa/pa64-hpux.h (LIB_SPEC): Likewise.
123         * gcc.c (GOMP_SELF_SPECS): Update comment about libgomp's
124         dependencies.
126 2013-11-07  Jakub Jelinek  <jakub@redhat.com>
128         * tree-ssa-loop-niter.c: Include tree-ssanames.h.
129         (determine_value_range): Add loop argument.  Use get_range_info to
130         improve range.
131         (bound_difference): Adjust caller.
133 2013-11-07  Richard Biener  <rguenther@suse.de>
134             Jakub Jelinek  <jakub@redhat.com>
136         * tree-vrp.c (find_assert_locations): Pre-seed live bitmaps for loop
137         latches from header PHI arguments from the latch edge.
139 2013-11-07  Paolo Carlini  <paolo.carlini@oracle.com>
141         PR c++/58176
142         * varasm.c (output_constant): Handle NULLPTR_TYPE.
144 2013-11-07  H.J. Lu  <hongjiu.lu@intel.com>
146         * config/i386/i386.c (ix86_expand_set_or_movmem): Don't set
147         misaligned_prologue_used when it has been set.
149 2013-11-07  Yury Gribov  <y.gribov@samsung.com>
150             Jakub Jelinek  <jakub@redhat.com>
152         PR sanitizer/59029
153         * asan.c (get_mem_refs_of_builtin_call): Allow
154         integer literals as addresses in instrumented builtins.
156 2013-11-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
158         * config/aarch64/aarch64.c (aarch64_legitimize_reload_address):
159         Explain why plus_constant is not used.
161 2013-11-07  Richard Biener  <rguenther@suse.de>
163         * tree-ssa-ccp.c (canonicalize_float_value): Rename to ...
164         (canonicalize_value): ... this.  Also handle stripping of
165         TREE_OVERFLOW.
166         (get_value, set_lattice_value, get_value_for_expr): Adjust.
167         * gimple-fold.c (canonicalize_constructor_val): Strip
168         TREE_OVERFLOW.
169         * tree-ssa-threadedge.c (set_ssa_name_value): Likewise.
171 2013-11-07  Richard Biener  <rguenther@suse.de>
173         * tree-dfa.c (get_ref_base_and_extent): Fix casting.
175 2013-11-07  H.J. Lu  <hongjiu.lu@intel.com>
177         PR target/59034
178         * config/i386/i386.md (push peepholer/splitter): Use Pmode
179         with stack_pointer_rtx.
181 2013-11-07  Bin Cheng  <bin.cheng@arm.com>
183         * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Check equality
184         using operand_equal_p.
186 2013-11-07  Bin Cheng  <bin.cheng@arm.com>
188         * tree-ssa-loop-ivopts.c (alloc_iv): Lower address expressions.
189         * tree-affine.c (get_inner_reference_aff): Return base.
190         * tree-affine.h (get_inner_reference_aff): Change prototype.
192 2013-11-06  Tobias Burnus  <burnus@net-b.de>
194         * doc/invoke.texi (Wdate-time): Fix typo.
196 2013-11-06  Oleg Endo  <olegendo@gcc.gnu.org>
198         * config/sh/sh.md (addsf3, divsf3, divsf3_i, rsqrtsf2, cmpgtdf_t,
199         cmpeqdf_t, *ieee_ccmpeqdf_t, negdf2, sqrtdf2, absdf2): Use
200         fp_arith_reg_operand instead of arith_reg_operand.
202 2013-11-06  Oleg Endo  <olegendo@gcc.gnu.org>
204         * config/sh/sh.md (adddi3): Remove empty constraints.
205         Remove can_create_pseudo_p and arith_reg_operand check.
206         (adddi3_compact, subdi3_compact, *negdi2): Remove constraints.
207         Split before reload.
209 2013-11-06  Jeff Law  <law@redhat.com>
210             Tom Tromey  <tromey@redhat.com>
212         * gdbinit.in: Disable strict type checking.
214 2013-11-06  Vladimir Makarov  <vmakarov@redhat.com>
216         * tree-pass.h (make_pass_live_range_shrinkage): New external.
217         * timevar.def (TV_LIVE_RANGE_SHRINKAGE): New.
218         * sched-rgn.c (gate_handle_live_range_shrinkage): New.
219         (rest_of_handle_live_range_shrinkage): Ditto
220         (class pass_live_range_shrinkage): Ditto.
221         (pass_data_live_range_shrinkage): Ditto.
222         (make_pass_live_range_shrinkage): Ditto.
223         * sched-int.h (initialize_live_range_shrinkage): New prototype.
224         (finish_live_range_shrinkage): Ditto.
225         * sched-deps.c (create_insn_reg_set): Make void return value.
226         * passes.def: Add pass_live_range_shrinkage.
227         * ira.c (update_equiv_regs): Don't move if
228         flag_live_range_shrinkage.
229         * haifa-sched.c (live_range_shrinkage_p): New.
230         (initialize_live_range_shrinkage, finish_live_range_shrinkage):
231         New functions.
232         (rank_for_schedule): Add code for pressure relief through live
233         range shrinkage.
234         (schedule_insn): Print more debug info.
235         (sched_init): Setup SCHED_PRESSURE_WEIGHTED for pressure relief
236         through live range shrinkage.
237         * doc/invoke.texi (-flive-range-shrinkage): New.
238         * common.opt (flive-range-shrinkage): New.
240 2013-11-06  Uros Bizjak  <ubizjak@gmail.com>
242         PR target/59021
243         * config/i386/i386.c (ix86_avx_u128_mode_needed): Require
244         AVX_U128_DIRTY mode for call_insn RTXes that use AVX256 registers.
245         (ix86_avx_u128_mode_needed): Return AVX_U128_DIRTY mode for call_insn
246         RTXes that return in AVX256 register.
248 2013-11-06  Richard Biener  <rguenther@suse.de>
250         PR tree-optimization/58653
251         * tree-predcom.c (ref_at_iteration): Rewrite to generate a MEM_REF.
252         (prepare_initializers_chain): Adjust.
254 2013-11-06  Andrew MacLeod  <amacleod@redhat.com>
256         * gimple.h (block_in_transaction): Move to basic-block.h and rename.
257         (gimple_in_transaction): Use bb_in_transaction.
258         * basic-block.h (bb_in_transaction): Relocate here and rename.
259         * tree-ssa-loop-im.c (execute_sm): Use bb_in_transaction.
261 2013-11-06  Richard Biener  <rguenther@suse.de>
263         * tree.c (drop_tree_overflow): New function.
264         * tree.h (drop_tree_overflow): Declare.
265         * gimplify.c (gimplify_expr): Drop TREE_OVERFLOW.
266         * tree-vrp.c (range_int_cst_singleton_p): Use
267         is_overflow_infinity instead of testing TREE_OVERFLOW.
268         (extract_range_from_assert): Likewise.
269         (zero_nonzero_bits_from_vr): Likewise.
270         (extract_range_basic): Likewise.
271         (register_new_assert_for): Use drop_tree_overflow.
272         (vrp_visit_phi_node): Likewise.
274 2013-11-06  Eric Botcazou  <ebotcazou@adacore.com>
276         * config/i386/i386.c (ix86_expand_prologue): Optimize stack
277         checking for leaf functions without dynamic stack allocation.
278         * config/ia64/ia64.c (ia64_emit_probe_stack_range): Adjust.
279         (ia64_expand_prologue): Likewise.
280         * config/mips/mips.c (mips_expand_prologue): Likewise.
281         * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
282         * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
283         (sparc_flat_expand_prologue): Likewise.
285 2013-11-06  James Greenhalgh  <james.greenhalgh@arm.com>
287         * config/aarch64/arm_neon.h (__ST2_LANE_FUNC): Better model data size.
288         (__ST3_LANE_FUNC): Likewise.
289         (__ST4_LANE_FUNC): Likewise.
291 2013-11-06  Nick Clifton  <nickc@redhat.com>
293         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define the
294         name returned by msp430_mcu_name.
295         (LIB_SPEC): If a -T option has not been specified then set a
296         default, mcu-specific, linker script.
297         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more mcu names.
298         * config/msp430/msp430.c (msp430x_names): Likewise.
299         Alpha sort the names for ease of comparison.
300         (msp430_mcu_name): New function:  Returns a string suitable for
301         use as a C preprocessor symbol based upon the name of the MCU
302         being targeted.
303         (msp430_option_override): Accept msp430x and msp430xv2 as generic
304         mcu names.
305         * config/msp430/msp430-protos.h (msp430_mcu_name): Prototype.
307         * gcc.c (do_spec_1): Do not insert a space after a %* substitution
308         unless it is the last part of a spec substring.
309         * doc/invoke.texi (Spec Files): Document space insertion
310         behaviour of %*.
312 2013-11-06  Christian Bruel  <christian.bruel@st.com>
314         * config/sh/sh-mem.cc (sh_expand_cmpnstr, sh_expand_cmpstr):
315         Factorize probabilities, Use adjust_address instead of
316         adjust_automodify_address when possible. Enable for optimize.
317         (sh_expand_strlen): New function.
318         * config/sh/sh-protos.h (sh_expand_strlen): Declare.
319         * config/sh/sh.md (strlensi): New pattern.
320         (UNSPEC_BUILTIN_STRLEN): Define.
322 2013-11-06  Jakub Jelinek  <jakub@redhat.com>
324         PR middle-end/58970
325         * expr.c (get_bit_range): Handle *offset == NULL_TREE.
326         (expand_assignment): If *bitpos is negative, set *offset
327         and adjust *bitpos, so that it is not negative.
329 2013-11-06  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
331         * config/i386/bdver3.md : Added two additional decoder units
332         to support issue rate of 4 and remodeled vector unit.
333         * config/i386/i386.c (ix86_issue_rate): Issue rate for BD
334         architectures is set to 4.
335         * config/i386/i386.c (ia32_multipass_dfa_lookahead): DFA
336         lookahead is set to 4 for BD architectures.
338 2013-11-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
340         * config/rs6000/rs6000.c (rs6000_option_override_internal):
341         Remove restriction against use of VSX instructions when generating
342         code for little endian mode.
344 2013-11-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
346         * config/rs6000/altivec.md (mulv4si3): Ensure we generate vmulouh
347         for both big and little endian.
348         (mulv8hi3): Swap input operands for merge high and merge low
349         instructions for little endian.
351 2013-11-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
353         * config/rs6000/altivec.md (vec_widen_umult_even_v16qi): Change
354         define_insn to define_expand that uses even patterns for big
355         endian and odd patterns for little endian.
356         (vec_widen_smult_even_v16qi): Likewise.
357         (vec_widen_umult_even_v8hi): Likewise.
358         (vec_widen_smult_even_v8hi): Likewise.
359         (vec_widen_umult_odd_v16qi): Likewise.
360         (vec_widen_smult_odd_v16qi): Likewise.
361         (vec_widen_umult_odd_v8hi): Likewise.
362         (vec_widen_smult_odd_v8hi): Likewise.
363         (altivec_vmuleub): New define_insn.
364         (altivec_vmuloub): Likewise.
365         (altivec_vmulesb): Likewise.
366         (altivec_vmulosb): Likewise.
367         (altivec_vmuleuh): Likewise.
368         (altivec_vmulouh): Likewise.
369         (altivec_vmulesh): Likewise.
370         (altivec_vmulosh): Likewise.
372 2013-11-05  Mike Stump  <mikestump@comcast.net>
374         * Makefile.in (mostlyclean): Remove c-family objects.
376 2013-11-05  Ian Lance Taylor  <iant@google.com>
378         * config/i386/sync.md (atomic_compare_and_swap<dwi>_doubleword):
379         If possible, add .cfi directives to record change to bx.
380         * config/i386/i386.c (ix86_emit_cfi): New function.
381         * config/i386/i386-protos.h (ix86_emit_cfi): Declare.
383 2013-11-05  Steven Bosscher  <steven@gcc.gnu.org>
385         * rtlanal.c (tablejump_p): Expect a JUMP_TABLE_DATA to always follow
386         immediately after a label for a tablejump pattern.
388         * config/arm/arm.c (is_jump_table): Remove.
389         (create_fix_barrier): Use tablejump_p instead.
390         (arm_reorg): Likewise.
391         (thumb1_output_casesi): Expect JUMP_TABLE_DATA to always be NEXT_INSN.
392         (thumb2_output_casesi): Likewise.
393         * config/aarch64/aarch64.c (aarch64_output_casesi): Likewise.
394         * config/sh/sh.md (casesi_worker_1, casesi_worker_2,
395         casesi_shift_media, casesi_load_media): Likewise.
396         * config/iq2000/iq2000.md: Likewise (in anonymous define_insn).
397         * config/microblaze/microblaze.md: Likewise.
399 2013-11-05  Tobias Burnus  <burnus@net-b.de>
401         * doc/invoke.texi (-Wdate-time): Document.
403 2013-11-05  Richard Sandiford  <rdsandiford@googlemail.com>
405         * double-int.c (lshift_double, rshift_double): Remove
406         SHIFT_COUNT_TRUNCATED handling.
408 2013-11-05  Jeff Law  <law@redhat.com>
410         * Makefile.in (OBJS): Add gimple-ssa-isolate-paths.o
411         * common.opt (-fisolate-erroneous-paths): Add option and documentation.
412         * gimple-ssa-isolate-paths.c: New file.
413         * gimple.c (check_loadstore): New function.
414         (infer_nonnull_range): Moved into gimple.c from tree-vrp.c
415         Verify OP is in the argument list and the argument corresponding
416         to OP is a pointer type.  Use operand_equal_p rather than
417         pointer equality when testing if OP is on the nonnull list.
418         Use check_loadstore rather than count_ptr_derefs.  Handle
419         GIMPLE_RETURN statements.
420         * tree-vrp.c (infer_nonnull_range): Remove.
421         * gimple.h (infer_nonnull_range): Declare.
422         * opts.c (default_options_table): Add OPT_fisolate_erroneous_paths.
423         * passes.def: Add pass_isolate_erroneous_paths.
424         * timevar.def (TV_ISOLATE_ERRONEOUS_PATHS): New timevar.
425         * tree-pass.h (make_pass_isolate_erroneous_paths): Declare.
426         * tree-ssa.c (struct count_ptr_d): Remove.
427         (count_ptr_derefs, count_uses_and_derefs): Remove.
428         * tree-ssa.h (count_uses_and_derefs): Remove.
430 2013-11-05  Jakub Jelinek  <jakub@redhat.com>
432         PR rtl-optimization/58997
433         * loop-iv.c (iv_subreg): For IV_UNKNOWN_EXTEND, expect
434         get_iv_value to be in iv->mode rather than iv->extend_mode.
435         (iv_extend): Likewise.  Otherwise, if iv->extend != extend,
436         use lowpart_subreg on get_iv_value before calling simplify_gen_unary.
437         * loop-unswitch.c (may_unswitch_on): Make sure op[i] is in the right
438         mode.
440 2013-11-05  Andrew MacLeod  <amacleod@redhat.com>
442         * gimple.h: Move some prototypes to gimple-expr.h and add to include
443         list.
444         (extract_ops_from_tree, gimple_call_addr_fndecl, is_gimple_reg_type):
445         Move to gimple-expr.h.
446         * gimple-expr.h: New file.  Relocate some prototypes from gimple.h.
447         (types_compatible_p, is_gimple_reg_type, is_gimple_variable,
448         is_gimple_id, virtual_operand_p, is_gimple_addressable,
449         is_gimple_constant, extract_ops_from_tree, gimple_call_addr_fndecl):
450         Relocate here.
451         * gimple.c (extract_ops_from_tree_1, gimple_cond_get_ops_from_tree,
452         gimple_set_body, gimple_body, gimple_has_body_p, is_gimple_lvalue,
453         is_gimple_condexpr, is_gimple_addressable, is_gimple_constant,
454         is_gimple_address, is_gimple_invariant_address,
455         is_gimple_ip_invariant_address, is_gimple_min_invariant,
456         is_gimple_ip_invariant, is_gimple_variable, is_gimple_id,
457         virtual_operand_p, is_gimple_reg, is_gimple_val, is_gimple_asm_val,
458         is_gimple_min_lval, is_gimple_call_addr, is_gimple_mem_ref_addr,
459         gimple_decl_printable_name, useless_type_conversion_p,
460         types_compatible_p, gimple_can_coalesce_p, copy_var_decl): Move to
461         gimple-expr.[ch].
462         * gimple-expr.c: New File.
463         (useless_type_conversion_p, gimple_set_body, gimple_body,
464         gimple_has_body_p, gimple_decl_printable_name, copy_var_decl,
465         gimple_can_coalesce_p, extract_ops_from_tree_1,
466         gimple_cond_get_ops_from_tree, is_gimple_lvalue, is_gimple_condexpr,
467         is_gimple_address, is_gimple_invariant_address,
468         is_gimple_ip_invariant_address, is_gimple_min_invariant,
469         is_gimple_ip_invariant, is_gimple_reg, is_gimple_val,
470         is_gimple_asm_val, is_gimple_min_lval, is_gimple_call_addr,
471         is_gimple_mem_ref_addr): Relocate here.
472         * Makefile.in (OBJS): Add gimple-expr.o.
474 2013-11-05  David Malcolm  <dmalcolm@redhat.com>
476         * gengtype-parse.c (struct_field_seq): Support empty structs.
478 2013-11-05  Uros Bizjak  <ubizjak@gmail.com>
480         * config/i386/t-rtems (MULTILIB_MATCHES): Fix option typos.
482 2013-11-05  Uros Bizjak  <ubizjak@gmail.com>
484         * config/i386/i386-c.c (ix86_target_macros): Define _SOFT_FLOAT
485         for !TARGET_80387.
486         * config/i386/rtemself.h (TARGET_OS_CPP_BUILTINS): Do not define
487         _SOFT_FLOAT here.
488         (LONG_DOUBLE_TYPE_SIZE): New define.
489         (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Ditto.
491 2013-11-05  Paolo Carlini  <paolo.carlini@oracle.com>
493         PR c++/58724
494         * doc/extend.texi [visibility ("visibility_type")]: Add example
495         about visibility attribute on namespace declaration.
497 2013-11-05  Richard Biener  <rguenther@suse.de>
499         PR ipa/58492
500         * passes.def (all_passes): Start with pass_fixup_cfg again.
502 2013-11-05  Richard Biener  <rguenther@suse.de>
504         PR tree-optimization/58955
505         * tree-loop-distribution.c (pg_add_dependence_edges): Fix
506         edge direction.
508 2013-11-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
510         * config/rs6000/vector.md (vec_pack_sfix_trunc_v2df): Adjust for
511         little endian.
512         (vec_pack_ufix_trunc_v2df): Likewise.
514 2013-11-05  H.J. Lu  <hongjiu.lu@intel.com>
516         PR middle-end/58981
517         * doc/md.texi (@code{movmem@var{m}}): Specify Pmode as mode of
518         pattern, instead of word_mode.
520         * expr.c (emit_block_move_via_movmem): Don't use mode wider than
521         Pmode for size.
522         (set_storage_via_setmem): Likewise.
524 2013-11-05  Andrew MacLeod  <amacleod@redhat.com>
526         * tree-outof-ssa.c (queue_phi_copy_p): Combine phi_ssa_name_p from
527         gimple.h and the rest of the condition in eliminate_build.
528         (eliminate_build): Call new routine.
529         * gimple.h (phi_ssa_name_p): Delete.
531 2013-11-05  Trevor Saunders  <tsaunders@mozilla.com>
533         * vec.c (vec_prefix::calculate_allocation): Don't try to handle the
534         case of no prefix and reserving zero slots, because when that's the
535         case we'll never get here.
536         * vec.h (va_heap::reserve): Don't try and handle
537         vec_prefix::calculate_allocation returning zero because that should
538         never happen.
540 2013-11-05  Richard Biener  <rguenther@suse.de>
542         PR middle-end/58941
543         * tree-dfa.c (get_ref_base_and_extent): Merge common code
544         in MEM_REF and TARGET_MEM_REF handling.  Make sure to
545         process trailing array detection before diving into the
546         view-converted object (and possibly apply some extra offset).
548 2013-11-05  Joseph Myers  <joseph@codesourcery.com>
550         * config/i386/i386.c (ix86_float_exceptions_rounding_supported_p):
551         New function.
552         (TARGET_FLOAT_EXCEPTIONS_ROUNDING_SUPPORTED_P): Define.
554 2013-11-05  Marc Glisse  <marc.glisse@inria.fr>
556         PR tree-optimization/58958
557         * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Use
558         get_addr_base_and_unit_offset instead of get_ref_base_and_extent.
560 2013-11-05  Marc Glisse  <marc.glisse@inria.fr>
562         * tree-ssa-alias.h (ranges_overlap_p): Handle negative offsets.
563         * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
565 2013-11-05  Jakub Jelinek  <jakub@redhat.com>
567         PR tree-optimization/58984
568         * ipa-prop.c (ipa_load_from_parm_agg_1): Add SIZE_P argument,
569         set *SIZE_P if non-NULL on success.
570         (ipa_load_from_parm_agg, ipa_analyze_indirect_call_uses): Adjust
571         callers.
572         (ipcp_transform_function): Likewise.  Punt if size of access
573         is different from TYPE_SIZE on v->value's type.
575 2013-11-05  Tobias Burnus  <burnus@net-b.de>
577         * doc/invoke.texi (-fopenmp-simd): Document new option.
578         * gimplify.c (gimplify_body): Accept -fopenmp-simd.
579         * omp-low.c (execute_expand_omp, execute_lower_omp): Ditto.
580         * tree.c (attribute_value_equal): Ditto.
582 2013-11-04  Wei Mi  <wmi@google.com>
584         * sched-rgn.c (add_branch_dependences): Keep insns in
585         a SCHED_GROUP at the end of BB to remain their location.
587 2013-11-04  Wei Mi  <wmi@google.com>
589         * config/i386/i386.c (memory_address_length): Extract a part
590         of code to rip_relative_addr_p.
591         (rip_relative_addr_p): New Function.
592         (ix86_macro_fusion_p): Ditto.
593         (ix86_macro_fusion_pair_p): Ditto.
594         * config/i386/i386.h: Add new tune features about macro-fusion.
595         * config/i386/x86-tune.def (DEF_TUNE): Ditto.
596         * doc/tm.texi: Generated.
597         * doc/tm.texi.in: Ditto.
598         * haifa-sched.c (try_group_insn): New Function.
599         (group_insns_for_macro_fusion): Ditto.
600         (sched_init): Call group_insns_for_macro_fusion.
601         * target.def: Add two hooks: macro_fusion_p and
602         macro_fusion_pair_p.
604 2013-11-04  Kostya Serebryany  <kcc@google.com>
606         Update to match the changed asan API.
607         * asan.c (asan_function_start): New function.
608         (asan_emit_stack_protection): Update the string stored in the
609         stack red zone to match new API.  Store the PC of the current
610         function in the red zone.
611         (asan_global_struct): Update the __asan_global definition to match
612         the new API.
613         (asan_add_global): Ditto.
614         * asan.h (asan_function_start): New prototype.
615         * final.c (final_start_function): Call asan_function_start.
616         * sanitizer.def (BUILT_IN_ASAN_INIT): Rename __asan_init_v1
617         to __asan_init_v3.
619 2013-11-04  Wei Mi  <wmi@google.com>
621         * config/i386/i386-c.c (ix86_target_macros_internal): Separate
622         PROCESSOR_COREI7_AVX out from PROCESSOR_COREI7.
623         * config/i386/i386.c (ix86_option_override_internal): Ditto.
624         (ix86_issue_rate): Ditto.
625         (ix86_adjust_cost): Ditto.
626         (ia32_multipass_dfa_lookahead): Ditto.
627         (ix86_sched_init_global): Ditto.
628         (get_builtin_code_for_version): Ditto.
629         * config/i386/i386.h (enum target_cpu_default): Ditto.
630         (enum processor_type): Ditto.
631         * config/i386/x86-tune.def (DEF_TUNE): Ditto.
633 2013-11-04  Vladimir Makarov  <vmakarov@redhat.com>
635         PR rtl-optimization/58967
636         * config/rs6000/rs6000.c (legitimate_lo_sum_address_p): Remove
637         !lra_in_progress for mode sizes bigger word.
639 2013-11-04  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
641         * config/rs6000/altivec.md (vec_widen_umult_hi_v16qi): Swap
642         arguments to merge instruction for little endian.
643         (vec_widen_umult_lo_v16qi): Likewise.
644         (vec_widen_smult_hi_v16qi): Likewise.
645         (vec_widen_smult_lo_v16qi): Likewise.
646         (vec_widen_umult_hi_v8hi): Likewise.
647         (vec_widen_umult_lo_v8hi): Likewise.
648         (vec_widen_smult_hi_v8hi): Likewise.
649         (vec_widen_smult_lo_v8hi): Likewise.
651 2013-11-04  Ian Lance Taylor  <iant@google.com>
653         * builtins.def (ATTR_NOTHROWCALL_LEAF_LIST): Define.
654         * sync-builtins.def: Use ATTR_NOTHROWCALL_LEAF_LIST for all sync
655         builtins that take pointers.
656         * lto-opts.c (lto_write_options): Write -fnon-call-exceptions if set.
657         * lto-wrapper.c (merge_and_complain): Collect OPT_fnon_call_exceptions.
658         (run_gcc): Pass -fnon-call-exceptions.
660 2013-11-04  Jakub Jelinek  <jakub@redhat.com>
662         * optabs.c (expand_vec_perm): Revert one incorrect line from
663         2013-10-31 change.
665         PR tree-optimization/58978
666         * tree-vrp.c (all_imm_uses_in_stmt_or_feed_cond): Don't modify
667         use_stmt by single_imm_use directly.  Only call single_imm_use
668         on SSA_NAMEs.
670 2013-11-04  Vladimir Makarov  <vmakarov@redhat.com>
672         PR rtl-optimization/58968
673         * lra-spills.c (return_regno_p): New function.
674         (lra_final_code_change): Use it.
676 2013-11-04  Joseph Myers  <joseph@codesourcery.com>
678         * doc/cpp.texi (__GCC_IEC_559, __GCC_IEC_559_COMPLEX): Document macros.
679         * target.def (float_exceptions_rounding_supported_p): New hook.
680         * targhooks.c (default_float_exceptions_rounding_supported_p): New
681         function.
682         * targhooks.h (default_float_exceptions_rounding_supported_p): Declare.
683         * doc/tm.texi.in (TARGET_FLOAT_EXCEPTIONS_ROUNDING_SUPPORTED_P):
684         New @hook.
685         * doc/tm.texi: Regenerate.
686         * config.gcc (powerpc*-*-linux*): Set extra_objs.
687         * config/rs6000/rs6000-linux.c: New file.
688         * config/rs6000/rs6000-protos.h
689         (rs6000_linux_float_exceptions_rounding_supported_p): Declare.
690         * config/rs6000/linux.h
691         (TARGET_FLOAT_EXCEPTIONS_ROUNDING_SUPPORTED_P): New macro.
692         * config/rs6000/linux64.h
693         (TARGET_FLOAT_EXCEPTIONS_ROUNDING_SUPPORTED_P): Likewise.
694         * config/rs6000/t-linux (rs6000-linux.o): New rule.
695         * config/rs6000/t-linux64 (rs6000-linux.o): Likewise.
697 2013-11-04  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
699         * config/rs6000/vsx.md (*vsx_le_perm_store_<mode> for VSX_D):
700         Replace the define_insn_and_split with a define_insn and two
701         define_splits, with the split after reload re-permuting the source
702         register to its original value.
703         (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
704         (*vsx_le_perm_store_v8hi): Likewise.
705         (*vsx_le_perm_store_v16qi): Likewise.
707 2013-11-04  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
709         * config/rs6000/vector.md (vec_pack_trunc_v2df):  Adjust for
710         little endian.
712 2013-11-04  Jakub Jelinek  <jakub@redhat.com>
714         PR tree-optimization/58946
715         * tree-ssa-reassoc.c (maybe_optimize_range_tests): Update all
716         bbs with bbinfo[idx].op != NULL before all blocks with
717         bbinfo[idx].op == NULL.
719 2013-11-04  Richard Sandiford  <rdsandiford@googlemail.com>
721         * config/avr/avr-log.c (avr_double_int_pop_digit): Delete.
722         (avr_dump_double_int_hex): Likewise.
723         (avr_log_vadump): Remove %D and %X handling.
724         * config/avr/avr.c (avr_double_int_push_digit): Delete.
725         (avr_map_op_t): Change map from double_int to unsigned int.
726         (avr_map_op): Update accordingly.
727         (avr_map, avr_map_metric, avr_has_nibble_0xf, avr_map_decompose)
728         (avr_move_bits, avr_out_insert_bits, avr_fold_builtin): Operate on
729         unsigned ints rather than double_ints.
731 2013-11-03  Marek Polacek  <polacek@redhat.com>
733         Implement -fsanitize=vla-bound.
734         * opts.c (common_handle_option): Handle vla-bound.
735         * sanitizer.def (BUILT_IN_UBSAN_HANDLE_VLA_BOUND_NOT_POSITIVE): Define.
736         * flag-types.h (enum sanitize_code): Add SANITIZE_VLA.
737         * asan.c (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR.
739 2013-11-02  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
741         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Adjust for
742         little endian.
744 2013-11-02  Uros Bizjak  <ubizjak@gmail.com>
746         * config/i386/constraints.md (Ts, Tv): New address constrains.
747         * config/i386/i386.md (*lea<mode>, *<mode>_<bndcheck>): Use Ts
748         constraint for address_no_seg_operand.
749         * config/i386/sse.md (*avx512pf_gatherpf<mode>_mask)
750         (*avx512pf_gatherpf<mode>, *avx512pf_scatterpf<mode>_mask)
751         (*avx512pf_scatterpf<mode>, *avx2_gathersi<mode>)
752         (*avx2_gathersi<mode>_2, *avx2_gatherdi<mode>, *avx2_gatherdi<mode>_2)
753         (*avx2_gatherdi<mode>_3, *avx2_gatherdi<mode>_4)
754         (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2)
755         (*avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2)
756         (*avx512f_scattersi<mode> *avx512f_scatterdi<mode>): Use Tv
757         constraint for vsib_address_operand.
759 2013-11-02  Steven Bosscher  <steven@gcc.gnu.org>
761         * gcse.c (pre_delete): Remove references to regmove from comments.
762         * recog.c: (validate_replace_rtx_1): Likewise.
763         * config/rl78/rl78.c: Likewise.
764         * config/v850/v850.h: Likewise, and remove unused ENABLE_REGMOVE_PASS.
765         * common/config/m32r/m32r-common.c: Don't manipulate OPT_fregmove.
766         * common/config/mmix/mmix-common.c: Likewise.
768 2013-11-01  Trevor Saunders  <tsaunders@mozilla.com>
770         * function.c (reorder_blocks): Convert block_stack to a stack_vec.
771         * gimplify.c (gimplify_compound_lval): Likewise.
772         * graphite-clast-to-gimple.c (gloog): Likewise.
773         * graphite-dependences.c (loop_is_parallel_p): Likewise.
774         * graphite-scop-detection.c (scopdet_basic_block_info): Likewise.
775         (limit_scop); Likewise.
776         (build_scops): Likewise.
777         (dot_scop): Likewise.
778         * graphite-sese-to-poly.c (sese_dom_walker): Likewise.
779         (build_scop_drs): Likewise.
780         (insert_stmts): Likewise.
781         (insert_out_of_ssa_copy): Likewise.
782         (remove_phi): Likewise.
783         (rewrite_commutative_reductions_out_of_ssa_close_phi): Likewise.
784         * hw-doloop.c (discover_loop): Likewise.
785         * tree-call-cdce.c (shrink_wrap_one_built_in_call): Likewise.
786         * tree-dfa.c (dump_enumerated_decls): Likewise.
787         * tree-if-conv.c (if_convertable_loop_p): Likewise.
788         * tree-inline.c (tree_function_versioning): Likewise.
789         * tree-loop-distribution.c (build_rdg): Likewise.
790         (rdg_flag_vertex_and_dependent): Likewise.
791         (distribute_loop): Likewise.
792         * tree-parloops.c (loop_parallel_p): Likewise.
793         (eliminate_local_variables): Likewise.
794         (separate_decls_in_region): Likewise.
795         * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
796         * tree-ssa-phiopt.c (cond_if_else_store_replacement): Likewise.
797         * tree-ssa-uncprop.c (uncprop_dom_walker): Likewise.
798         * tree-vect-loop.c (vect_analyze_scaler_cycles_1): Likewise.
799         * tree-vect-patterns.c (vect_pattern_recog): Likewise.
800         * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
801         (vectorizable_condition): Likewise.
803 2013-11-01  Uros Bizjak  <ubizjak@gmail.com>
805         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): Always define as 0/1.
806         * configure: Regenerate.
807         * config/i386/i386.md (*movdi_internal): Change
808         HAVE_AS_IX86_INTERUNIT_MOVQ to runtime check.
809         (*movdf_internal): Ditto.
810         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
811         * config/i386/sse.md (vec_concatv2di): Output interunit movq
812         for HAVE_AS_IX86_INTERUNIT_MOVQ targets.
814 2013-10-31  Robert Suchanek  <Robert.Suchanek@imgtec.com>
816         * lra-spills.c (assign_spill_hard_regs): Remove statement terminator
817         after comment.
819 2013-10-31  David Malcolm  <dmalcolm@redhat.com>
821         Automated part of renaming of symtab_node_base to symtab_node.
823         Patch autogenerated by rename_symtab.py from
824         https://github.com/davidmalcolm/gcc-refactoring-scripts
825         revision 58bb219cc090b2f4516a9297d868c245495ee622
826         with ChangeLog entry fixed up by hand.
828         * cgraph.c (x_cgraph_nodes_queue): Rename symtab_node_base to
829         symtab_node.
830         (cgraph_node_for_asm): Likewise.
831         * cgraph.h (symtab_node_base): Likewise.
832         (cgraph_node): Likewise.
833         (varpool_node): Likewise.
834         (is_a_helper <cgraph_node>::test): Likewise.
835         (is_a_helper <varpool_node>::test): Likewise.
836         (symtab_nodes): Likewise.
837         (symtab_register_node): Likewise.
838         (symtab_unregister_node): Likewise.
839         (symtab_remove_node): Likewise.
840         (symtab_get_node): Likewise.
841         (symtab_node_for_asm): Likewise.
842         (symtab_node_asm_name): Likewise.
843         (symtab_node_name): Likewise.
844         (symtab_insert_node_to_hashtable): Likewise.
845         (symtab_add_to_same_comdat_group): Likewise.
846         (symtab_dissolve_same_comdat_group_list): Likewise.
847         (dump_symtab_node): Likewise.
848         (debug_symtab_node): Likewise.
849         (dump_symtab_base): Likewise.
850         (verify_symtab_node): Likewise.
851         (verify_symtab_base): Likewise.
852         (symtab_used_from_object_file_p): Likewise.
853         (symtab_alias_ultimate_target): Likewise.
854         (symtab_resolve_alias): Likewise.
855         (fixup_same_cpp_alias_visibility): Likewise.
856         (symtab_for_node_and_aliases): Likewise.
857         (symtab_nonoverwritable_alias): Likewise.
858         (availability symtab_node_availability): Likewise.
859         (symtab_semantically_equivalent_p): Likewise.
860         (fixup_same_cpp_alias_visibility): Likewise.
861         (symtab_prevail_in_asm_name_hash): Likewise.
862         (cgraph): Likewise.
863         (varpool): Likewise.
864         (varpool_first_variable): Likewise.
865         (varpool_next_variable): Likewise.
866         (varpool_first_static_initializer): Likewise.
867         (varpool_next_static_initializer): Likewise.
868         (varpool_first_defined_variable): Likewise.
869         (varpool_next_defined_variable): Likewise.
870         (cgraph_first_defined_function): Likewise.
871         (cgraph_next_defined_function): Likewise.
872         (cgraph_first_function): Likewise.
873         (cgraph_next_function): Likewise.
874         (cgraph_first_function_with_gimple_body): Likewise.
875         (cgraph_next_function_with_gimple_body): Likewise.
876         (symtab_alias_target): Likewise.
877         (symtab_real_symbol_p): Likewise.
878         (symtab_can_be_discarded): Likewise.
879         * cgraphbuild.c (mark_address): Likewise.
880         (mark_load): Likewise.
881         (mark_store): Likewise.
882         * cgraphunit.c (decide_is_symbol_needed): Likewise.
883         (first): Likewise.
884         (enqueue_node): Likewise.
885         (referred_to_p): Likewise.
886         (cgraph_process_same_body_aliases): Likewise.
887         (analyze_functions): Likewise.
888         (handle_alias_pairs): Likewise.
889         (output_weakrefs): Likewise.
890         (compile): Likewise.
891         * gimple-fold.c (can_refer_decl_in_current_unit_p): Likewise.
892         * ipa-inline-analysis.c (inline_write_summary): Likewise.
893         * ipa-prop.c (remove_described_reference): Likewise.
894         (try_decrement_rdesc_refcount): Likewise.
895         (ipa_edge_duplication_hook): Likewise.
896         * ipa-ref.c (ipa_record_reference): Likewise.
897         (ipa_maybe_record_reference): Likewise.
898         (ipa_clone_ref): Likewise.
899         (ipa_clone_references): Likewise.
900         (ipa_clone_referring): Likewise.
901         (ipa_find_reference): Likewise.
902         (ipa_remove_stmt_references): Likewise.
903         (ipa_clear_stmts_in_references): Likewise.
904         * ipa-ref.h (symtab_node_base): Likewise.
905         (ipa_ref): Likewise.
906         (ipa_record_reference): Likewise.
907         (ipa_maybe_record_reference): Likewise.
908         (ipa_clone_references): Likewise.
909         (ipa_clone_referring): Likewise.
910         (ipa_clone_ref): Likewise.
911         (ipa_find_reference): Likewise.
912         (ipa_remove_stmt_references): Likewise.
913         (ipa_clear_stmts_in_references): Likewise.
914         * ipa-reference.c (ipa_reference_write_optimization_summary):
915         Likewise.
916         * ipa.c (enqueue_node): Likewise.
917         (process_references): Likewise.
918         (walk_polymorphic_call_targets): Likewise.
919         (symtab_remove_unreachable_nodes): Likewise.
920         (address_taken_from_non_vtable_p): Likewise.
921         (comdat_can_be_unshared_p_1): Likewise.
922         (comdat_can_be_unshared_p): Likewise.
923         (can_replace_by_local_alias): Likewise.
924         (function_and_variable_visibility): Likewise.
925         * is-a.h: Likewise (within example in comment).
926         * lto-cgraph.c (input_cgraph_opt_summary): Likewise.
927         (lto_symtab_encoder_encode): Likewise.
928         (lto_symtab_encoder_delete_node): Likewise.
929         (lto_symtab_encoder_in_partition_p): Likewise.
930         (lto_set_symtab_encoder_in_partition): Likewise.
931         (output_refs): Likewise.
932         (compute_ltrans_boundary): Likewise.
933         (output_symtab): Likewise.
934         (input_node): Likewise.
935         (input_ref): Likewise.
936         (input_edge): Likewise.
937         (input_cgraph_1): Likewise.
938         (input_refs): Likewise.
939         (output_cgraph_opt_summary): Likewise.
940         (input_node_opt_summary): Likewise.
941         (input_cgraph_opt_section): Likewise.
942         * lto-section-in.c (lto_free_function_in_decl_state_for_node):
943         Likewise.
944         * lto-streamer-out.c (lto_output): Likewise.
945         (output_symbol_p): Likewise.
946         (produce_symtab): Likewise.
947         * lto-streamer.h (lto_encoder_entry): Likewise.
948         (lto_free_function_in_decl_state_for_node): Likewise.
949         (lto_symtab_encoder_encode): Likewise.
950         (lto_symtab_encoder_delete_node): Likewise.
951         (lto_symtab_encoder_in_partition_p): Likewise.
952         (lto_set_symtab_encoder_in_partition): Likewise.
953         (lto_symtab_encoder_lookup): Likewise.
954         (lsei_node): Likewise.
955         (lto_symtab_encoder_deref): Likewise.
956         * symtab.c (symtab_hash): Likewise.
957         (assembler_name_hash): Likewise.
958         (symtab_nodes): Likewise.
959         (hash_node): Likewise.
960         (eq_node): Likewise.
961         (hash_node_by_assembler_name): Likewise.
962         (eq_assembler_name): Likewise.
963         (insert_to_assembler_name_hash): Likewise.
964         (unlink_from_assembler_name_hash): Likewise.
965         (symtab_prevail_in_asm_name_hash): Likewise.
966         (symtab_register_node): Likewise.
967         (symtab_insert_node_to_hashtable): Likewise.
968         (symtab_unregister_node): Likewise.
969         (symtab_get_node): Likewise.
970         (symtab_remove_node): Likewise.
971         (symtab_initialize_asm_name_hash): Likewise.
972         (symtab_node_for_asm): Likewise.
973         (symtab_add_to_same_comdat_group): Likewise.
974         (symtab_dissolve_same_comdat_group_list): Likewise.
975         (symtab_node_asm_name): Likewise.
976         (symtab_node_name): Likewise.
977         (dump_symtab_base): Likewise.
978         (dump_symtab_node): Likewise.
979         (dump_symtab): Likewise.
980         (debug_symtab_node): Likewise.
981         (verify_symtab_base): Likewise.
982         (verify_symtab_node): Likewise.
983         (verify_symtab): Likewise.
984         (symtab_used_from_object_file_p): Likewise.
985         (symtab_node_availability): Likewise.
986         (symtab_alias_ultimate_target): Likewise.
987         (fixup_same_cpp_alias_visibility): Likewise.
988         (symtab_resolve_alias): Likewise.
989         (symtab_for_node_and_aliases): Likewise.
990         (symtab_for_node_and_aliases): Likewise.
991         (symtab_nonoverwritable_alias_1): Likewise.
992         (symtab_nonoverwritable_alias): Likewise.
993         (symtab_semantically_equivalent_p): Likewise.
994         * value-prof.c (init_node_map): Likewise.
995         * varasm.c (find_decl): Likewise.
996         * varpool.c (varpool_node_for_asm): Likewise.
997         (varpool_remove_unreferenced_decls): Likewise.
999 2013-10-31  David Malcolm  <dmalcolm@redhat.com>
1001         Manual part of renaming of symtab_node_base to symtab_node.
1003         * ipa-ref.h (symtab_node): Remove typedef to pointer type, as it
1004         clashes with the preferred name for the base class.
1005         (const_symtab_node): Remove redundant typedef.
1007 2013-10-31  Jakub Jelinek  <jakub@redhat.com>
1009         * optabs.c (expand_vec_perm): Avoid vector mode punning
1010         SUBREGs in SET_DEST.
1011         * expmed.c (store_bit_field_1): Likewise.
1012         * config/i386/sse.md (movdi_to_sse, vec_pack_sfix_trunc_v2df,
1013         vec_pack_sfix_v2df, vec_shl_<mode>, vec_shr_<mode>,
1014         vec_interleave_high<mode>, vec_interleave_low<mode>): Likewise.
1015         * config/i386/i386.c (ix86_expand_vector_move_misalign,
1016         ix86_expand_sse_movcc, ix86_expand_int_vcond, ix86_expand_vec_perm,
1017         ix86_expand_sse_unpack, ix86_expand_args_builtin,
1018         ix86_expand_vector_init_duplicate, ix86_expand_vector_set,
1019         emit_reduc_half, expand_vec_perm_blend, expand_vec_perm_pshufb,
1020         expand_vec_perm_interleave2, expand_vec_perm_pshufb2,
1021         expand_vec_perm_vpshufb2_vpermq,
1022         expand_vec_perm_vpshufb2_vpermq_even_odd, expand_vec_perm_even_odd_1,
1023         expand_vec_perm_broadcast_1, expand_vec_perm_vpshufb4_vpermq2,
1024         ix86_expand_sse2_mulv4si3, ix86_expand_pinsr): Likewise.
1025         (expand_vec_perm_palignr): Likewise.  Modify a copy of *d rather
1026         than *d itself.
1028 2013-10-31  Uros Bizjak  <ubizjak@gmail.com>
1030         * config/i386/i386.c (ix86_expand_sse2_abs): Rename function arguments.
1031         Use gcc_unreachable for unhandled modes.  Do not check results of
1032         expand_simple_binop.  If not expanded to target, move the result.
1034 2013-10-31  Chung-Ju Wu  <jasonwucj@gmail.com>
1035             Shiva Chen  <shiva0217@gmail.com>
1037         * config.gcc (nds32*-*-*): Add nds32 target.
1038         * config/nds32/nds32.c: New file.
1039         * config/nds32/nds32.h: New file.
1040         * config/nds32/nds32.md: New file.
1041         * config/nds32/constants.md: New file.
1042         * config/nds32/constraints.md: New file.
1043         * config/nds32/iterators.md: New file.
1044         * config/nds32/nds32-doubleword.md: New file.
1045         * config/nds32/nds32-intrinsic.md: New file.
1046         * config/nds32/nds32_intrinsic.h: New file.
1047         * config/nds32/nds32-modes.def: New file.
1048         * config/nds32/nds32-multiple.md: New file.
1049         * config/nds32/nds32.opt: New file.
1050         * config/nds32/nds32-opts.h: New file.
1051         * config/nds32/nds32-protos.h: New file.
1052         * config/nds32/nds32-peephole2.md: New file.
1053         * config/nds32/pipelines.md: New file.
1054         * config/nds32/predicates.md: New file.
1055         * config/nds32/t-mlibs: New file.
1056         * common/config/nds32: New directory and files.
1058         * doc/invoke.texi (NDS32 options): Document nds32 specific options.
1059         * doc/md.texi (NDS32 family): Document nds32 specific constraints.
1060         * doc/install.texi (Cross-Compiler-Specific Options): Document
1061         --with-nds32-lib for nds32 target.
1062         * doc/extend.texi (Function Attributes, Target Builtins): Document
1063         nds32 specific attributes.
1065 2013-10-31  Vladimir Makarov  <vmakarov@redhat.com>
1067         * lra-constraints (process_alt_operands): Use the result
1068         elimination register for operand when matching constraints.
1070 2013-10-31  Jakub Jelinek  <jakub@redhat.com>
1072         * tree-vrp.c (maybe_set_nonzero_bits): New function.
1073         (remove_range_assertions): Call it.
1075         * tree.c (tree_ctz): New function.
1076         * tree.h (tree_ctz): New prototype.
1077         * tree-ssanames.h (get_range_info, get_nonzero_bits): Change
1078         first argument from tree to const_tree.
1079         * tree-ssanames.c (get_range_info, get_nonzero_bits): Likewise.
1080         * tree-vectorizer.h (vect_generate_tmps_on_preheader): New prototype.
1081         * tree-vect-loop-manip.c (vect_generate_tmps_on_preheader): No longer
1082         static.
1083         * expr.c (highest_pow2_factor): Reimplemented using tree_ctz.
1084         * tree-vect-loop.c (vect_analyze_loop_operations,
1085         vect_transform_loop): Don't force scalar loop for bound just because
1086         number of iterations is unknown, only do it if it is not known to be
1087         a multiple of vectorization_factor.
1088         * builtins.c (get_object_alignment_2): Use tree_ctz on offset.
1090         * gimple-pretty-print.c (dump_ssaname_info): Print newline also
1091         in case of VR_VARYING.  Print get_nonzero_bits if not all ones.
1092         * tree-ssanames.h (struct range_info_def): Add nonzero_bits field.
1093         (set_nonzero_bits, get_nonzero_bits): New prototypes.
1094         * tree-ssa-ccp.c (get_default_value): Use get_range_info to see if
1095         a default def isn't partially constant.
1096         (ccp_finalize): If after IPA, set_range_info if integral SSA_NAME
1097         is known to be partially zero.
1098         (evaluate_stmt): If we'd return otherwise VARYING, use get_range_info
1099         to see if a default def isn't partially constant.
1100         * tree-ssanames.c (set_range_info): Initialize nonzero_bits upon
1101         creation of a range, if VR_RANGE, try to improve nonzero_bits from
1102         the range.
1103         (set_nonzero_bits, get_nonzero_bits): New functions.
1105         * tree-cfg.c (assert_unreachable_fallthru_edge_p): New function.
1106         * tree-cfg.h (assert_unreachable_fallthru_edge_p): New prototype.
1107         * tree-vrp.c (all_imm_uses_in_stmt_or_feed_cond): New function.
1108         (remove_range_assertions): If ASSERT_EXPR_VAR has no other immediate
1109         uses but in the condition and ASSERT_EXPR and the other successor of
1110         the predecessor bb is __builtin_unreachable (), set_range_info of the
1111         ASSERT_EXPR_VAR to the range info of the ASSERT_EXPR's lhs.
1113 2013-10-31  Martin Jambor  <mjambor@suse.cz>
1115         PR rtl-optimization/58934
1116         Revert:
1117         2013-10-30  Martin Jambor  <mjambor@suse.cz>
1118         PR rtl-optimization/10474
1119         * ira.c (find_moveable_pseudos): Do not calculate dominance info
1120         nor df analysis.
1121         (interesting_dest_for_shprep): New function.
1122         (split_live_ranges_for_shrink_wrap): Likewise.
1123         (ira): Calculate dominance info and df analysis. Call
1124         split_live_ranges_for_shrink_wrap.
1126 2013-10-31  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
1127             Yury Gribov  <y.gribov@samsung.com>
1129         PR sanitizer/58543
1130         * asan.c (asan_clear_shadow): Allocate a new vreg for temporary
1131         shadow pointer to avoid clobbering the main one.
1133 2013-10-31  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
1135         * lower-subreg.c (resolve_simple_move): Copy REG_INC note.
1137 2013-10-30  Vladimir Makarov  <vmakarov@redhat.com>
1139         PR bootstrap/58933
1140         * ira-color.c (update_costs_from_copies): Add new parameter.  Use
1141         it for calling update_costs_from_allocno.
1142         (assign_hard_reg): Call restore_costs_from_copies only for
1143         !retry_p.  Pass new argument to update_costs_from_copies.
1144         (color_pass): Pass new argument to update_costs_from_copies.
1145         (ira_mark_allocation_change): Ditto.
1147 2013-10-30  Sharad Singhai  <singhai@google.com>
1149         PR middle-end/58134
1150         * opts.c (common_handle_option): Remove deprecated option
1151         -ftree-vectorizer-verbose.
1152         * doc/invoke.texi (Debugging Options): Ditto.
1153         * opts-global.c (handle_common_deferred_options): Ditto.
1154         (dump_remap_tree_vectorizer_verbose): Delete.
1155         * common.opt: Set -ftree-vectorizer-verbose as an ignored option.
1157 2013-10-30  DJ Delorie  <dj@redhat.com>
1159         * config/rx/rx.c (ADD_RX_BUILTIN0): New macro, used for builtins
1160         that take no arguments.
1162 2013-10-30  Joern Rennecke  <joern.rennecke@embecosm.com>
1164         PR other/58545
1165         * reload1.c (update_eliminables_and_spill): New function, broken
1166         out of reload.
1167         (reload): Use it.  Check for frame size change after frame size
1168         alignment, and call update_eliminables_and_spill first if continue-ing.
1170 2013-10-30  Cong Hou  <congh@google.com>
1172         PR target/58762
1173         * config/i386/i386-protos.h (ix86_expand_sse2_abs): New function.
1174         * config/i386/i386.c (ix86_expand_sse2_abs): New function.
1175         * config/i386/sse.md: Add SSE2 support to abs (8/16/32-bit-int).
1177 2013-10-18  Mikael Pettersson  <mikpelinux@gmail.com>
1179         PR rtl-optimization/58369
1180         * reload1.c (compute_reload_subreg_offset): New function.
1181         (choose_reload_regs): Use it to pass endian-correct
1182         offset to subreg_regno_offset.
1184 2013-10-30  Tobias Burnus  <burnus@net-b.de>
1186         PR other/33426
1187         * tree-cfg.c (replace_loop_annotate): Replace warning by
1188         warning_at.
1190 2013-10-30  Jason Merrill  <jason@redhat.com>
1192         * configure.ac (loose_warn): Add -Wno-format if
1193         --disable-build-format-warnings.
1195 2013-10-30  David Malcolm  <dmalcolm@redhat.com>
1197         * cgraphunit.c (analyze_functions): Split symtab_node declarations
1198         onto multiple lines to make things easier for rename_symtab.py.
1200         * symtab.c (symtab_dissolve_same_comdat_group_list): Likewise.
1201         (symtab_semantically_equivalent_p): Likewise.
1203 2013-10-30  Vladimir Makarov  <vmakarov@redhat.com>
1205         PR target/58784
1206         * lra.c (check_rtl): Remove address check before LRA work.
1208 2013-10-30  Marc Glisse  <marc.glisse@inria.fr>
1210         * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Look for a
1211         POINTER_PLUS_EXPR in the defining statement.
1213 2013-10-30  Vladimir Makarov  <vmakarov@redhat.com>
1215         * regmove.c: Remove.
1216         * tree-pass.h (make_pass_regmove): Remove.
1217         * timevar.def (TV_REGMOVE): Remove.
1218         * passes.def (pass_regmove): Remove.
1219         * opts.c (default_options_table): Remove entry for regmove.
1220         * doc/passes.texi: Remove regmove pass description.
1221         * doc/invoke.texi (-foptimize-register-move, -fregmove): Remove
1222         options.
1223         (-fdump-rtl-regmove): Ditto.
1224         * common.opt (foptimize-register-move, fregmove): Ignore.
1225         * Makefile.in (OBJS): Remove regmove.o.
1226         * regmove.c: Remove.
1227         * ira-int.h (struct ira_allocno_pref, ira_pref_t): New structure
1228         and type.
1229         (struct ira_allocno) New member allocno_prefs.
1230         (ALLOCNO_PREFS): New macro.
1231         (ira_prefs, ira_prefs_num): New external vars.
1232         (ira_setup_alts, ira_get_dup_out_num, ira_debug_pref): New prototypes.
1233         (ira_debug_prefs, ira_debug_allocno_prefs, ira_create_pref): Ditto.
1234         (ira_add_allocno_pref, ira_remove_pref, ira_remove_allocno_prefs):
1235         Ditto.
1236         (ira_add_allocno_copy_to_list): Remove prototype.
1237         (ira_swap_allocno_copy_ends_if_necessary): Ditto.
1238         (ira_pref_iterator): New type.
1239         (ira_pref_iter_init, ira_pref_iter_cond): New functions.
1240         (FOR_EACH_PREF): New macro.
1241         * ira.c (commutative_constraint_p): Move from ira-conflicts.c.
1242         (ira_get_dup_out_num): Ditto. Rename from get_dup_num.  Modify the
1243         code.
1244         (ira_setup_alts): New function.
1245         (decrease_live_ranges_number): New function.
1246         (ira): Call the above function.
1247         * ira-build.c (ira_prefs, ira_prefs_num): New global vars.
1248         (ira_create_allocno): Initialize allocno prefs.
1249         (pref_pool, pref_vec): New static vars.
1250         (initiate_prefs, find_allocno_pref, ira_create_pref): New
1251         functions.
1252         (add_allocno_pref_to_list, ira_add_allocno_pref, print_pref): Ditto.
1253         (ira_debug_pref, print_prefs, ira_debug_prefs): Ditto.
1254         (print_allocno_prefs, ira_debug_allocno_prefs, finish_pref): Ditto.
1255         (ira_remove_pref, ira_remove_allocno_prefs, finish_prefs): Ditto.
1256         (ira_add_allocno_copy_to_list): Make static.  Rename to
1257         add_allocno_copy_to_list.
1258         (ira_swap_allocno_copy_ends_if_necessary): Make static.  Rename to
1259         swap_allocno_copy_ends_if_necessary.
1260         (remove_unnecessary_allocnos, remove_low_level_allocnos): Call
1261         ira_remove_allocno_prefs.
1262         (ira_flattening): Ditto.
1263         (ira_build): Call initiate_prefs, print_prefs.
1264         (ira_destroy): Call finish_prefs.
1265         * ira-color.c (struct update_cost_record): New.
1266         (struct allocno_color_data): Add new member update_cost_records.
1267         (update_cost_record_pool): New static var.
1268         (init_update_cost_records, get_update_cost_record): New functions.
1269         (free_update_cost_record_list, finish_update_cost_records): Ditto.
1270         (struct update_cost_queue_elem): Add member from.
1271         (initiate_cost_update): Call init_update_cost_records.
1272         (finish_cost_update): Call finish_update_cost_records.
1273         (queue_update_cost, get_next_update_cost): Add new param from.
1274         (Update_allocno_cost, update_costs_from_allocno): New functions.
1275         (update_costs_from_prefs): Ditto.
1276         (update_copy_costs): Rename to update_costs_from_copies.
1277         (restore_costs_from_copies): New function.
1278         (update_conflict_hard_regno_costs): Don't go back.
1279         (assign_hard_reg): Call restore_costs_from_copies.  Add printing
1280         more debug info.
1281         (pop_allocnos): Add priniting more debug info.
1282         (color_allocnos): Remove prefs for conflicting hard regs.
1283         Call update_costs_from_prefs.
1284         * ira-conflicts.c (commutative_constraint_p): Move to ira.c
1285         (get_dup_num): Rename, modify, and move to ira.c
1286         (process_regs_for_copy): Add prefs.
1287         (add_insn_allocno_copies): Put src as first arg of
1288         process_regs_for_copy.  Remove dead code.  Call ira_setup_alts.
1289         * ira-costs.c (record_reg_classes): Modify and move code into
1290         record_operands_costs.
1291         (find_costs_and_classes): Create prefs for the hard reg of small
1292         reg class.
1293         (process_bb_node_for_hard_reg_moves): Add prefs.
1295 2013-10-30  Richard Biener  <rguenther@suse.de>
1297         PR middle-end/57100
1298         * basic-block.h (pre_and_rev_post_order_compute_fn): New function.
1299         * cfganal.c (pre_and_rev_post_order_compute_fn): New function
1300         as worker for ...
1301         (pre_and_rev_post_order_compute): ... which now wraps it.
1302         * graph.c (draw_cfg_nodes_no_loops): Use
1303         pre_and_rev_post_order_compute_fn to avoid ICEing and dependence
1304         on cfun.
1306 2013-10-30  Christian Bruel  <christian.bruel@st.com>
1308         * config/sh/sh-mem.cc (sh_expand_cmpnstr): New function.
1309         (sh_expand_cmpstr): Handle known align and schedule improvements.
1310         * config/sh/sh-protos.h (sh_expand_cmpstrn): Declare.
1311         * config/sh/sh.md (cmpstrnsi): New pattern.
1313 2013-10-30  Martin Jambor  <mjambor@suse.cz>
1315         PR rtl-optimization/10474
1316         * ira.c (find_moveable_pseudos): Do not calculate dominance info
1317         nor df analysis.
1318         (interesting_dest_for_shprep): New function.
1319         (split_live_ranges_for_shrink_wrap): Likewise.
1320         (ira): Calculate dominance info and df analysis. Call
1321         split_live_ranges_for_shrink_wrap.
1323 2013-10-30  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1325         PR target/58854
1326         * config/arm/arm.c (arm_expand_epilogue_apcs_frame): Emit blockage.
1328 2013-10-30  Ilya Enkovich  <ilya.enkovich@intel.com>
1330         * tree-core.h (tree_index): Add TI_POINTER_BOUNDS_TYPE.
1331         * tree.h (POINTER_BOUNDS_P): New.
1332         (BOUNDED_TYPE_P): New.
1333         (BOUNDED_P): New.
1334         (pointer_bounds_type_node): New.
1335         * tree.c (build_common_tree_nodes): Initialize
1336         pointer_bounds_type_node.
1337         * gimple.h (gimple_call_get_nobnd_arg_index): New.
1338         (gimple_call_num_nobnd_args): New.
1339         (gimple_call_nobnd_arg): New.
1340         (gimple_return_retbnd): New.
1341         (gimple_return_set_retbnd): New
1342         * gimple.c (gimple_build_return): Increase number of ops
1343         for return statement.
1344         (gimple_call_get_nobnd_arg_index): New.
1345         * gimple-pretty-print.c (dump_gimple_return): Print second op.
1347 2013-10-30  Ilya Enkovich  <ilya.enkovich@intel.com>
1349         * ipa.c (cgraph_build_static_cdtor_1): Support contructors
1350         with "chkp ctor" and "bnd_legacy" attributes.
1351         * gimplify.c (gimplify_init_constructor): Avoid infinite
1352         loop during gimplification of bounds initializer.
1354 2013-10-30  Ilya Enkovich  <ilya.enkovich@intel.com>
1356         * c-family/c-common.c (handle_bnd_variable_size_attribute): New.
1357         (handle_bnd_legacy): New.
1358         (c_common_attribute_table): Add bnd_variable_size and bnd_legacy.
1359         * doc/extend.texi: Document bnd_variable_size and bnd_legacy
1360         attributes.
1362 2013-10-29  Ilya Enkovich  <ilya.enkovich@intel.com>
1364         * builtin-types.def (BT_FN_VOID_CONST_PTR): New.
1365         (BT_FN_PTR_CONST_PTR): New.
1366         (BT_FN_CONST_PTR_CONST_PTR): New.
1367         (BT_FN_PTR_CONST_PTR_SIZE): New.
1368         (BT_FN_PTR_CONST_PTR_CONST_PTR): New.
1369         (BT_FN_VOID_PTRPTR_CONST_PTR): New.
1370         (BT_FN_VOID_CONST_PTR_SIZE): New.
1371         (BT_FN_PTR_CONST_PTR_CONST_PTR_SIZE): New.
1372         * chkp-builtins.def: New.
1373         * builtins.def: include chkp-builtins.def.
1374         (DEF_CHKP_BUILTIN): New.
1375         * builtins.c (expand_builtin): Support BUILT_IN_CHKP_INIT_PTR_BOUNDS,
1376         BUILT_IN_CHKP_NULL_PTR_BOUNDS, BUILT_IN_CHKP_COPY_PTR_BOUNDS,
1377         BUILT_IN_CHKP_CHECK_PTR_LBOUNDS, BUILT_IN_CHKP_CHECK_PTR_UBOUNDS,
1378         BUILT_IN_CHKP_CHECK_PTR_BOUNDS, BUILT_IN_CHKP_SET_PTR_BOUNDS,
1379         BUILT_IN_CHKP_NARROW_PTR_BOUNDS, BUILT_IN_CHKP_STORE_PTR_BOUNDS,
1380         BUILT_IN_CHKP_GET_PTR_LBOUND, BUILT_IN_CHKP_GET_PTR_UBOUND,
1381         BUILT_IN_CHKP_BNDMK, BUILT_IN_CHKP_BNDSTX, BUILT_IN_CHKP_BNDCL,
1382         BUILT_IN_CHKP_BNDCU, BUILT_IN_CHKP_BNDLDX, BUILT_IN_CHKP_BNDRET,
1383         BUILT_IN_CHKP_INTERSECT, BUILT_IN_CHKP_ARG_BND, BUILT_IN_CHKP_NARROW,
1384         BUILT_IN_CHKP_EXTRACT_LOWER, BUILT_IN_CHKP_EXTRACT_UPPER.
1385         * common.opt (fcheck-pointer-bounds): New.
1386         * toplev.c (process_options): Check Pointer Bounds Checker is
1387         supported.
1388         * doc/extend.texi: Document Pointer Bounds Checker built-in functions.
1390 2013-10-30  Ilya Enkovich  <ilya.enkovich@intel.com>
1392         * target.def (builtin_chkp_function): New.
1393         (chkp_bound_type): New.
1394         (chkp_bound_mode): New.
1395         (fn_abi_va_list_bounds_size): New.
1396         (load_bounds_for_arg): New.
1397         (store_bounds_for_arg): New.
1398         * targhooks.h (default_load_bounds_for_arg): New.
1399         (default_store_bounds_for_arg): New.
1400         (default_fn_abi_va_list_bounds_size): New.
1401         (default_chkp_bound_type): New.
1402         (default_chkp_bound_mode): New.
1403         (default_builtin_chkp_function): New.
1404         * targhooks.c (default_load_bounds_for_arg): New.
1405         (default_store_bounds_for_arg): New.
1406         (default_fn_abi_va_list_bounds_size): New.
1407         (default_chkp_bound_type): New.
1408         (default_chkp_bound_mode); New.
1409         (default_builtin_chkp_function): New.
1410         * doc/tm.texi.in (TARGET_FN_ABI_VA_LIST_BOUNDS_SIZE): New.
1411         (TARGET_LOAD_BOUNDS_FOR_ARG): New.
1412         (TARGET_STORE_BOUNDS_FOR_ARG): New.
1413         (TARGET_BUILTIN_CHKP_FUNCTION): New.
1414         (TARGET_CHKP_BOUND_TYPE): New.
1415         (TARGET_CHKP_BOUND_MODE): New.
1416         * doc/tm.texi: Regenerated.
1417         * langhooks.h (lang_hooks): Add chkp_supported field.
1418         * langhooks-def.h (LANG_HOOKS_CHKP_SUPPORTED): New.
1419         (LANG_HOOKS_INITIALIZER); Add LANG_HOOKS_CHKP_SUPPORTED.
1421 2013-10-29  Andrew Pinski <apinski@cavium.com>
1423         * tree-ssa-ifcombine.c: Include rtl.h and tm_p.h.
1424         (ifcombine_ifandif): Handle cases where maybe_fold_and_comparisons
1425         fails, combining the branches anyways.
1426         (tree_ssa_ifcombine): Inverse the order of the basic block walk,
1427         increases the number of combinings.
1428         * gimple.h (gsi_start_nondebug_after_labels_bb): New function.
1430 2013-10-29  Mike Stump  <mikestump@comcast.net>
1432         * machmode.def (PARTIAL_INT_MODE): Add precision and name.
1433         * genmodes.c (PARTIAL_INT_MODE): Add precision and name.
1434         (make_vector_mode): Increase namebuf to 16.
1435         (emit_insn_modes_h): When processing BImode, don't
1436         also match partial int modes.
1437         (emit_class_narrowest_mode): Likewise.
1439         * config/bfin/bfin-modes.def: Add precision to PDI.
1440         * config/m32c/m32c-modes.def: Add precision to PSI.
1441         * config/msp430/msp430-modes.def: Add precision to PSI.
1442         * config/rs6000/rs6000-modes.def: Add precision to PTI.
1443         * config/sh/sh-modes.def: Add precision to PSI and PDI.
1445 2013-10-29  Oleg Endo  <olegendo@gcc.gnu.org>
1447         PR target/54236
1448         * config/sh/sh.md (*addc): Rename existing variations to ...
1449         (*addc_r_r_1, *addc_2r_1, *addc_r_1): ... these.
1450         (*addc_r_lsb, *addc_r_r_lsb, *addc_r_lsb_r, *addc_2r_lsb, *addc_r_msb,
1451         *addc_r_r_msb, *addc_2r_msb): New insn_and_split patterns.
1452         * config/sh/sh.c (addsubcosts): Handle some addc special cases.
1454 2013-10-29  Teresa Johnson  <tejohnson@google.com>
1456         PR ipa/58862
1457         * tree-ssa-tail-merge.c (replace_block_by): Tolerate profile
1458         insanities when updating probabilities.
1460 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
1462         * gdbhooks.py (CGraphNodePrinter.to_string): Update gdb
1463         prettyprinter for cgraph_node to reflect the conversion of the
1464         symtable types to a C++ class hierarchy: it now *is* a
1465         symtab_node_base, rather than having one (named "symbol").
1467 2013-10-29  Balaji V. Iyer  <balaji.v.iyer@intel.com>
1469         * builtins.c (is_builtin_name): Added a check for __cilkrts_detach and
1470         __cilkrts_pop_frame.  If matched, then return true for built-in
1471         function name.
1472         (expand_builtin): Added BUILT_IN_CILK_DETACH and
1473         BUILT_IN_CILK_POP_FRAME case.
1474         * langhooks-def.h (lhd_install_body_with_frame_cleanup): New prototype.
1475         (lhs_cilk_detect_spawn): Likewise.
1476         (LANG_HOOKS_DECLS): Added LANG_HOOKS_CILKPLUS.
1477         (LANG_HOOKS_CILKPLUS_DETECT_SPAWN_AND_UNWRAP): New #define.
1478         (LANG_HOOKS_CILKPLUS_FRAME_CLEANUP): Likewise.
1479         (LANG_HOOKS_CILKPLUS_GIMPLIFY_SPAWN): Likewise.
1480         (LANG_HOOKS_CILKPLUS): Likewise.
1481         * tree.h (CILK_SPAWN_FN): Likewise.
1482         * builtin.def (DEF_CILK_BUILTIN_STUB): Likewise.
1483         * Makefile.in (C_COMMON_OBJS): Added c-family/cilk.o.
1484         (OBJS): Added cilk-common.o.
1485         (BUILTINS_DEF): Added cilk-builtins.def.
1486         * langhooks.c (lhd_install_body_with_frame_cleanup): New function.
1487         (lhd_cilk_detect_spawn): Likewise.
1488         * langhooks.h (lang_hooks_for_cilkplus): New struct.
1489         (struct lang_hooks): Added new field called "cilkplus."
1490         * cilk-common.c: New file.
1491         * cilk.h: Likewise.
1492         * cilk-builtins.def: Likewise.
1493         * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Added
1494         "__cilk" macro and set it to 200.
1495         * function.h (struct function::cilk_frame_decl): New field.
1496         (struct function::is_cilk_function): Likewise.
1497         (struct function::calls_cilk_spawn): Likewise.
1498         * gimplify.c (gimplify_call_expr): Added a check if the function call
1499         being gimplified is a spawn detach point.  If so, then add pop_frame
1500         and detach function calls.
1501         (gimplify_expr): Added a CILK_SPAWN_STMT and CILK_SYNC_STMT case
1502         for gimplifying _Cilk_spawn and _Cilk_sync statements.
1503         (gimplify_return_expr): Added a check for _Cilk_spawn usage in
1504         function.  If so, added a _Cilk_sync and gimplified it.
1505         (gimplify_modify_expr): Added a check for _Cilk_spawn in MODIFY and
1506         INIT_EXPRs.  If so, then call gimplify_cilk_spawn.
1507         * ipa-inline-analysis (initialize_inline_failed): Prevent inlining of
1508         spawner function.
1509         (can_inline_edge_p): Prevent inling of spawnee function.
1510         * ira.c (ira_setup_eliminable_regset): Force usage of frame pointer
1511         for functions that use Cilk keywords.
1512         * tree-inline.h (struct copy_body_data::remap_var_for_cilk): New field.
1513         * tree-pretty-print.c (dump_generic_node): Added CILK_SPAWN_STMT and
1514         CILK_SYNC_STMT cases.
1515         * tree.def (DEFTREECODE): Added CILK_SPAWN_STMT and CILK_SYNC_STMT
1516         trees.
1517         * generic.texi (CILK_SPAWN_STMT): Added documentation for _Cilk_spawn.
1518         (CILK_SYNC_STMT): Added documentation for _Cilk_sync.
1519         * passes.texi (Cilk Keywords): New section that describes the compiler
1520         code changes for handling Cilk Keywords.
1522 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
1524         Patch autogenerated by refactor_symtab.py from
1525         https://github.com/davidmalcolm/gcc-refactoring-scripts
1526         revision 58bb219cc090b2f4516a9297d868c245495ee622
1528         * asan.c (asan_finish_file): Update for conversion of symtab types to
1529         a true class hierarchy.
1530         * cfgexpand.c (estimated_stack_frame_size): Likewise.
1531         * cgraph.c (cgraph_get_body): Likewise.
1532         (cgraph_get_create_real_symbol_node): Likewise.
1533         (verify_cgraph_node): Likewise.
1534         (verify_edge_corresponds_to_fndecl): Likewise.
1535         (verify_edge_count_and_frequency): Likewise.
1536         (cgraph_will_be_removed_from_program_if_no_direct_calls): Likewise.
1537         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
1538         (cgraph_can_remove_if_no_direct_calls_and_refs_p): Likewise.
1539         (cgraph_node_cannot_return): Likewise.
1540         (cgraph_set_pure_flag_1): Likewise.
1541         (cgraph_set_const_flag_1): Likewise.
1542         (cgraph_set_nothrow_flag_1): Likewise.
1543         (cgraph_make_node_local_1): Likewise.
1544         (cgraph_for_node_and_aliases): Likewise.
1545         (cgraph_for_node_thunks_and_aliases): Likewise.
1546         (cgraph_node_can_be_local_p): Likewise.
1547         (cgraph_node_cannot_be_local_p_1): Likewise.
1548         (cgraph_function_body_availability): Likewise.
1549         (dump_cgraph_node): Likewise.
1550         (cgraph_rtl_info): Likewise.
1551         (cgraph_mark_address_taken_node): Likewise.
1552         (cgraph_remove_node): Likewise.
1553         (cgraph_release_function_body): Likewise.
1554         (cgraph_update_edges_for_call_stmt_node): Likewise.
1555         (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
1556         (cgraph_make_edge_direct): Likewise.
1557         (cgraph_resolve_speculation): Likewise.
1558         (cgraph_speculative_call_info): Likewise.
1559         (cgraph_turn_edge_to_speculative): Likewise.
1560         (cgraph_create_edge_1): Likewise.
1561         (cgraph_set_call_stmt): Likewise.
1562         (cgraph_node_for_asm): Likewise.
1563         (cgraph_add_thunk): Likewise.
1564         (cgraph_same_body_alias): Likewise.
1565         (cgraph_create_function_alias): Likewise.
1566         (cgraph_create_node): Likewise.
1567         (cgraph_create_empty_node): Likewise.
1568         (record_function_versions): Likewise.
1569         (used_from_object_file_p): Likewise.
1570         * cgraph.h (symtab_can_be_discarded): Likewise.
1571         (symtab_real_symbol_p): Likewise.
1572         (cgraph_mark_force_output_node): Likewise.
1573         (cgraph_edge_recursive_p): Likewise.
1574         (symtab_alias_target): Likewise.
1575         (varpool_all_refs_explicit_p): Likewise.
1576         (varpool_can_remove_if_no_refs): Likewise.
1577         (cgraph_only_called_directly_or_aliased_p): Likewise.
1578         (cgraph_next_function_with_gimple_body): Likewise.
1579         (cgraph_first_function_with_gimple_body): Likewise.
1580         (cgraph_function_with_gimple_body_p): Likewise.
1581         (cgraph_next_function): Likewise.
1582         (cgraph_first_function): Likewise.
1583         (cgraph_next_defined_function): Likewise.
1584         (cgraph_first_defined_function): Likewise.
1585         (varpool_next_defined_variable): Likewise.
1586         (varpool_first_defined_variable): Likewise.
1587         (varpool_next_static_initializer): Likewise.
1588         (varpool_first_static_initializer): Likewise.
1589         (varpool_next_variable): Likewise.
1590         (varpool_first_variable): Likewise.
1591         (varpool_node_name): Likewise.
1592         (varpool): Likewise.
1593         (cgraph): Likewise.
1594         (is_a_helper <varpool_node>::test): Likewise.
1595         (is_a_helper <cgraph_node>::test): Likewise.
1596         (varpool_variable_node): Likewise.
1597         (cgraph_function_or_thunk_node): Likewise.
1598         (varpool_alias_target): Likewise.
1599         (cgraph_alias_target): Likewise.
1600         (cgraph_node_name): Likewise.
1601         (varpool_node_asm_name): Likewise.
1602         (cgraph_node_asm_name): Likewise.
1603         * cgraphbuild.c (remove_cgraph_callee_edges): Likewise.
1604         (cgraph_rebuild_references): Likewise.
1605         (rebuild_cgraph_edges): Likewise.
1606         (record_eh_tables): Likewise.
1607         (build_cgraph_edges): Likewise.
1608         (mark_store): Likewise.
1609         (mark_load): Likewise.
1610         (mark_address): Likewise.
1611         (record_type_list): Likewise.
1612         (record_reference): Likewise.
1613         * cgraphclones.c (cgraph_materialize_all_clones): Likewise.
1614         (cgraph_materialize_clone): Likewise.
1615         (cgraph_function_versioning): Likewise.
1616         (cgraph_copy_node_for_versioning): Likewise.
1617         (update_call_expr): Likewise.
1618         (cgraph_find_replacement_node): Likewise.
1619         (cgraph_create_virtual_clone): Likewise.
1620         (cgraph_clone_node): Likewise.
1621         * cgraphunit.c (compile): Likewise.
1622         (output_weakrefs): Likewise.
1623         (output_in_order): Likewise.
1624         (expand_function): Likewise.
1625         (assemble_thunks_and_aliases): Likewise.
1626         (expand_thunk): Likewise.
1627         (mark_functions_to_output): Likewise.
1628         (handle_alias_pairs): Likewise.
1629         (analyze_functions): Likewise.
1630         (walk_polymorphic_call_targets): Likewise.
1631         (varpool_finalize_decl): Likewise.
1632         (process_function_and_variable_attributes): Likewise.
1633         (cgraph_process_same_body_aliases): Likewise.
1634         (analyze_function): Likewise.
1635         (cgraph_add_new_function): Likewise.
1636         (cgraph_finalize_function): Likewise.
1637         (referred_to_p): Likewise.
1638         (cgraph_reset_node): Likewise.
1639         (cgraph_process_new_functions): Likewise.
1640         (enqueue_node): Likewise.
1641         (decide_is_symbol_needed): Likewise.
1642         * coverage.c (coverage_compute_profile_id): Likewise.
1643         * dbxout.c (dbxout_expand_expr): Likewise.
1644         * dwarf2out.c (premark_types_used_by_global_vars_helper): Likewise.
1645         (reference_to_unused): Likewise.
1646         * gimple-fold.c (can_refer_decl_in_current_unit_p): Likewise.
1647         * gimplify.c (unvisit_body): Likewise.
1648         (unshare_body): Likewise.
1649         * ipa-cp.c (ipcp_generate_summary): Likewise.
1650         (ipcp_decision_stage): Likewise.
1651         (identify_dead_nodes): Likewise.
1652         (decide_whether_version_node): Likewise.
1653         (decide_about_value): Likewise.
1654         (perhaps_add_new_callers): Likewise.
1655         (create_specialized_node): Likewise.
1656         (update_profiling_info): Likewise.
1657         (ipcp_propagate_stage): Likewise.
1658         (estimate_local_effects): Likewise.
1659         (good_cloning_opportunity_p): Likewise.
1660         (devirtualization_time_bonus): Likewise.
1661         (propagate_constants_accross_call): Likewise.
1662         (initialize_node_lattices): Likewise.
1663         (ipcp_cloning_candidate_p): Likewise.
1664         (determine_versionability): Likewise.
1665         (print_all_lattices): Likewise.
1666         (print_lattice): Likewise.
1667         (ipcp_discover_new_direct_edges): Likewise.
1668         * ipa-devirt.c (ipa_devirt): Likewise.
1669         (likely_target_p): Likewise.
1670         (update_type_inheritance_graph): Likewise.
1671         (possible_polymorphic_call_target_p): Likewise.
1672         (dump_possible_polymorphic_call_targets): Likewise.
1673         (devirt_variable_node_removal_hook): Likewise.
1674         (record_binfo): Likewise.
1675         (maybe_record_node): Likewise.
1676         (build_type_inheritance_graph): Likewise.
1677         * ipa-inline-analysis.c (inline_write_summary): Likewise.
1678         (inline_generate_summary): Likewise.
1679         (inline_analyze_function): Likewise.
1680         (do_estimate_growth): Likewise.
1681         (simple_edge_hints): Likewise.
1682         (estimate_node_size_and_time): Likewise.
1683         (estimate_edge_devirt_benefit): Likewise.
1684         (compute_inline_parameters): Likewise.
1685         (estimate_function_body_sizes): Likewise.
1686         (compute_bb_predicates): Likewise.
1687         (initialize_inline_failed): Likewise.
1688         (dump_inline_summary): Likewise.
1689         (dump_inline_edge_summary): Likewise.
1690         * ipa-inline-transform.c (inline_transform): Likewise.
1691         (preserve_function_body_p): Likewise.
1692         (save_inline_function_body): Likewise.
1693         (inline_call): Likewise.
1694         (clone_inlined_nodes): Likewise.
1695         (can_remove_node_now_p): Likewise.
1696         (can_remove_node_now_p_1): Likewise.
1697         * ipa-inline.c (early_inliner): Likewise.
1698         (early_inline_small_functions): Likewise.
1699         (inline_always_inline_functions): Likewise.
1700         (ipa_inline): Likewise.
1701         (flatten_function): Likewise.
1702         (inline_small_functions): Likewise.
1703         (speculation_useful_p): Likewise.
1704         (recursive_inlining): Likewise.
1705         (update_caller_keys): Likewise.
1706         (reset_edge_caches): Likewise.
1707         (update_edge_key): Likewise.
1708         (edge_badness): Likewise.
1709         (relative_time_benefit): Likewise.
1710         (want_inline_self_recursive_call_p): Likewise.
1711         (want_inline_small_function_p): Likewise.
1712         (want_early_inline_function_p): Likewise.
1713         (num_calls): Likewise.
1714         (can_early_inline_edge_p): Likewise.
1715         (can_inline_edge_p): Likewise.
1716         (report_inline_failed_reason): Likewise.
1717         * ipa-profile.c (ipa_profile): Likewise.
1718         (ipa_propagate_frequency): Likewise.
1719         (ipa_propagate_frequency_1): Likewise.
1720         (ipa_profile_generate_summary): Likewise.
1721         * ipa-prop.c (ipcp_transform_function): Likewise.
1722         (read_replacements_section): Likewise.
1723         (ipa_prop_read_section): Likewise.
1724         (ipa_modify_call_arguments): Likewise.
1725         (ipa_print_node_params): Likewise.
1726         (propagate_controlled_uses): Likewise.
1727         (update_indirect_edges_after_inlining): Likewise.
1728         (remove_described_reference): Likewise.
1729         (ipa_make_edge_direct_to_target): Likewise.
1730         (ipa_analyze_node): Likewise.
1731         (ipa_analyze_params_uses): Likewise.
1732         (ipa_compute_jump_functions): Likewise.
1733         (ipa_get_callee_param_type): Likewise.
1734         (ipa_print_node_jump_functions): Likewise.
1735         (ipa_initialize_node_params): Likewise.
1736         (ipa_populate_param_decls): Likewise.
1737         (ipa_func_spec_opts_forbid_analysis_p): Likewise.
1738         (write_agg_replacement_chain): Likewise.
1739         (ipa_write_node_info): Likewise.
1740         (ipa_edge_duplication_hook): Likewise.
1741         (try_decrement_rdesc_refcount): Likewise.
1742         * ipa-pure-const.c (propagate_nothrow): Likewise.
1743         (propagate_pure_const): Likewise.
1744         (pure_const_read_summary): Likewise.
1745         (pure_const_write_summary): Likewise.
1746         (analyze_function): Likewise.
1747         * ipa-ref-inline.h (ipa_ref_referred_ref_list): Likewise.
1748         (ipa_ref_referring_ref_list): Likewise.
1749         * ipa-ref.c (ipa_clear_stmts_in_references): Likewise.
1750         (ipa_remove_stmt_references): Likewise.
1751         (ipa_find_reference): Likewise.
1752         (ipa_dump_referring): Likewise.
1753         (ipa_dump_references): Likewise.
1754         (ipa_record_reference): Likewise.
1755         * ipa-reference.c (ipa_reference_read_optimization_summary): Likewise.
1756         (ipa_reference_write_optimization_summary): Likewise.
1757         (write_node_summary_p): Likewise.
1758         (propagate): Likewise.
1759         (read_write_all_from_decl): Likewise.
1760         (generate_summary): Likewise.
1761         (analyze_function): Likewise.
1762         (propagate_bits): Likewise.
1763         (ipa_reference_get_not_written_global): Likewise.
1764         (ipa_reference_get_not_read_global): Likewise.
1765         * ipa-split.c (execute_split_functions): Likewise.
1766         (split_function): Likewise.
1767         * ipa-utils.c (ipa_merge_profiles): Likewise.
1768         (dump_cgraph_node_set): Likewise.
1769         (ipa_reverse_postorder): Likewise.
1770         (ipa_edge_within_scc): Likewise.
1771         (ipa_get_nodes_in_cycle): Likewise.
1772         (ipa_free_postorder_info): Likewise.
1773         (ipa_reduced_postorder): Likewise.
1774         (searchc): Likewise.
1775         (recursive_call_p): Likewise.
1776         * ipa.c (ipa_cdtor_merge): Likewise.
1777         (record_cdtor_fn): Likewise.
1778         (function_and_variable_visibility): Likewise.
1779         (varpool_externally_visible_p): Likewise.
1780         (cgraph_externally_visible_p): Likewise.
1781         (comdat_can_be_unshared_p): Likewise.
1782         (comdat_can_be_unshared_p_1): Likewise.
1783         (address_taken_from_non_vtable_p): Likewise.
1784         (ipa_discover_readonly_nonaddressable_vars): Likewise.
1785         (symtab_remove_unreachable_nodes): Likewise.
1786         (walk_polymorphic_call_targets): Likewise.
1787         (process_references): Likewise.
1788         (enqueue_node): Likewise.
1789         (has_addr_references_p): Likewise.
1790         (cgraph_non_local_node_p_1): Likewise.
1791         * is-a.h (varpool_analyze_node): Likewise.
1792         * lto-cgraph.c (input_symtab): Likewise.
1793         (merge_profile_summaries): Likewise.
1794         (input_cgraph_1): Likewise.
1795         (input_edge): Likewise.
1796         (input_varpool_node): Likewise.
1797         (input_node): Likewise.
1798         (input_overwrite_node): Likewise.
1799         (compute_ltrans_boundary): Likewise.
1800         (output_refs): Likewise.
1801         (lto_output_varpool_node): Likewise.
1802         (lto_output_node): Likewise.
1803         (reachable_from_other_partition_p): Likewise.
1804         (referenced_from_other_partition_p): Likewise.
1805         (lto_output_edge): Likewise.
1806         (output_node_opt_summary): Likewise.
1807         (add_node_to): Likewise.
1808         (reachable_from_this_partition_p): Likewise.
1809         (lto_set_symtab_encoder_in_partition): Likewise.
1810         (lto_symtab_encoder_in_partition_p): Likewise.
1811         (lto_set_symtab_encoder_encode_initializer): Likewise.
1812         (lto_symtab_encoder_encode_initializer_p): Likewise.
1813         (lto_set_symtab_encoder_encode_body): Likewise.
1814         (lto_symtab_encoder_encode_body_p): Likewise.
1815         * lto-section-in.c (lto_free_function_in_decl_state_for_node):
1816         Likewise.
1817         * lto-streamer-in.c (lto_read_body): Likewise.
1818         (fixup_call_stmt_edges): Likewise.
1819         (fixup_call_stmt_edges_1): Likewise.
1820         * lto-streamer-out.c (produce_symtab): Likewise.
1821         (output_symbol_p): Likewise.
1822         (write_symbol): Likewise.
1823         (lto_output): Likewise.
1824         (copy_function): Likewise.
1825         (output_function): Likewise.
1826         * passes.c (function_called_by_processed_nodes_p): Likewise.
1827         (ipa_write_optimization_summaries): Likewise.
1828         (ipa_write_summaries): Likewise.
1829         (do_per_function_toporder): Likewise.
1830         (do_per_function): Likewise.
1831         (dump_passes): Likewise.
1832         * symtab.c (symtab_semantically_equivalent_p): Likewise.
1833         (symtab_nonoverwritable_alias): Likewise.
1834         (symtab_nonoverwritable_alias_1): Likewise.
1835         (symtab_for_node_and_aliases): Likewise.
1836         (symtab_resolve_alias): Likewise.
1837         (fixup_same_cpp_alias_visibility): Likewise.
1838         (symtab_alias_ultimate_target): Likewise.
1839         (symtab_used_from_object_file_p): Likewise.
1840         (verify_symtab_base): Likewise.
1841         (dump_symtab_base): Likewise.
1842         (symtab_node_name): Likewise.
1843         (symtab_node_asm_name): Likewise.
1844         (symtab_dissolve_same_comdat_group_list): Likewise.
1845         (symtab_add_to_same_comdat_group): Likewise.
1846         (symtab_unregister_node): Likewise.
1847         (symtab_insert_node_to_hashtable): Likewise.
1848         (symtab_register_node): Likewise.
1849         (unlink_from_assembler_name_hash): Likewise.
1850         (insert_to_assembler_name_hash): Likewise.
1851         (eq_assembler_name): Likewise.
1852         (hash_node_by_assembler_name): Likewise.
1853         (eq_node): Likewise.
1854         (hash_node): Likewise.
1855         * toplev.c (wrapup_global_declaration_2): Likewise.
1856         * trans-mem.c (ipa_tm_execute): Likewise.
1857         (ipa_tm_transform_clone): Likewise.
1858         (ipa_tm_transform_transaction): Likewise.
1859         (ipa_tm_transform_calls_redirect): Likewise.
1860         (ipa_tm_insert_gettmclone_call): Likewise.
1861         (ipa_tm_insert_irr_call): Likewise.
1862         (ipa_tm_create_version): Likewise.
1863         (ipa_tm_create_version_alias): Likewise.
1864         (ipa_tm_mark_forced_by_abi_node): Likewise.
1865         (ipa_tm_mark_force_output_node): Likewise.
1866         (ipa_tm_diagnose_tm_safe): Likewise.
1867         (ipa_tm_mayenterirr_function): Likewise.
1868         (ipa_tm_scan_irr_function): Likewise.
1869         (ipa_tm_note_irrevocable): Likewise.
1870         (ipa_tm_scan_calls_clone): Likewise.
1871         (get_cg_data): Likewise.
1872         * tree-eh.c (tree_could_trap_p): Likewise.
1873         * tree-emutls.c (ipa_lower_emutls): Likewise.
1874         (create_emultls_var): Likewise.
1875         (lower_emutls_function_body): Likewise.
1876         (gen_emutls_addr): Likewise.
1877         (emutls_decl): Likewise.
1878         (new_emutls_decl): Likewise.
1879         * tree-inline.c (tree_function_versioning): Likewise.
1880         (optimize_inline_calls): Likewise.
1881         (expand_call_inline): Likewise.
1882         (estimate_num_insns): Likewise.
1883         (copy_bb): Likewise.
1884         (delete_unreachable_blocks_update_callgraph): Likewise.
1885         * tree-nested.c (gimplify_all_functions): Likewise.
1886         (create_nesting_tree): Likewise.
1887         (check_for_nested_with_variably_modified): Likewise.
1888         * tree-pretty-print.c (dump_function_header): Likewise.
1889         * tree-profile.c (tree_profiling): Likewise.
1890         * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
1891         (modify_function): Likewise.
1892         (convert_callers): Likewise.
1893         (convert_callers_for_node): Likewise.
1894         * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
1895         (associate_varinfo_to_alias): Likewise.
1896         (create_variable_info_for): Likewise.
1897         (get_constraint_for_ssa_var): Likewise.
1898         * tree-vectorizer.c (increase_alignment): Likewise.
1899         * tree.c (find_decls_types_in_var): Likewise.
1900         (find_decls_types_in_node): Likewise.
1901         (free_lang_data_in_decl): Likewise.
1902         * value-prof.c (gimple_ic_transform): Likewise.
1903         (gimple_ic): Likewise.
1904         (check_ic_target): Likewise.
1905         (init_node_map): Likewise.
1906         * varasm.c (decl_binds_to_current_def_p): Likewise.
1907         (default_binds_local_p_1): Likewise.
1908         (dump_tm_clone_pairs): Likewise.
1909         (assemble_alias): Likewise.
1910         (find_decl): Likewise.
1911         (mark_decl_referenced): Likewise.
1912         * varpool.c (varpool_for_node_and_aliases): Likewise.
1913         (varpool_extra_name_alias): Likewise.
1914         (varpool_create_variable_alias): Likewise.
1915         (add_new_static_var): Likewise.
1916         (varpool_finalize_named_section_flags): Likewise.
1917         (varpool_remove_unreferenced_decls): Likewise.
1918         (enqueue_node): Likewise.
1919         (varpool_assemble_decl): Likewise.
1920         (assemble_aliases): Likewise.
1921         (varpool_analyze_node): Likewise.
1922         (cgraph_variable_initializer_availability): Likewise.
1923         (varpool_add_new_variable): Likewise.
1924         (ctor_for_folding): Likewise.
1925         (dump_varpool_node): Likewise.
1926         (varpool_remove_initializer): Likewise.
1927         (varpool_remove_node): Likewise.
1928         (varpool_node_for_decl): Likewise.
1929         (varpool_create_empty_node): Likewise.
1930         * config/i386/i386.c (ix86_generate_version_dispatcher_body): Likewise.
1931         (ix86_get_function_versions_dispatcher): Likewise.
1933 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
1935         * cgraph.h (symtab_node_base): Convert to a class;
1936         add GTY((desc ("%h.type"), tag ("SYMTAB_SYMBOL"))), and take
1937         chain_next/prev from symtab_node_def.
1938         (cgraph_node): Inherit from symtab_node; add GTY option
1939         tag ("SYMTAB_FUNCTION").
1940         (varpool_node): Inherit from symtab_node; add GTY option
1941         tag ("SYMTAB_VARIABLE").
1942         (symtab_node_def): Remove.
1943         (is_a_helper <cgraph_node>::test (symtab_node_def *)): Convert to...
1944         (is_a_helper <cgraph_node>::test (symtab_node_base *)): ...this.
1945         (is_a_helper <varpool_node>::test (symtab_node_def *)): Convert to...
1946         (is_a_helper <varpool_node>::test (symtab_node_base *)): ...this.
1948         * ipa-ref.h (symtab_node_def): Drop.
1949         (symtab_node): Change underlying type from symtab_node_def to
1950         symtab_node_base.
1951         (const_symtab_node): Likwise.
1953         * is-a.h: Update examples in comment.
1955         * symtab.c (symtab_hash): Change symtab_node_def to symtab_node_base.
1956         (assembler_name_hash): Likewise.
1958 2013-10-29  Martin Liska  <marxin.liska@gmail.com>
1960         * doc/tree-ssa.texi (gimple_phi_result): Document.
1961         (gimple_phi_num_args, gimple_phi_arg): Likewise.
1962         (gimple_phi_arg_edge, gimple_phi_arg_def): Likewise.
1963         (PHI_RESULT, PHI_NUM_ARGS): Remove.
1964         (PHI_ARG_ELT, PHI_ARG_EDGE, PHI_ARG_DEF): Likewise.
1966 2013-10-29  Andrew MacLeod  <amacleod@redhat.com>
1968         * expr.h: Revert change and include tree-core.h.
1969         * rtl.h: Revert change and don't include tree-core.h.
1971 2013-10-29  Andrew MacLeod  <amacleod@redhat.com>
1973         * config/darwin.c: Include gimple.h.
1974         * config/i386/winnt.c: Likewise.
1976 2013-10-29  Marc Glisse  <marc.glisse@inria.fr>
1978         PR tree-optimization/19831
1979         * tree-ssa-alias.c (stmt_kills_ref_p_1): Handle BUILT_IN_FREE.
1981 2013-10-29  Andrew MacLeod  <amacleod@redhat.com>
1983         * tree-outof-ssa.h: Remove include files.
1984         * tree-outof-ssa.c: Add required include files from tree-outof-ssa.h.
1985         * expr.c: Likewise.
1986         * tree-ssa-coalesce.c: Likewise.
1987         * cfgexpand.c: Likewise.
1988         * tree-ssa-ter.c: Likewise.
1989         * ipa-prop.h: Remove gimple.h and tree-core.h from include list.
1990         * lto-streamer.h: Likewise.
1991         * cgraphbuild.c: Add gimple.h to include list.
1992         * data-streamer-in.c: Likewise.
1993         * ipa-cp.c: Likewise.
1994         * tree-streamer.c: Likewise.
1995         * lto-compress.c: Likewise.
1996         * ipa-reference.c: Likewise.
1997         * data-streamer-out.c: Likewise.
1998         * lto-cgraph.c: Likewise.
1999         * cgraphclones.c: Likewise.
2000         * ipa-utils.c: Likewise.
2001         * data-streamer.c: Likewise.
2002         * ipa-split.c: Likewise.
2003         * lto-section-in.c: Likewise.
2004         * tree-streamer-out.c: Likewise.
2005         * ipa-prop.c: Likewise.
2006         * tree-streamer-in.c: Likewise.
2007         * symtab.c: Likewise.
2008         * opts-global.c: Likewise.
2009         * lto-opts.c: Likewise.
2010         * lto-section-out.c: Likewise.
2011         * lto-streamer.c: Likewise.
2012         * rtl.h: Add tree-core.h to include list.
2013         * expr.h: Remove tree-core.h from include list.
2014         * gimple.h: Likewise.
2015         * ipa-utils.h: Likewise.
2016         * streamer-hooks.h: Likewise.
2017         * streamer-hooks.c: Include input.h.
2019 2013-10-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2021         * config/arm/arm.c (cortexa7_extra_costs): New table.
2022         (arm_cortex_a7_tune): New.
2023         * config/arm/arm-cores.def: Use cortex_a7 tuning for cortex-a7.
2025 2013-10-29  Eric Botcazou  <ebotcazou@adacore.com>
2027         * expr.c (expand_expr_real_1) <MEM_EXPR>: Eliminate small redundancy.
2029 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
2031         * doc/gty.texi ("Inheritance and GTY"): Make it clear that
2032         to use autogenerated markers for a class-hierarchy, every class
2033         must have a GTY marker.
2034         * gengtype.h (struct type): Add linked list of subclasses to
2035         the "s" member of the union.
2036         (add_subclass): New decl.
2037         * gengtype-state.c (read_state_struct_type): Set up subclass
2038         linked list.
2039         * gengtype.c (get_ultimate_base_class): New.
2040         (add_subclass): New.
2041         (new_structure): Set up subclass linked list.
2042         (set_gc_used_type): Propagate usage information to subclasses.
2043         (output_mangled_typename): Use get_ultimate_base_class.
2044         (walk_subclasses): Use the subclass linked list, avoiding an
2045         O(N^2) when writing out all types.
2046         (walk_type): Issue an error if the base class is missing a tag,
2047         rather than generating bogus C code.  Add a gcc_unreachable
2048         default case, in case people omit tags from concrete subclasses,
2049         or get the values wrong.
2050         (write_func_for_structure): Issue an error for subclasses for
2051         which the base doesn't have a "desc", since otherwise the
2052         autogenerated routines for the base would silently fail to visit
2053         any subclass fields.
2054         (write_root): Use get_ultimate_base_class, tweaking constness of
2055         tp to match that function's signature.
2057 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
2059         * doc/gty.texi (GTY Options): Add note about inheritance to
2060         description of desc and tag.
2061         (Inheritance and GTY): New.
2063 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
2065         * gengtype-parse.c (opts_have): Drop "static" so that
2066         we can use this from gengtype.c.
2067         * gengtype.c (set_gc_used_type): Mark any base class as used;
2068         update field traversal to visit inherited fields.
2069         (output_mangled_typename):  Convert references to classes within
2070         an inheritance hierarchy to reference the ultimate base class,
2071         since only it will have gt_ functions.
2072         (get_string_option): New.
2073         (walk_subclasses): New.
2074         (walk_type): Treat GTY structs that have a "desc" as being the
2075         root of an inheritance hierarchy.  Generate a switch on it
2076         within the marking function which walks all subclasses, adding
2077         cases for them via walk_subclasses.  For subclasses, visit all
2078         fields of the type (including inherited ones).
2079         (write_func_for_structure): Don't write fns for subclasses, only
2080         for the ultimate base class within an inheritance hierarchy.
2081         Subclasses-marking will be handled by the base class marking functions.
2082         (write_types): Likewise.
2083         (write_local_func_for_structure): Likewise.
2084         (USED_BY_TYPED_GC_P): Emit allocators for subclasses that have
2085         a "tag" option (and are thus concrete subclasses).
2086         (write_root): Use the marker function for the ultimate base class.
2087         * gengtype.h (FOR_ALL_INHERITED_FIELDS): New.
2088         (opts_have): Add declaration.
2090 2013-10-28  Vladimir Makarov  <vmakarov@redhat.com>
2092         * lra-spills.c (lra_final_code_change): Remove useless move insns
2093         originated from moves of pseudos.
2095 2013-10-28  Jeff Law  <law@redhat.com>
2097         * lower-subreg.c (resolve_simple_move): Fix comment typo.
2099 2013-10-28  Trevor Saunders  <tsaunders@mozilla.com>
2101         * df-scan.c (df_collection_rec): Adjust.
2102         (copy_defs): New constant.
2103         (copy_uses): Likewise.
2104         (copy_eq_uses): Likewise.
2105         (copy_mw): Likewise.
2106         (copy_all): Likewise.
2107         (df_insn_rescan): Adjust.
2108         (df_notes_rescan): Likewise.
2109         (df_swap_refs): Likewise.
2110         (df_sort_and_compress_refs): Likewise.
2111         (df_sort_and_compress_mws): Likewise.
2112         (df_install_refs): Likewise.
2113         (df_install_mws): Likewise.
2114         (df_refs_add_to_chains): Add flags parameter controlling which vectors
2115         are coppied.
2116         (df_bb_refs_record): Adjust.
2117         (df_record_entry_block_defs): Likewise.
2118         (df_record_exit_block_defs): Likewise.
2119         (df_refs_verify): Likewise.
2120         (df_mws_verify): Likewise.
2121         (df_insn_refs_verify): Likewise.
2122         (df_bb_verify): Likewise.
2123         * ipa-pure-const.c (finish_state): Remove.
2124         (propagate): Adjust.
2125         * tree-data-ref.c tree-ssa-alias.c tree-ssa-loop-ivcanon.c
2126         tree-ssa-threadedge.c tree-vect-loop-manip.c tree-vect-slp.c
2127         var-tracking.c: Adjust.
2128         * vec.c (stack_vecs): Remove.
2129         (register_stack_vec): Likewise.
2130         (stack_vec_register_index): Likewise.
2131         (unregister_stack_vec): Likewise.
2132         * vec.h (struct va_stack): Remove.
2133         (struct vec<T, A, vl_ptr>): Specialize as
2134         struct vec<T, va_heap, vl_ptr> instead since va_heap is the only
2135         allocation strategy compatable with the vl_ptr layout.
2136         (struct vec<T, va_gc, vl_ptr>): Remove because it now gets an empty
2137         specialization anyway.
2138         (class stack_vec): New class.
2139         (vec_stack_alloc): Remove.
2140         (vec<T, va_heap, vl_ptr>::using_auto_storage): New method.
2142 2013-10-28  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2143             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2144             Sergey Lega  <sergey.s.lega@intel.com>
2145             Anna Tikhonova  <anna.tikhonova@intel.com>
2146             Ilya Tocar  <ilya.tocar@intel.com>
2147             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2148             Ilya Verbin  <ilya.verbin@intel.com>
2149             Kirill Yukhin  <kirill.yukhin@intel.com>
2150             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2152         * config/i386/i386.md (prefetch): Allow TARGET_AVX512PF.
2153         (*prefetch_avx512pf_<mode>): New.
2154         * config/i386/sse.md (avx512f_vmcmp<mode>3): Ditto.
2155         (avx512f_maskcmp<mode>3): Ditto.
2156         (vashrv16si3): Ditto.
2158 2013-10-28  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2159             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2160             Sergey Lega  <sergey.s.lega@intel.com>
2161             Anna Tikhonova  <anna.tikhonova@intel.com>
2162             Ilya Tocar  <ilya.tocar@intel.com>
2163             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2164             Ilya Verbin  <ilya.verbin@intel.com>
2165             Kirill Yukhin  <kirill.yukhin@intel.com>
2166             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2168         * config/i386/i386.md (any_truncate): New.
2169         (trunsuffix): Ditto.
2170         * config/i386/predicates.md (const_8_to_9_operand): New.
2171         (const_10_to_11_operand): Ditto.
2172         (const_12_to_13_operand): Ditto.
2173         (const_14_to_15_operand): Ditto.
2174         (const_16_to_19_operand): Ditto.
2175         (const_20_to_23_operand): Ditto.
2176         (const_24_to_27_operand): Ditto.
2177         (const_28_to_31_operand): Ditto.
2178         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_FIX_NOTRUNC.
2179         (cvtusi2<ssescalarmodesuffix>32): New.
2180         (cvtusi2<ssescalarmodesuffix>64): Ditto.
2181         (ufloatv16siv16sf2): Ditto.
2182         (avx512f_fix_notruncv16sfv16si): Ditto.
2183         (avx512f_ufix_notruncv16sfv16si): Ditto.
2184         (avx512f_vcvtss2usi): Ditto.
2185         (avx512f_vcvtss2usiq): Ditto.
2186         (avx512f_vcvttss2usi): Ditto.
2187         (avx512f_vcvttss2usiq): Ditto.
2188         (avx512f_vcvtsd2usi): Ditto.
2189         (avx512f_vcvtsd2usiq): Ditto.
2190         (avx512f_vcvttsd2usi): Ditto.
2191         (avx512f_vcvttsd2usiq): Ditto.
2192         (ufloatv8siv8df): Ditto.
2193         (avx512f_cvtdq2pd512_2): Ditto.
2194         (avx512f_cvtpd2dq512): Ditto.
2195         (avx512f_ufix_notruncv8dfv8si): Ditto.
2196         (avx512f_cvtpd2ps512): Ditto.
2197         (vec_unpacks_lo_v16sf): Ditto.
2198         (vec_unpacks_hi_v16sf): Ditto.
2199         (vec_unpacks_float_hi_v16si): Ditto.
2200         (vec_unpacks_float_lo_v16si): Ditto.
2201         (avx512f_unpckhps512): Ditto.
2202         (avx512f_unpcklps512): Ditto.
2203         (avx512f_movshdup512): Ditto.
2204         (avx512f_movsldup512): Ditto.
2205         (vec_extract_lo_v32hi): Ditto.
2206         (vec_extract_hi_v32hi): Ditto.
2207         (vec_extract_lo_v64qi): Ditto.
2208         (vec_extract_hi_v64qi): Ditto.
2209         (avx512f_unpckhpd512): Ditto.
2210         (avx512f_movddup512): Ditto.
2211         (avx512f_unpcklpd512): Ditto.
2212         (*avx512f_unpcklpd512): Ditto.
2213         (avx512f_shufps512_1): Ditto.
2214         (avx512f_shufpd512_1): Ditto.
2215         (avx512f_interleave_highv8di): Ditto.
2216         (avx512f_interleave_lowv8di): Ditto.
2217         (PMOV_DST_MODE): Ditto.
2218         (pmov_src_mode): Ditto.
2219         (pmov_src_lower): Ditto.
2220         (pmov_suff): Ditto.
2221         (*avx512f_<code><pmov_src_lower><mode>2): Ditto.
2222         (*avx512f_<code>v8div16qi2): Ditto.
2223         (*avx512f_<code>v8div16qi2_store): Ditto.
2224         (vec_widen_umult_even_v16si): Ditto.
2225         (*vec_widen_umult_even_v16si): Ditto.
2226         (vec_widen_smult_even_v16si): Ditto.
2227         (*vec_widen_smult_even_v16si): Ditto.
2228         (avx512f_interleave_highv16si): Ditto.
2229         (avx512f_interleave_lowv16si): Ditto.
2230         (avx512f_<code>v16qiv16si2): Ditto.
2231         (avx512f_<code>v16hiv16si2): Ditto.
2232         (avx512f_<code>v8qiv8di2): Ditto.
2233         (avx512f_<code>v8hiv8di2): Ditto.
2234         (avx512f_<code>v8siv8di2): Ditto.
2235         (avx512cd_maskb_vec_dupv8di): Ditto.
2236         (avx512cd_maskw_vec_dupv16si): Ditto.
2237         (avx512f_vcvtph2ps512): Ditto.
2238         (avx512f_vcvtps2ph512): Ditto.
2239         (VEC_EXTRACT_MODE): Extened with wider modes.
2240         (VEC_PERM_AVX2): Ditto.
2241         (VEC_PERM_CONST): Ditto.
2243 2013-10-28  Joern Rennecke  <joern.rennecke@embecosm.com>
2245         * config/arc/arc.c (arc_ccfsm_post_advance):
2246         Add comment about TYPE_RETURN.
2248 2013-10-28  Bin Cheng  <bin.cheng@arm.com>
2250         * tree-ssa-loop-ivopts.c (strip_offset_1): Change parameter type.
2251         Count DECL_FIELD_BIT_OFFSET for COMPONENT_REF.
2252         (strip_offset): Convert offset to unsigned number.
2254 2013-10-27  Tom de Vries  <tom@codesourcery.com>
2256         * cfgexpand.c (gimple_expand_cfg): Remove test for parm_birth_insn.
2257         Don't commit insertions after NOTE_INSN_FUNCTION_BEG.
2259 2013-10-27  Oleg Endo  <olegendo@gcc.gnu.org>
2261         * config/sh/sh.c (MSW, LSW): Move and rename macros to...
2262         * config/sh/sh.h (SH_REG_MSW_OFFSET, SH_REG_LSW_OFFSET): ... here.
2263         (TARGET_BIG_ENDIAN): New macro.
2264         * config/sh/sh.md: Use it instead of !TARGET_LITTLE_ENDIAN.
2265         Use SH_REG_MSW_OFFSET and SH_REG_LSW_OFFSET.
2266         * config/sh/sh.c: Likewise.
2267         * config/sh/sh.h: Likewise.
2269 2013-10-27  Hans-Peter Nilsson  <hp@axis.com>
2271         * config/cris/cris.c (cris_emit_trap_for_misalignment): Replace the
2272         removed PRED_MUDFLAP with PRED_NORETURN.  Correct file-path in comment.
2274 2013-10-26  Oleg Endo  <olegendo@gcc.gnu.org>
2276         * config/sh/sh.md (movmemsi): Remove empty constraints and predicates.
2277         Fix formatting.
2278         (cmpstr_t, cmpstrsi): Fix formatting.
2280 2013-10-26  Oleg Endo  <olegendo@gcc.gnu.org>
2282         PR target/52483
2283         * config/sh/predicates.md (general_movdst_operand): Allow reg+reg
2284         addressing, do not use general_operand for memory operands.
2286 2013-10-26  Vladimir Makarov  <vmakarov@redhat.com>
2288         Revert:
2289         2013-10-25  Vladimir Makarov  <vmakarov@redhat.com>
2290         * lra-spills.c (lra_final_code_change): Remove useless move insns.
2292 2013-10-26  Jeff Law  <law@redhat.com>
2294         * predict.c (PRED_MUDFLAP): Remove.
2295         * targhooks.c (build_va_arg_indirect_ref): Remove mudflap support.
2297         * Makefile.in (C_COMMON_OBJS): Remove tree-mudflap.
2298         (OBJS): Remove tree-nomudflap.o
2299         (GTFILES): Remove tree-mudflap.c
2300         * builtins.c (expand_builtin_alloc): Remove mudflap support.
2301         * gcc.c (MFWRAP_SPEC, MFLIB_SPEC): Likewise.
2302         (mfwrap_spec, mflib_spec): Likewise.
2303         (cpp_unique_options, cc1_options, static_specs): Likewise.
2304         * gimplify (gimplify_vla_decl, build_va_arg_indirect_ref): Likewise.
2305         * passes.def: Likewise.
2306         * toplev.c (compile_file, process_options): Likewise.
2307         * tree-inline.c (copy_tree_r): Likewise.
2308         * tree-pass.,h (make_pass_mudflap_1, make_pass_mudflap_2): Likewise.
2309         * varasm.c (make_decl_rtl, make_decl_rtl_for_debug): Likewise.
2310         (build_constant_desc, output_constant_def_contents): Likewise.
2311         (categorize_decl_for_section): Likewise.
2312         * tree-mudflap.c: Removed.
2313         * tree-mudflap.h: Removed.
2314         * tree-nomudflap.c: Removed.
2315         * bfin/uclinux.h (MFWRAP_SPEC): Remove.
2316         * moxie/uclinux.h (MFWRAP_SPEC): Likewise.
2317         * rs6000/aix.h (MFWRAP_SPEC, MFLIB_SPEC): Likewise.
2318         * config/sol2.h (MFLIB_SPEC): Likewise.
2319         * doc/install.texi: Remove mudflap references.
2320         * doc/passes.texi: Similarly.
2321         * doc/sourcebuild.texi: Similarly.
2322         * doc/invoke.texi: Remove mudlfap related options.
2324 2013-10-25  Vladimir Makarov  <vmakarov@redhat.com>
2326         PR rtl-optimization/58759
2327         * lra-constraints.c (lra_constraints): Remove wrong condition to
2328         remove insn setting up an equivalent pseudo.
2330 2013-10-25  Vladimir Makarov  <vmakarov@redhat.com>
2332         * config/rs6000/rs6000-protos.h
2333         (rs6000_secondary_memory_needed_mode): New prototype.
2334         * config/rs6000/rs6000.c: Include ira.h.
2335         (TARGET_LRA_P): Redefine.
2336         (rs6000_legitimate_offset_address_p): Call
2337         legitimate_constant_pool_address_p in strict mode for LRA.
2338         (rs6000_legitimate_address_p): Ditto.
2339         (legitimate_lo_sum_address_p): Add code for LRA.  Use lra_in_progress.
2340         (rs6000_emit_move): Add LRA version of code to generate load/store
2341         of SDmode values.
2342         (rs6000_secondary_memory_needed_mode): New.
2343         (rs6000_alloc_sdmode_stack_slot): Do nothing for LRA.
2344         (rs6000_secondary_reload_class): Return NO_REGS for LRA for
2345         constants, memory, and FP registers.
2346         (rs6000_lra_p): New.
2347         * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): New macro.
2348         * config/rs6000/rs6000.opt (mlra): New option.
2349         * lra-spills.c (lra_final_code_change): Remove useless move insns.
2351 2013-10-25  Yufeng Zhang  <yufeng.zhang@arm.com>
2353         * tree-ssa-math-opts.c (convert_plusminus_to_widen): Call
2354         has_single_use () and not do the conversion if has_single_use ()
2355         returns false for the multiplication result.
2357 2013-10-25  David Malcolm  <dmalcolm@redhat.com>
2359         * tree.h (EXCEPTIONAL_CLASS_P): Rename parameter from "CODE"
2360         to "NODE", since this works on a "tree", not an
2361         "enum tree_code".
2362         (CONSTANT_CLASS_P): Likewise.
2363         (TYPE_P): Likewise.
2364         (DECL_P): Likewise.
2365         (INDIRECT_REF_P): Likewise.
2366         (REFERENCE_CLASS_P): Likewise.
2367         (COMPARISON_CLASS_P): Likewise.
2368         (UNARY_CLASS_P): Likewise.
2369         (BINARY_CLASS_P): Likewise.
2370         (STATEMENT_CLASS_P): Likewise.
2371         (VL_EXP_CLASS_P): Likewise.
2372         (EXPRESSION_CLASS_P): Likewise.
2373         (IS_TYPE_OR_DECL_P): Likewise.
2375 2013-10-25  Marc Glisse  <marc.glisse@inria.fr>
2377         * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Look for an
2378         ADDR_EXPR in the defining statement.
2380 2013-10-25  Richard Biener  <rguenther@suse.de>
2382         PR tree-optimization/58626
2383         * tree-loop-distribution.c (enum rdg_dep_type): Remove
2384         anti_dd, output_dd and input_dd.
2385         (struct rdg_edge): Remove level and relation members.
2386         (RDGE_LEVEL, RDGE_RELATION): Remove.
2387         (dot_rdg_1): Adjust.
2388         (create_rdg_edge_for_ddr): Remove.
2389         (create_rdg_edges_for_scalar): Adjust.
2390         (create_edge_for_control_dependence): Likewise.
2391         (create_rdg_edges): Split into ...
2392         (create_rdg_flow_edges): ... this
2393         (create_rdg_cd_edges): ... and this.
2394         (free_rdg): Adjust.
2395         (build_rdg): Likewise, do not compute data dependences or
2396         add edges for them.
2397         (pg_add_dependence_edges): New function.
2398         (pgcmp): Likewise.
2399         (distribute_loop): First apply all non-dependence based
2400         partition mergings.  Then compute dependences between partitions
2401         and merge and order partitions according to them.
2403 2013-10-25  Eric Botcazou  <ebotcazou@adacore.com>
2405         PR rtl-optimization/58831
2406         * alias.c (init_alias_analysis): At the beginning of each iteration,
2407         set the reg_seen[N] bit if static_reg_base_value[N] is non-null.
2409 2013-10-25  Eric Botcazou  <ebotcazou@adacore.com>
2411         * recog.c (search_ofs): New static variable moved from...
2412         (peep2_find_free_register): ...here.
2413         (peephole2_optimize): Initialize it.
2415 2013-10-25  Tobias Burnus  <burnus@net-b.de>
2417         * doc/invoke.texi (fopenmp): Change supported OpenMP version to 4.0.
2419 2013-10-25  Uros Bizjak  <ubizjak@gmail.com>
2421         * config/i386/i386.h (TARGET_MPX): New define.
2422         (TARGET_MPX_P): Ditto.
2424 2013-10-24  Ilya Enkovich  <ilya.enkovich@intel.com>
2426         * config/i386/constraints.md (B): New.
2427         (Ti): New.
2428         (Tb): New.
2429         * config/i386/i386-c.c (ix86_target_macros_internal): Add __MPX__.
2430         * config/i386/i386-modes.def (BND32): New.
2431         (BND64): New.
2432         * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): New.
2433         * config/i386/i386.c (isa_opts): Add mmpx.
2434         (regclass_map): Add bound registers.
2435         (dbx_register_map): Likewise.
2436         (dbx64_register_map): Likewise.
2437         (svr4_dbx_register_map): Likewise.
2438         (PTA_MPX): New.
2439         (ix86_option_override_internal): Support MPX ISA.
2440         (ix86_conditional_register_usage): Support bound registers.
2441         (print_reg): Likewise.
2442         (ix86_code_end): Add MPX bnd prefix.
2443         (output_set_got): Likewise.
2444         (ix86_output_call_insn): Likewise.
2445         (ix86_print_operand): Add '!' (MPX bnd) print prefix support.
2446         (ix86_print_operand_punct_valid_p): Likewise.
2447         (ix86_print_operand_address): Support UNSPEC_BNDMK_ADDR and
2448         UNSPEC_BNDMK_ADDR.
2449         (ix86_class_likely_spilled_p): Add bound regs support.
2450         (ix86_hard_regno_mode_ok): Likewise.
2451         (x86_order_regs_for_local_alloc): Likewise.
2452         (ix86_bnd_prefixed_insn_p): New.
2453         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Fix to new value.
2454         (FIXED_REGISTERS): Add bound registers.
2455         (CALL_USED_REGISTERS): Likewise.
2456         (REG_ALLOC_ORDER): Likewise.
2457         (HARD_REGNO_NREGS): Likewise.
2458         (TARGET_MPX): New.
2459         (VALID_BND_REG_MODE): New.
2460         (FIRST_BND_REG): New.
2461         (LAST_BND_REG): New.
2462         (reg_class): Add BND_REGS.
2463         (REG_CLASS_NAMES): Likewise.
2464         (REG_CLASS_CONTENTS): Likewise.
2465         (BND_REGNO_P): New.
2466         (ANY_BND_REG_P): New.
2467         (BNDmode): New.
2468         (HI_REGISTER_NAMES): Add bound registers.
2469         * config/i386/i386.md (UNSPEC_BNDMK): New.
2470         (UNSPEC_BNDMK_ADDR): New.
2471         (UNSPEC_BNDSTX): New.
2472         (UNSPEC_BNDLDX): New.
2473         (UNSPEC_BNDLDX_ADDR): New.
2474         (UNSPEC_BNDCL): New.
2475         (UNSPEC_BNDCU): New.
2476         (UNSPEC_BNDCN): New.
2477         (UNSPEC_MPX_FENCE): New.
2478         (BND0_REG): New.
2479         (BND1_REG): New.
2480         (type): Add mpxmov, mpxmk, mpxchk, mpxld, mpxst.
2481         (length_immediate): Likewise.
2482         (prefix_0f): Likewise.
2483         (memory): Likewise.
2484         (prefix_rep): Check for bnd prefix.
2485         (length_nobnd): New.
2486         (length): Use length_nobnd if specified.
2487         (BND): New.
2488         (bnd_ptr): New.
2489         (BNDCHECK): New.
2490         (bndcheck): New.
2491         (*jcc_1): Add bnd prefix and rename length attr to length_nobnd.
2492         (*jcc_2): Likewise.
2493         (jump): Likewise.
2494         (simple_return_internal): Likewise.
2495         (simple_return_pop_internal): Likewise.
2496         (*indirect_jump): Add MPX bnd prefix.
2497         (*tablejump_1): Likewise.
2498         (simple_return_internal_long): Likewise.
2499         (simple_return_indirect_internal): Likewise.
2500         (<mode>_mk): New.
2501         (*<mode>_mk): New.
2502         (mov<mode>): New.
2503         (*mov<mode>_internal_mpx): New.
2504         (<mode>_<bndcheck>): New.
2505         (*<mode>_<bndcheck>): New.
2506         (<mode>_ldx): New.
2507         (*<mode>_ldx): New.
2508         (<mode>_stx): New.
2509         (*<mode>_stx): New.
2510         * config/i386/predicates.md (lea_address_operand): Rename to...
2511         (address_no_seg_operand): ... this.
2512         (address_mpx_no_base_operand): New.
2513         (address_mpx_no_index_operand): New.
2514         (bnd_mem_operator): New.
2515         * config/i386/i386.opt (mmpx): New.
2516         * doc/invoke.texi: Add documentation for the flags -mmpx, -mno-mpx.
2517         * doc/rtl.texi Add documentation for BND32mode and BND64mode.
2519 2013-10-24  Ilya Enkovich  <ilya.enkovich@intel.com>
2521         * mode-classes.def (MODE_POINTER_BOUNDS): New.
2522         * tree.def (POINTER_BOUNDS_TYPE): New.
2523         * genmodes.c (complete_mode): Support MODE_POINTER_BOUNDS.
2524         (POINTER_BOUNDS_MODE): New.
2525         (make_pointer_bounds_mode): New.
2526         * machmode.h (POINTER_BOUNDS_MODE_P): New.
2527         * stor-layout.c (int_mode_for_mode): Support MODE_POINTER_BOUNDS.
2528         (layout_type): Support POINTER_BOUNDS_TYPE.
2529         * tree-pretty-print.c (dump_generic_node): Support POINTER_BOUNDS_TYPE.
2530         * tree.c (build_int_cst_wide): Support POINTER_BOUNDS_TYPE.
2531         (type_contains_placeholder_1): Likewise.
2532         * tree.h (POINTER_BOUNDS_TYPE_P): New.
2533         * varasm.c (output_constant): Support POINTER_BOUNDS_TYPE.
2534         * doc/rtl.texi (MODE_POINTER_BOUNDS): New.
2536 2013-10-24  Igor Shevlyakov  <igor.shevlyakov@gmail.com>
2538         * expr.c (expand_expr_real_1): Use mode of memory reference rather than
2539         mode of address computation when calling memory_address_addr_space.
2541 2013-08-24  Richard Henderson  <rth@twiddle.net>
2543         PR rtl/58542
2544         * optabs.c (maybe_emit_atomic_exchange): Use create_input_operand
2545         instead of create_convert_operand_to.
2546         (maybe_emit_sync_lock_test_and_set): Likewise.
2547         (expand_atomic_compare_and_swap): Likewise.
2548         (maybe_emit_compare_and_swap_exchange_loop): Don't convert_modes.
2550 2013-08-24  Sriraman Tallam  <tmsriram@google.com>
2552         * cgraph.c (cgraph_fnver_htab): Move GTY((...)) to be before htab_t.
2553         Change param_is to use the struct and not the pointer to the struct.
2555 2013-10-24  Andrew MacLeod  <amacleod@redhat.com>
2557         * builtins.c (dummy_object, gimplify_va_arg_expr): Move to gimplify.c.
2558         * gimplify.c (build_va_arg_indirect_ref, std_gimplify_va_arg_expr):
2559         Move to targhooks.c.
2560         (dummy_object, gimplify_va_arg_expr): Relocate from builtins.c.
2561         * targhooks.c (build_va_arg_indirect_ref, std_gimplify_va_arg_expr):
2562         Relocate from gimplify.c.
2563         * targhooks.h: Add 2 prototypes.
2564         * tree.h: Delete 2 prototypes.
2566 2013-10-24  Igor Shevlyakov  <igor.shevlyakov@gmail.com>
2568         * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p ): Check both
2569         [reg+mult*reg] and [mult*reg] to determine if multiplier is allowed.
2571 2013-10-24  Cong Hou  <congh@google.com>
2573         * convert.c (convert_to_real): Guard those unsafe math function
2574         convertions with flag_unsafe_math_optimizations.  Handle sqrt()
2575         specially.
2577 2013-10-24  Markus Trippelsdorf  <markus@trippelsdorf.de>
2579         PR ipa/58712
2580         * cgraph.c (cgraph_create_edge_1): Add indirect_unknown_callee
2581         as argument.
2582         (cgraph_create_edge): Use the new argument.
2583         (cgraph_create_indirect_edge): Likewise.
2585 2013-10-24  Joern Rennecke  <joern.rennecke@embecosm.com>
2587         * config/arc/arc.c (arc_ccfsm_post_advance): Also handle
2588         TYPE_UNCOND_BRANCH.
2589         (arc_ifcvt) <case 1 and 2>: Check that arc_ccfsm_post_advance
2590         changes statep->state.
2592 2013-10-24  Tobias Burnus  <burnus@net-b.de>
2594         PR other/33426
2595         * tree-cfg.c (replace_loop_annotate): New function.
2596         (execute_build_cfg): Call it.
2597         * gimplify.c (gimple_boolify, gimplify_expr): Handle ANNOTATE_EXPR.
2598         * internal-fn.c (expand_ANNOTATE): New function.
2599         * internal-fn.def (ANNOTATE): Define as new internal function.
2600         * tree-core.h (tree_node_kind): Add annot_expr_ivdep_kind.
2601         * tree-pretty-print.c (dump_generic_node): Handle ANNOTATE_EXPR.
2602         * tree.def (ANNOTATE_EXPR): New DEFTREECODE.
2603         * doc/extend.texi (Pragmas): Document #pragma ivdep.
2604         * doc/generic.texi (Expressions): Document ANNOTATE_EXPR.
2606 2013-10-17  Ian Bolton  <ian.bolton@arm.com>
2607             Marcus Shawcroft  <marcus.shawcroft@arm.com>
2609         * config/aarch64/aarch64.c (aarch64_preferred_reload_class):
2610         Special case reload SP+C into none GENERAL_REGS.
2612 2013-10-24  Michael Matz  <matz@suse.de>
2614         * gengtype.c (is_file_equal): Check that files will be same length.
2616 2013-10-25  Christian Bruel  <christian.bruel@st.com>
2618         * config.gcc (sh-*): Add sh-mem.o to extra_obj.
2619         * config/sh/t-sh (sh-mem.o): New rule.
2620         * config/sh/sh-mem.cc (expand_block_move): Moved here.
2621         (sh_expand_cmpstr): New function.
2622         * config/sh/sh.c (force_into, expand_block_move): Move to sh-mem.c.
2623         * config/sh/sh-protos.h (sh_expand_cmpstr): Declare.
2624         * config/sh/sh.md (cmpstrsi, cmpstr_t): New patterns.
2625         (rotlhi3_8): Rename.
2627 2013-10-24  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2629         * configure.ac (ZW_PROG_COMPILER_DEPENDENCIES): Use CXX instead of CC.
2630         * Makefile.in (CXXDEPMODE): Assign and change users.
2631         (CCDEPMODE): Delete.
2632         * configure: Regenerate.
2634 2013-10-23  David Malcolm  <dmalcolm@redhat.com>
2636         * gengtype-parse.c (require_without_advance): New.
2637         (type): For GTY-marked types that are not GTY((user)), parse any
2638         base classes, requiring them to be single-inheritance, and not
2639         be templates.  For non-GTY-marked types and GTY((user)),
2640         continue to skip over any C++ inheritance specification.
2641         * gengtype-state.c (state_writer::write_state_struct_type):
2642         Write base class of type (if any).
2643         (read_state_struct_type): Read base class of type (if any).
2644         * gengtype.c (new_structure): Add a "base_class" parameter.
2645         (create_optional_field_): Update for new parameter to new_structure.
2646         (adjust_field_rtx_def): Likewise.
2647         (adjust_field_tree_exp): Likewise.
2648         * gengtype.h (struct type): Add "base_class" field to the s
2649         union field.
2650         (new_structure): Add "base" parameter.
2652 2013-10-23  Sriraman Tallam  <tmsriram@google.com>
2654         PR target/57756
2655         * config/i386/i386.c (ix86_option_override_internal):
2656         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...)
2657         (ix86_valid_target_attribute_tree):
2658         Change TARGET_64BIT to TARGET_64BIT_P (opts->...)
2659         Change TARGET_SSE to TARGET_SSE_P (opts->...)
2661 2013-10-23  Andrew MacLeod  <amacleod@redhat.com>
2663         * tree-ssa-loop.h: Remove include files.
2664         * gengtype.c (open_base_files): Adjust include list for gtype-desc.c.
2665         * cfgloopmanip.c: Move required includes from tree-ssa-loop.h.
2666         * graphite-clast-to-gimple.c: Likewise.
2667         * graphite-scop-detection.c: Likewise.
2668         * graphite-sese-to-poly.c: Likewise.
2669         * ipa-inline-analysis.c: Likewise.
2670         * ipa-pure-const.c: Likewise.
2671         * loop-init.c: Likewise.
2672         * passes.c: Likewise.
2673         * predict.c: Likewise.
2674         * tree-cfg.c: Likewise.
2675         * tree-cfgcleanup.c: Likewise.
2676         * tree-chrec.c: Likewise.
2677         * tree-data-ref.c: Likewise.
2678         * tree-loop-distribution.c: Likewise.
2679         * tree-parloops.c: Likewise.
2680         * tree-predcom.c: Likewise.
2681         * tree-scalar-evolution.c: Likewise.
2682         * tree-ssa-address.c: Likewise.
2683         * tree-ssa.c: Likewise.
2684         * tree-ssa-dce.c: Likewise.
2685         * tree-ssa-loop.c: Likewise.
2686         * tree-ssa-loop-im.c: Likewise.
2687         * tree-ssa-loop-ivcanon.c: Likewise.
2688         * tree-ssa-loop-ivopts.c: Likewise.
2689         * tree-ssa-loop-manip.c: Likewise.
2690         * tree-ssa-loop-niter.c: Likewise.
2691         * tree-ssa-loop-prefetch.c: Likewise.
2692         * tree-ssa-loop-unswitch.c: Likewise.
2693         * tree-ssa-reassoc.c: Likewise.
2694         * tree-vect-data-refs.c: Likewise.
2695         * tree-vect-loop.c: Likewise.
2696         * tree-vect-loop-manip.c: Likewise.
2697         * tree-vectorizer.c: Likewise.
2698         * tree-vect-stmts.c: Likewise.
2699         * tree-vrp.c: Likewise.
2701 2013-10-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2703         * config/rs6000/altivec.md (mulv8hi3): Adjust for little endian.
2705 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
2707         PR tree-optimization/58775
2708         PR tree-optimization/58791
2709         * tree-ssa-reassoc.c (reassoc_stmt_dominates_stmt_p): New function.
2710         (insert_stmt_after): Rewritten, don't move the stmt, but really
2711         insert it.
2712         (get_stmt_uid_with_default): Remove.
2713         (build_and_add_sum): Use insert_stmt_after and
2714         reassoc_stmt_dominates_stmt_p.  Fix up uid if bb contains only labels.
2715         (update_range_test): Set uid on stmts added by
2716         force_gimple_operand_gsi.  Don't immediately modify statements
2717         in inter-bb optimization, just update oe->op values.
2718         (optimize_range_tests): Return bool whether any changed have been made.
2719         (update_ops): New function.
2720         (struct inter_bb_range_test_entry): New type.
2721         (maybe_optimize_range_tests): Perform statement changes here.
2722         (not_dominated_by, appears_later_in_bb, get_def_stmt,
2723         ensure_ops_are_available): Remove.
2724         (find_insert_point): Rewritten.
2725         (rewrite_expr_tree): Remove MOVED argument, add CHANGED argument,
2726         return LHS of the (new resp. old) stmt.  Don't call
2727         ensure_ops_are_available, don't reuse SSA_NAMEs, recurse first
2728         instead of last, move new stmt at the right place.
2729         (linearize_expr, repropagate_negates): Don't reuse SSA_NAMEs.
2730         (negate_value): Likewise.  Set uids.
2731         (break_up_subtract_bb): Initialize uids.
2732         (reassociate_bb): Adjust rewrite_expr_tree caller.
2733         (do_reassoc): Don't call renumber_gimple_stmt_uids.
2735 2013-10-23  David Edelsohn  <dje.gcc@gmail.com>
2737         PR target/58838
2738         * config/rs6000/rs6000.md (mulsi3_internal1 and splitter): Add
2739         TARGET_32BIT final condition.
2740         (mulsi3_internal2 and splitter): Same.
2742 2013-10-23  Jeff Law  <law@redhat.com>
2744         * tree-ssa-threadedge.c (thread_across_edge): Do not allow threading
2745         through joiner blocks with abnormal outgoing edges.
2747         * tree-ssa-threadupdate.c (thread_block_1): Renamed from thread_block.
2748         Add parameter JOINERS, to allow/disallow threading through joiner
2749         blocks.
2750         (thread_block): New.  Call thread_block_1.
2751         (mark_threaded_blocks): Remove code to filter out certain cases
2752         of threading through joiner blocks.
2753         (thread_through_all_blocks): Document how we can have a dangling
2754         edge AUX field and clear it.
2756 2013-10-23  Ian Lance Taylor  <iant@google.com>
2758         * doc/invoke.texi (Option Summary): Remove -fno-default-inline.
2759         (C++ Dialect Options): Likewise.
2760         (Optimize Options): Likewise.
2762 2013-10-23  Tom de Vries  <tom@codesourcery.com>
2764         PR tree-optimization/58805
2765         * tree-ssa-tail-merge.c (stmt_local_def): Add gimple_vdef check.
2767 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
2769         * tree-vect-patterns.c (vect_recog_divmod_pattern): Optimize
2770         sequence based on get_range_info returned range.
2772 2013-10-23  Andrew MacLeod  <amacleod@redhat.com>
2774         * tree-ssa.h: Remove all #include's
2775         * gengtype.c (open_base_files): Adjust include list for gtype-desc.c.
2776         * alias.c: Move required includes from tree-ssa.h.
2777         * asan.c: Likewise.
2778         * builtins.c: Likewise.
2779         * calls.c: Likewise.
2780         * cfgexpand.c: Likewise.
2781         * cfghooks.c: Likewise.
2782         * cfgloop.c: Likewise.
2783         * cfgloopmanip.c: Likewise.
2784         * cgraph.c: Likewise.
2785         * cgraphbuild.c: Likewise.
2786         * cgraphclones.c: Likewise.
2787         * cgraphunit.c: Likewise.
2788         * dse.c: Likewise.
2789         * except.c: Likewise.
2790         * expr.c: Likewise.
2791         * final.c: Likewise.
2792         * fold-const.c: Likewise.
2793         * ggc-page.c: Likewise.
2794         * gimple-builder.c: Likewise.
2795         * gimple-fold.c: Likewise.
2796         * gimple-iterator.c: Likewise.
2797         * gimple-low.c: Likewise.
2798         * gimple-pretty-print.c: Likewise.
2799         * gimple-ssa-strength-reduction.c: Likewise.
2800         * gimple-streamer-in.c: Likewise.
2801         * gimple-streamer-out.c: Likewise.
2802         * gimplify.c: Likewise.
2803         * graphite-blocking.c: Likewise.
2804         * graphite-clast-to-gimple.c: Likewise.
2805         * graphite-dependences.c: Likewise.
2806         * graphite-interchange.c: Likewise.
2807         * graphite-optimize-isl.c: Likewise.
2808         * graphite-poly.c: Likewise.
2809         * graphite-scop-detection.c: Likewise.
2810         * graphite-sese-to-poly.c: Likewise.
2811         * graphite.c: Likewise.
2812         * ipa-cp.c: Likewise.
2813         * ipa-inline-analysis.c: Likewise.
2814         * ipa-inline-transform.c: Likewise.
2815         * ipa-inline.c: Likewise.
2816         * ipa-prop.c: Likewise.
2817         * ipa-pure-const.c: Likewise.
2818         * ipa-reference.c: Likewise.
2819         * ipa-split.c: Likewise.
2820         * ipa-utils.c: Likewise.
2821         * loop-init.c: Likewise.
2822         * lto-cgraph.c: Likewise.
2823         * lto-section-in.c: Likewise.
2824         * lto-section-out.c: Likewise.
2825         * lto-streamer-in.c: Likewise.
2826         * lto-streamer-out.c: Likewise.
2827         * lto-streamer.c: Likewise.
2828         * omp-low.c: Likewise.
2829         * passes.c: Likewise.
2830         * predict.c: Likewise.
2831         * print-tree.c: Likewise.
2832         * profile.c: Likewise.
2833         * sese.c: Likewise.
2834         * targhooks.c: Likewise.
2835         * tracer.c: Likewise.
2836         * trans-mem.c: Likewise.
2837         * tree-call-cdce.c: Likewise.
2838         * tree-cfg.c: Likewise.
2839         * tree-cfgcleanup.c: Likewise.
2840         * tree-chrec.c: Likewise.
2841         * tree-complex.c: Likewise.
2842         * tree-data-ref.c: Likewise.
2843         * tree-dfa.c: Likewise.
2844         * tree-eh.c: Likewise.
2845         * tree-emutls.c: Likewise.
2846         * tree-if-conv.c: Likewise.
2847         * tree-inline.c: Likewise.
2848         * tree-into-ssa.c: Likewise.
2849         * tree-loop-distribution.c: Likewise.
2850         * tree-mudflap.c: Likewise.
2851         * tree-nested.c: Likewise.
2852         * tree-nrv.c: Likewise.
2853         * tree-object-size.c: Likewise.
2854         * tree-outof-ssa.c: Likewise.
2855         * tree-parloops.c: Likewise.
2856         * tree-phinodes.c: Likewise.
2857         * tree-predcom.c: Likewise.
2858         * tree-pretty-print.c: Likewise.
2859         * tree-profile.c: Likewise.
2860         * tree-scalar-evolution.c: Likewise.
2861         * tree-sra.c: Likewise.
2862         * tree-ssa-address.c: Likewise.
2863         * tree-ssa-alias.c: Likewise.
2864         * tree-ssa-ccp.c: Likewise.
2865         * tree-ssa-coalesce.c: Likewise.
2866         * tree-ssa-copy.c: Likewise.
2867         * tree-ssa-copyrename.c: Likewise.
2868         * tree-ssa-dce.c: Likewise.
2869         * tree-ssa-dom.c: Likewise.
2870         * tree-ssa-dse.c: Likewise.
2871         * tree-ssa-forwprop.c: Likewise.
2872         * tree-ssa-ifcombine.c: Likewise.
2873         * tree-ssa-live.c: Likewise.
2874         * tree-ssa-loop-ch.c: Likewise.
2875         * tree-ssa-loop-im.c: Likewise.
2876         * tree-ssa-loop-ivcanon.c: Likewise.
2877         * tree-ssa-loop-ivopts.c: Likewise.
2878         * tree-ssa-loop-manip.c: Likewise.
2879         * tree-ssa-loop-niter.c: Likewise.
2880         * tree-ssa-loop-prefetch.c: Likewise.
2881         * tree-ssa-loop-unswitch.c: Likewise.
2882         * tree-ssa-loop.c: Likewise.
2883         * tree-ssa-math-opts.c: Likewise.
2884         * tree-ssa-operands.c: Likewise.
2885         * tree-ssa-phiopt.c: Likewise.
2886         * tree-ssa-phiprop.c: Likewise.
2887         * tree-ssa-pre.c: Likewise.
2888         * tree-ssa-propagate.c: Likewise.
2889         * tree-ssa-reassoc.c: Likewise.
2890         * tree-ssa-sccvn.c: Likewise.
2891         * tree-ssa-sink.c: Likewise.
2892         * tree-ssa-strlen.c: Likewise.
2893         * tree-ssa-structalias.c: Likewise.
2894         * tree-ssa-tail-merge.c: Likewise.
2895         * tree-ssa-ter.c: Likewise.
2896         * tree-ssa-threadedge.c: Likewise.
2897         * tree-ssa-threadupdate.c: Likewise.
2898         * tree-ssa-uncprop.c: Likewise.
2899         * tree-ssa-uninit.c: Likewise.
2900         * tree-ssa.c: Likewise.
2901         * tree-ssanames.c: Likewise.
2902         * tree-stdarg.c: Likewise.
2903         * tree-streamer-in.c: Likewise.
2904         * tree-switch-conversion.c: Likewise.
2905         * tree-tailcall.c: Likewise.
2906         * tree-vect-data-refs.c: Likewise.
2907         * tree-vect-generic.c: Likewise.
2908         * tree-vect-loop-manip.c: Likewise.
2909         * tree-vect-loop.c: Likewise.
2910         * tree-vect-patterns.c: Likewise.
2911         * tree-vect-slp.c: Likewise.
2912         * tree-vect-stmts.c: Likewise.
2913         * tree-vectorizer.c: Likewise.
2914         * tree-vrp.c: Likewise.
2915         * tree.c: Likewise.
2916         * tsan.c: Likewise.
2917         * value-prof.c: Likewise.
2918         * var-tracking.c: Likewise.
2919         * varpool.c: Likewise.
2920         * vtable-verify.c: Likewise.
2922 2013-10-23  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2924         * config/tilegx/tilegx.c: Include "tree.h".
2926 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
2928         * gimple-pretty-print.c (dump_ssaname_info): Always print "# " before
2929         the info, not after it.
2930         (gump_gimple_phi): Add COMMENT argument, if true, print "# " after
2931         dump_ssaname_info call.
2932         (pp_gimple_stmt_1): Adjust caller.
2933         (dump_phi_nodes): Likewise.  Don't print "# " here.
2935 2013-10-22  Jan Hubicka  <jh@suse.cz>
2937         * i386.h (TARGET_MISALIGNED_MOVE_STRING_PROLOGUES_EPILOGUES): New
2938         tuning flag.
2939         * x86-tune.def (TARGET_MISALIGNED_MOVE_STRING_PROLOGUES): Define it.
2940         * i386.c (expand_small_movmem_or_setmem): New function.
2941         (expand_set_or_movmem_prologue_epilogue_by_misaligned_moves): New
2942         function.
2943         (alg_usable_p): Add support for value ranges; cleanup.
2944         (ix86_expand_set_or_movmem): Add support for misaligned moves.
2946 2013-10-22  Sterling Augustine  <saugustine@google.com>
2948         * doc/invoke.texi: Document -ggnu-pubnames.
2949         * common.opt: Add new option -ggnu-pubnames and modify -gpubnames
2950         logic.
2951         * dwarf2out.c: Include gdb/gdb-index.h.
2952         (DEBUG_PUBNAMES_SECTION, DEBUG_PUBTYPES_SECTION): Handle
2953         debug_generate_pub_sections.
2954         (is_java, output_pubtables, output_pubname): New functions.
2955         (include_pubname_in_output): Handle debug_generate_pub_sections at
2956         level 2.
2957         (size_of_pubnames): Use new local space_for_flags based on
2958         debug_generate_pub_sections.
2959         (output_pubnames): Unify pubnames and pubtypes output logic.
2960         Genericize comments.  Call output_pubname.
2961         (dwarf2out_finish): Move logic to output_pubnames and call it.
2963 2013-10-22  Uros Bizjak  <ubizjak@gmail.com>
2965         PR target/58779
2966         * config/i386/i386.c (put_condition_code) <case GTU, case LEU>:
2967         Remove CCCmode handling.
2968         <case LTU>: Return 'c' suffix for CCCmode.
2969         <case GEU>: Return 'nc' suffix for CCCmode.
2970         (ix86_cc_mode) <case GTU, case LEU>: Do not generate overflow checks.
2971         * config/i386/i386.md (*sub<mode>3_cconly_overflow): Remove.
2972         (*sub<mode>3_cc_overflow): Ditto.
2973         (*subsi3_zext_cc_overflow): Ditto.
2975 2013-10-22  Steve Ellcey  <sellcey@mips.com>
2977         * config/mips/mips.c (mips_rtx_costs):  Fix cost estimate for nor
2978         (AND (NOT OP1) (NOT OP2)).
2980 2013-10-22  Bill Schmidt  <wschmidt@vnet.ibm.com>
2982         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
2983         meaning of merge-high and merge-low masks for little endian; avoid
2984         use of vector-pack masks for little endian for mismatched modes.
2986 2013-10-22  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2988         * config/tilepro/tilepro.c: Include "tree.h".
2990 2013-10-22  Andreas Schwab  <schwab@suse.de>
2992         * config/m68k/m68k.c (notice_update_cc): Handle register conflict
2993         with PRE_DEC.
2995 2013-10-22  Paolo Carlini  <paolo.carlini@oracle.com>
2997         * doc/contrib.texi ([Fran@,{c}ois Dumont], [Tim Shen],
2998         [Ed Smith-Rowland]): New entries.
2999         ([Stephen M. Webb]): Update.
3001 2013-10-22  Andrew MacLeod  <amacleod@redhat.com>
3003         * tree-ssa-ter.h: Remove duplicate copy of file contents.
3005 2013-10-21  Marek Polacek  <polacek@redhat.com>
3007         PR middle-end/58809
3008         * fold-const.c (fold_range_test): Return 0 if the type is not
3009         an integral type.
3011 2013-10-21  Richard Sandiford  <rdsandiford@googlemail.com>
3013         * system.h: Move hwint.h include further down.
3014         * hwint.h (sext_hwi, zext_hwi): Define unconditionally.  Add
3015         gcc_checking_asserts.
3016         * hwint.c (sext_hwi, zext_hwi): Delete ENABLE_CHECKING versions.
3018 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
3020         Fix volatile issues in optimize_bit_field_compare.
3021         * fold-const.c (optimize_bit_field_compare): Bail out if
3022         lvolatilep or rvolatilep.
3024 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
3026         Fix DECL_BIT_FIELD depencency on flag_strict_volatile_bitfields
3027         and get_inner_reference returning different pmode for non-volatile
3028         bit-field members dependent on flag_strict_volatile_bitfields.
3029         * stor-layout.c (layout_decl): Remove special handling of
3030         flag_strict_volatile_bitfields.
3031         * expr.c (get_inner_reference): Don't use DECL_BIT_FIELD
3032         if flag_strict_volatile_bitfields > 0 and TREE_THIS_VOLATILE.
3034 2013-10-21  Paulo Matos  <pmatos@broadcom.com>
3036         * ipa-inline.c (edge_badness): Cap edge->count at max_count for badness
3037         calculations.
3039 2013-10-21  Jeff Law  <law@redhat.com>
3041         * tree-ssa-threadedge.c (thread_through_normal_block): New
3042         argument VISITED.  Remove VISISTED as a local variable.  When we
3043         have a threadable jump, verify the destination of the jump has not
3044         been visised.
3045         (thread_across_edge): Allocate VISITED bitmap once at function
3046         scope and use it throughout.  Make sure to set appropriate bits in
3047         VISITED for E (start of jump thread path).
3048         * tree-ssa-threadupdate.c (mark_threaded_blocks): Reject threading
3049         through a joiner if any edge on the path has a recorded jump thread.
3051 2013-10-21  Ian Lance Taylor  <iant@google.com>
3053         * doc/invoke.texi (Optimize Options): For -fno-toplevel-reorder,
3054         don't imply that attributes can solve all problems.
3055         (Directory Options): Fix typo.
3057 2013-10-21  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3059         * config/arm/arm.c (cortexa9_extra_costs): Update mult costs for
3060         extend and extend_add.
3062 2013-10-21  Richard Biener  <rguenther@suse.de>
3064         PR tree-optimization/58794
3065         * fold-const.c (operand_equal_p): Compare FIELD_DECL operand
3066         of COMPONENT_REFs with OEP_CONSTANT_ADDRESS_OF left in place.
3068 2013-10-21  Richard Biener  <rguenther@suse.de>
3070         PR middle-end/58742
3071         * fold-const.c (fold_binary_loc): Fold ((T) (X /[ex] C)) * C
3072         to (T) X for sign-changing conversions (or no conversion).
3074 2013-10-20  Uros Bizjak  <ubizjak@gmail.com>
3076         * config/i386/i386.md (kxnor<mode>): Add FLAGS_REG clobber.
3078 2013-10-20  Jan Hubicka  <jh@suse.cz>
3080         * config/i386/i386-tune.def: Add comment; organize into categories
3082 2013-10-21  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
3084         * config/i386/i386.c (expand_set_or_movmem_via_loop): Add issetmem
3085         argument.  Update function comment.
3086         (expand_set_or_movmem_via_rep): New function combining
3087         expand_movmem_via_rep_mov and expand_setmem_via_rep_stos.
3088         (expand_movmem_via_rep_mov): Remove.
3089         expand_setmem_via_rep_stos): Remove.
3090         (expand_movmem_epilogue): Update calls correspondingly.
3091         (expand_setmem_epilogue_via_loop): Likewise.
3092         (emit_memset): New.
3093         (expand_setmem_epilogue): Add VEC_VALUE argument, refactor.
3094         (expand_set_or_movmem_prologue): New function combining
3095         expand_movmem_prologue and expand_setmem_prologue.
3096         (expand_movmem_prologue): Remove.
3097         (expand_setmem_prologue): Remove.
3098         (expand_set_or_movmem_constant_prologue): New function combining
3099         expand_constant_movmem_prologue and expand_constant_setmem_prologue.
3100         (expand_constant_movmem_prologue): Remove.
3101         (expand_constant_setmem_prologue): Remove.
3102         (promote_duplicated_reg): Allow vector-const0 value.
3103         (ix86_expand_set_or_movmem): New function combining ix86_expand_movmem
3104         and ix86_expand_setmem.
3105         (ix86_expand_movmem): Call ix86_expand_set_or_movmem.
3106         (ix86_expand_setmem): Call ix86_expand_set_or_movmem.
3108 2013-10-21  Diego Novillo  <dnovillo@google.com>
3110         * asan.c: Include tree.h
3111         * bb-reorder.c: Likewise.
3112         * cfgcleanup.c: Likewise.
3113         * cfgloopmanip.c: Likewise.
3114         * data-streamer-in.c: Likewise.
3115         * data-streamer-out.c: Likewise.
3116         * data-streamer.c: Likewise.
3117         * dwarf2cfi.c: Likewise.
3118         * graphite-blocking.c: Likewise.
3119         * graphite-clast-to-gimple.c: Likewise.
3120         * graphite-dependences.c: Likewise.
3121         * graphite-interchange.c: Likewise.
3122         * graphite-optimize-isl.c: Likewise.
3123         * graphite-poly.c: Likewise.
3124         * graphite-scop-detection.c: Likewise.
3125         * graphite-sese-to-poly.c: Likewise.
3126         * graphite.c: Likewise.
3127         * ipa-devirt.c: Likewise.
3128         * ipa-profile.c: Likewise.
3129         * ipa.c: Likewise.
3130         * ira.c: Likewise.
3131         * loop-init.c: Likewise.
3132         * loop-unroll.c: Likewise.
3133         * lower-subreg.c: Likewise.
3134         * lto/lto-object.c: Likewise.
3135         * recog.c: Likewise.
3136         * reginfo.c: Likewise.
3137         * tree-loop-distribution.c: Likewise.
3138         * tree-parloops.c: Likewise.
3139         * tree-ssa-strlen.c: Likewise.
3140         * tree-streamer.c: Likewise.
3141         * value-prof.c: Likewise.
3142         * target-globals.c: Likewise.
3143         * expr.h: Include tree-core.h instead of tree.h.
3144         * gimple.h: Likewise.
3145         * ipa-prop.h: Likewise.
3146         * ipa-utils.h: Likewise.
3147         * lto-streamer.h: Likewise.
3148         * streamer-hooks.h: Likewise.
3149         * ipa-reference.h: Include cgraph.h instead of tree.h.
3150         * cgraph.h: Include basic-block.h instead of tree.h.
3151         * tree-streamer.h: Do not include tree.h.
3152         * genattrtab.c (write_header): Generate inclusion of
3153         tree.h.
3154         * genautomata.c (main): Likewise.
3155         * genemit.c: Likewise.
3156         * genopinit.c: Likewise.
3157         * genoutput.c (output_prologue): Likewise.
3158         * genpeep.c: Likewise.
3160 2013-10-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3162         * config/rs6000/altivec.md (vec_unpacku_hi_v16qi): Adjust for
3163         little endian.
3164         (vec_unpacku_hi_v8hi): Likewise.
3165         (vec_unpacku_lo_v16qi): Likewise.
3166         (vec_unpacku_lo_v8hi): Likewise.
3168 2013-10-20  Jan Hubicka  <jh@suse.cz>
3170         * config/i386/x86-tune.def (X86_TUNE_SLOW_IMUL_IMM32_MEM,
3171         X86_TUNE_SLOW_IMUL_IMM8): Keep enabled only for K8 and AMDFAM10.
3172         (X86_TUNE_USE_VECTOR_FP_CONVERTS): Disable for generic.
3174 2013-10-20  Richard Sandiford  <rdsandiford@googlemail.com>
3176         * config/mips/mips.h (ISA_HAS_WSBH): Define.
3177         * config/mips/mips.md (UNSPEC_WSBH, UNSPEC_DSBH, UNSPEC_DSHD): New
3178         constants.
3179         (bswaphi2, bswapsi2, bswapdi2, wsbh, dsbh, dshd): New patterns.
3181 2013-10-19  John David Anglin  <danglin@gcc.gnu.org>
3183         PR target/58603
3184         * system.h: Undef m_slot.
3186 2013-10-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3188         * config/rs6000/rs6000.c (vspltis_constant): Make sure we check
3189         all elements for both endian flavors.
3191 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
3193         PR target/58792
3194         * config/i386/i386.c (ix86_function_value_regno): Add DX_REG,
3195         ST1_REG and XMM1_REG for 32bit and 64bit targets.  Also add DI_REG
3196         and SI_REG for 64bit SYSV ABI targets.
3198 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
3200         * mode-switching.c (create_pre_exit): Rename maybe_builtin_apply
3201         to multi_reg_return.  Clarify that we are skipping USEs of multiple
3202         return registers.  Use bool type where appropriate.
3204 2013-10-18  Jan Hubicka  <jh@suse.cz>
3206         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Disable accumulation
3207         for cold functions.
3208         * x86-tune.def (X86_TUNE_USE_LEAVE): Update comment.
3209         (X86_TUNE_PUSH_MEMORY): Likewise.
3210         (X86_TUNE_AVX256_UNALIGNED_LOAD_OPTIMAL,
3211         X86_TUNE_AVX256_UNALIGNED_STORE_OPTIMAL): New.
3212         (X86_TUNE_ACCUMULATE_OUTGOING_ARGS, X86_TUNE_ALWAYS_FANCY_MATH_387):
3213         New.
3214         * i386.c (x86_accumulate_outgoing_args, x86_arch_always_fancy_math_387,
3215         x86_avx256_split_unaligned_load, x86_avx256_split_unaligned_store):
3216         Remove.
3217         (ix86_option_override_internal): Update to use tune features instead
3218         of variables.
3220 2013-10-18  Cong Hou  <congh@google.com>
3222         PR tree-optimization/58508
3223         * tree-vect-loop-manip.c (vect_loop_versioning): Hoist loop invariant
3224         statement that contains data refs with zero-step.
3226 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
3228         * tree-ssa.h: Don't include gimple-low.h, tree-ssa-address.h,
3229         sbitmap.h, tree-ssa-threadedge.h, tree-ssa-dom.h and tree-cfgcleanup.h.
3230         * gimple-low.c (gimple_check_call_arg,
3231         gimple_check_call_matching_types): Move to cgraph.c.
3232         * gimple-low.h: Remove prototype.
3233         * cgraph.c: (gimple_check_call_arg, gimple_check_call_matching_types):
3234         Relocate from gimple-low.c.
3235         * cgraph.h: Add prototype.  Don't include basic-block.h.
3236         * gimplify.c: Add gimple-low to include list.
3237         * omp-low.c: Add gimple-low and tree-cfgcleanup.h to include list.
3238         * tree-eh.c: Add gimple-low to include list.
3239         * tree-nested.c: Likewise.
3240         * cfgexpand.c: Add tree-ssa-address.h to include list.
3241         * expr.c: Likewise.
3242         * gimple-fold.c: Likewise.
3243         * gimple-ssa-strength-reduction.c: Likewise.
3244         * trans-mem.c: Likewise.
3245         * tree-mudflap.c: Likewise.
3246         * tree-ssa-loop-ivopts.c: Likewise.
3247         * tree-ssa-dom.c: Include tree-ssa-threadedge.h and tree-ssa-dom.h.
3248         (degenerate_phi_result): Move to tree-phinodes.c.
3249         * tree-ssa-loop-ch.c: Include tree-ssa-threadedge.h.
3250         * tree-ssa-threadedge.c: Likewise.
3251         * tree-vrp.c: Likewise.
3252         * tree-phinodes.c (degenerate_phi_result): Relocate here.
3253         * tree-ssa-dom.h (degenerate_phi_result): Remove Prototype.
3254         * tree-phinodes.h (degenerate_phi_result): Add prototype.
3255         * tree-ssa-copy.c: Include tree-ssa-dom.h.
3256         * tree-ssa-forwprop.c: Likewise.
3257         * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing,
3258         pass_data_cleanup_cfg_post_optimizing,
3259         make_pass_cleanup_cfg_post_optimizing): Relocate from tree-optimize.c.
3260         * tree-optimize.c: Delete File.
3261         * graphite.c: Include tree-cfgcleanup.h.
3262         * passes.c: Likewise.
3263         * tree-cfg.c: Likewise.
3264         * tree-profile.c: Likewise.
3265         * tree-ssa-dse.c: Likewise.
3266         * tree-ssa-loop-ivcanon.c: Likewise.
3267         * tree-switch-conversion.c: Don't include tree-ssa-operands.h.
3268         * tree-outof-ssa.c: Include sbitmap.h.
3269         * tree-ssa-live.c: Likewise.
3270         * tree-ssa-propagate.c: Likewise.
3271         * tree-ssa-structalias.c: Likewise.
3272         * tree-stdarg.c: Likewise.
3273         * Makefile.in (OBJS): Delete tree-optimize.o.
3274         * basic-block.h (gcov_type, gcov_type_unsigned): Move to coretypes.h.
3275         * coretypes.h (gcov_type, gcov_type_unsigned): Relocate here.
3276         * varasm.c: Include basic-block.h.
3277         * cfgloop.h: Include function.h instead of basic-block.h
3278         (bb_loop_depth): Move to cfgloop.c.
3279         * cfgloop.c (bb_loop_depth): Relocate from cfgloop.h.
3281 2013-10-18  Teresa Johnson  <tejohnson@google.com>
3283         * predict.c (probably_never_executed): Compare frequency-based
3284         count to number of training runs.
3285         * params.def (UNLIKELY_BB_COUNT_FRACTION): New parameter.
3287 2013-10-18  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3289         * config/arm/arm.c (cortexa9_extra_costs): New table.
3290         (arm_cortex_a9_tune): Use cortexa9_extra_costs.
3292 2013-10-18  Jeff Law  <law@redhat.com>
3294         * tree-ssa-threadupdate.c: Do not include "tm.h" or "tm_p.h".
3296         * tree-ssa-threadupdate.c: Include "dbgcnt.h".
3297         (register_jump_thread): Add "registered_jump_thread" debug
3298         counter support.
3299         * dbgcnt.def (registered_jump_thread): New debug counter.
3301 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
3303         * config/rs6000/rs6000.c: Include cgraph.h.
3305 2013-10-18  Teresa Johnson  <tejohnson@google.com>
3307         * tree-ssa-tail-merge.c (replace_block_by): Update edge
3308         weights during merging.
3310 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
3312         * tree-cfg.h: Rename from tree-flow.h.  Remove #includes.
3313         * tree-ssa.h: Relocate required #includes from tree-cfg.h.
3314         * tree-ssa-operands.h: Remove prototype.
3315         * tree-ssa-operands.c (virtual_operand_p): Move to gimple.c.
3316         * gimple.c (virtual_operand_p): Relocate from gimple.c.
3317         * gimple.h: Add prototype.
3318         * gimple-ssa.h: Include tree-ssa-operands.h.
3319         * tree-dump.c: Add tree-cfg.h to include list.
3320         * tree-ssa-alias.c: Add ipa-reference.h to include list.
3321         * config/alpha/alpha.c: Include gimple-ssa.h instead of tree-flow.h.
3322         * config/i386/i386.c: Don't include tree-flow.h.
3323         * config/rs6000/rs6000.c: Likewise.
3325 2013-10-18  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3327         * config/frv/frv.c (frv_init_cumulative_args): Fix wrong cast.
3329 2013-10-18  Richard Biener  <rguenther@suse.de>
3331         * stor-layout.c (layout_type): Do not change TYPE_PRECISION
3332         or TYPE_UNSIGNED of integral types.
3333         (set_min_and_max_values_for_integral_type): Leave TYPE_MIN/MAX_VALUE
3334         NULL_TREE for zero-precision integral types.
3336 2013-10-18  James Greenhalgh  <james.greenhalgh@arm.com>
3338         * config/aarch64/arm_neon.h
3339         (vcvt<ds>_n_<fsu><32,64>_<fsu><32,64>): Correct argument types.
3341 2013-10-17  Sriraman Tallam  <tmsriram@google.com>
3343         PR target/57756
3344         * opth-gen.awk: Define target_flags_explicit.
3346 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
3348         * config/rs6000/rs6000.c (enum rs6000_reload_reg_type): Add new
3349         fields to the reg_addr array that describes the valid addressing
3350         mode for any register, general purpose registers, floating point
3351         registers, and Altivec registers.
3352         (FIRST_RELOAD_REG_CLASS): Likewise.
3353         (LAST_RELOAD_REG_CLASS): Likewise.
3354         (struct reload_reg_map_type): Likewise.
3355         (reload_reg_map_type): Likewise.
3356         (RELOAD_REG_VALID): Likewise.
3357         (RELOAD_REG_MULTIPLE): Likewise.
3358         (RELOAD_REG_INDEXED): Likewise.
3359         (RELOAD_REG_OFFSET): Likewise.
3360         (RELOAD_REG_PRE_INCDEC): Likewise.
3361         (RELOAD_REG_PRE_MODIFY): Likewise.
3362         (reg_addr): Likewise.
3363         (mode_supports_pre_incdec_p): New helper functions to say whether
3364         a given mode supports PRE_INC, PRE_DEC, and PRE_MODIFY.
3365         (mode_supports_pre_modify_p): Likewise.
3366         (rs6000_debug_vector_unit): Rearrange the -mdebug=reg output to
3367         print the valid address mode bits for each mode.
3368         (rs6000_debug_print_mode): Likewise.
3369         (rs6000_debug_reg_global): Likewise.
3370         (rs6000_setup_reg_addr_masks): New function to set up the address
3371         mask bits for each type.
3372         (rs6000_init_hard_regno_mode_ok): Use memset to clear arrays.
3373         Call rs6000_setup_reg_addr_masks to set up the address mask bits.
3374         (rs6000_legitimate_address_p): Use mode_supports_pre_incdec_p and
3375         mode_supports_pre_modify_p to determine if PRE_INC, PRE_DEC, and
3376         PRE_MODIFY are supported.
3377         (rs6000_output_move_128bit): Change to use {src,dest}_vmx_p for altivec
3378         registers, instead of {src,dest}_av_p.
3379         (rs6000_print_options_internal): Tweak the debug output slightly.
3381 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
3383         * config/i386/sse.md (*vec_widen_smult_even_v8si): Remove
3384         isa attribute.
3386 2013-10-17  Andrew MacLeod  <amacleod@redhat.com>
3388         * tree-flow.h (struct omp_region): Move to omp-low.c.
3389         Remove omp_ prototypes and variables.
3390         * gimple.h (omp_reduction_init): Move prototype to omp-low.h.
3391         (copy_var_decl): Relocate prototype from tree-flow.h.
3392         * gimple.c (copy_var_decl): Relocate from omp-low.c.
3393         * tree.h: Move prototype to omp-low.h.
3394         * omp-low.h: New File.  Relocate prototypes here.
3395         * omp-low.c (struct omp_region): Make local here.
3396         (root_omp_region): Make static.
3397         (copy_var_decl) Move to gimple.c.
3398         (new_omp_region): Make static.
3399         (make_gimple_omp_edges): New.  Refactored from tree-cfg.c make_edges.
3400         * tree-cfg.c: Include omp-low.h.
3401         (make_edges): Factor out OMP specific bits to make_gimple_omp_edges.
3402         * gimplify.c: Include omp-low.h.
3403         * tree-parloops.c: Likewise.
3405 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
3407         * config/i386/i386.c (ix86_fixup_binary_operands): When both source
3408         operands are in memory, prefer to force non-matched operand 1 to
3409         the register.
3411 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
3413         PR target/58673
3414         * config/rs6000/rs6000.c (rs6000_legitimate_address_p): Only
3415         restrict TImode addresses to single indirect registers if both
3416         -mquad-memory and -mvsx-timode are used.
3417         (rs6000_output_move_128bit): Use quad_load_store_p to determine if
3418         we should emit load/store quad.  Remove using %y for quad memory
3419         addresses.
3421         * config/rs6000/rs6000.md (mov<mode>_ppc64, TI/PTImode): Add
3422         constraints to allow load/store quad on machines where TImode is
3423         not allowed in VSX registers.  Use 'n' instead of 'F' constraint
3424         for TImode to load integer constants.
3426 2013-10-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3428         * config/aarch64/aarch64.c (aarch64_print_operand): Handle 'c'.
3430 2013-10-17  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3432         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Adjust
3433         handling of STACK_REG.
3435 2013-10-17  Richard Biener  <rguenther@suse.de>
3437         PR tree-optimization/58143
3438         * tree-ssa-loop-im.c (arith_code_with_undefined_signed_overflow):
3439         New function.
3440         (rewrite_to_defined_overflow): Likewise.
3441         (move_computations_dom_walker::before_dom): Rewrite stmts
3442         with undefined signed overflow that are not always executed
3443         into unsigned arithmetic.
3445 2013-10-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
3447         PR target/57756
3448         * config/rs6000/rs6000.opt (rs6000_isa_flags_explicit): Move the
3449         explicit isa flag to be an options variable, instead of using
3450         global_options_set.  Remove define from rs6000.h.
3451         * config/rs6000/rs6000.h (rs6000_isa_flags_explicit): Likewise.
3453         * config/rs6000/rs6000.c (rs6000_option_override_internal):
3454         Initialize rs6000_isa_flags_explicit.
3455         (rs6000_function_specific_save): Add gcc_options* parameter, so
3456         that the powerpc builds after the 2013-10-15 changes.
3457         (rs6000_function_specific_restore): Likewise.
3459 2013-10-16  DJ Delorie  <dj@redhat.com>
3461         * config/rl78/rl78.c (rl78_alloc_address_registers_macax): Verify
3462         op is a REG before checking REGNO.
3463         (rl78_alloc_physical_registers): Verify pattern is a SET before
3464         checking SET_SRC.
3466 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3468         * config/rs6000/vector.md (vec_unpacks_hi_v4sf): Correct for
3469         endianness.
3470         (vec_unpacks_lo_v4sf): Likewise.
3471         (vec_unpacks_float_hi_v4si): Likewise.
3472         (vec_unpacks_float_lo_v4si): Likewise.
3473         (vec_unpacku_float_hi_v4si): Likewise.
3474         (vec_unpacku_float_lo_v4si): Likewise.
3476 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3478         * config/rs6000/vsx.md (vsx_concat_<mode>): Adjust output for LE.
3479         (vsx_concat_v2sf): Likewise.
3481 2013-10-16  James Greenhalgh  <james.greenhalgh@arm.com>
3483         * config/aarch64/aarch64.md
3484         (*mov<mode>_aarch64): Fix output template for DUP (element) Scalar.
3486 2013-10-16  Andrew MacLeod  <amacleod@redhat.com>
3488         PR tree-optimization/58697
3489         * cfgloop.c (get_estimated_loop_iterations_int): Rename from
3490         estimated_loop_iterations_int.
3491         (max_stmt_executions_int): Call get_max_loop_iterations_int.
3492         (get_max_loop_iterations_int): New.  HWINT version of
3493         get_max_loop_iterations.
3494         * cfgloop.h: Add prototypes.
3495         * loop-iv.c (find_simple_exit): call get_estimated_loop_iterations_int.
3496         * loop-unroll.c (decide_peel_once_rolling): Call
3497         get_estimated_loop_iterations_int.
3498         * tree-ssa-loop-niter.c (estimated_loop_iterations_int): Add back.
3499         * tree-ssa-loop-niter.h: Tweak prototypes.
3501 2013-10-16  David Malcolm  <dmalcolm@redhat.com>
3503         * gengtype-parse.c (struct_field_seq): Ignore access-control
3504         keywords ("public:" etc).
3506 2013-10-16  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3508         * config/aarch64/aarch64.c (aarch64_regno_regclass): Classify
3509         FRAME_POINTER_REGNUM and ARG_POINTER_REGNUM as POINTER_REGS.
3511 2013-10-16  Yvan Roux  <yvan.roux@linaro.org>
3513         * config/arm/arm.opt (mlra): New option.
3514         * config/arm/arm.c (arm_lra_p): New function.
3515         (TARGET_LRA_P): Define.
3517 2013-10-16  Paulo Matos  <pmatos@broadcom.com>
3519         * tree-core.h (tree_code_name): Remove.
3520         * tree.h (get_tree_code_name): New prototype.
3521         * tree.c (tree_code_name): Make static.
3522         (get_tree_code_name): New function.
3523         (dump_tree_statistics, tree_check_failed, tree_not_check_failed,
3524         tree_class_check_failed, tree_range_check_failed,
3525         tree_not_class_check_failed, omp_clause_check_failed,
3526         tree_contains_struct_check_failed, tree_operand_check_failed): Use new
3527         wrapper get_tree_code_name instead of calling tree_code_name directly.
3528         * tree-vrp.c (dump_asserts_for): Likewise.
3529         * tree-dump.c (dequeue_and_dump): Likewise.
3530         * tree-pretty-print.c (do_niy, dump_generic_node): Likewise.
3531         * tree-pretty-print.h (pp_unsupported_tree): Likewise.
3532         * lto-streamer-out.c (lto_write_tree, DFS_write_tree): Likewise.
3533         * tree-ssa-dom.c (print_expr_hash_elt): Likewise.
3534         * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs,
3535         dump_ternary_rhs, dump_gimple_assign, dump_gimple_cond,
3536         dump_gimple_omp_for): Likewise.
3537         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
3538         * tree-ssa-pre.c (print_pre_expr): Likewise.
3539         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
3540         * print-tree.c (print_node_brief, print_node): Likewise.
3541         * gimple.c (gimple_check_failed): Likewise.
3542         * lto-streamer.c (lto_tag_name, print_lto_report): Likewise.
3543         * config/frv/frv.c (frv_init_cumulative_args): Likewise.
3544         * config/mep/mep.c (mep_validate_vliw): Likewise.
3545         * config/iq2000/iq2000.c (init_cumulative_args): Likewise.
3546         * config/rs6000/rs6000.c (init_cumulative_args): Likewise.
3548 2013-10-16 Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
3550         * config/i386/i386.c (ix86_option_override_internal): Enable FMA4
3551         for AMD bdver3.
3553 2013-10-16  Hans-Peter Nilsson  <hp@axis.com>
3555         * config/cris/t-elfmulti (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
3556         (MULTILIB_MATCHES): Add multilib for -march=v8.
3558 2013-10-15 Sriraman Tallam  <tmsriram@google.com>
3560         PR target/57756
3561         * optc-save-gen.awk: Add extra parameter to the save and restore
3562         target calls.
3563         * opth-gen.awk: Generate new TARGET_* macros  to accept a parameter.
3564         * tree.c (build_optimization_node): New parameter.  Add extra parameter
3565         to call to cl_optimization_save.
3566         (build_target_option_node): New parameter. Add extra parameter
3567         to call to cl_target_option_save.
3568         * tree.h (build_optimization_node): New parameter.
3569         (build_target_option_node): New parameter.
3570         * c-family/c-common.c (handle_optimize_attribute): Fix calls to
3571         build_optimization_node and build_target_option_node.
3572         * c-family/c-pragma.c (handle_pragma_optimize): Ditto.
3573         (handle_pragma_push_options): Ditto.
3574         * toplev.c (process_options): Ditto.
3575         * opts.c (init_options_struct): Check for opts_set non-null.
3576         * target.def (target_option.save): New parameter.
3577         (target_option.restore): New parameter.
3578         * tm.texi: Generate.
3579         * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
3580         (ix86_pragma_target_parse): Ditto.
3581         * config/i386/i386-protos.h (ix86_valid_target_attribute_tree): New
3582         parameters.
3583         * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix calls
3584         to  build_optimization_node and build_target_option_node.
3585         (rs6000_valid_attribute_p): Ditto.
3586         (rs6000_pragma_target_parse): Ditto.
3587         * config/i386/i386.opt (x_ix86_target_flags_explicit): New TargetSave
3588         data.
3589         * config/i386/i386.h:
3590         TARGET_64BIT_P: New Macro
3591         TARGET_MMX_P: New Macro.
3592         TARGET_3DNOW_P: New Macro.
3593         TARGET_3DNOW_A_P: New Macro.
3594         TARGET_SSE_P: New Macro.
3595         TARGET_SSE2_P: New Macro.
3596         TARGET_SSE3_P: New Macro.
3597         TARGET_SSSE3_P: New Macro.
3598         TARGET_SSE4_1_P: New Macro.
3599         TARGET_SSE4_2_P: New Macro.
3600         TARGET_AVX_P: New Macro.
3601         TARGET_AVX2_P: New Macro.
3602         TARGET_AVX512F_P: New Macro.
3603         TARGET_AVX512PF_P: New Macro.
3604         TARGET_AVX512ER_P: New Macro.
3605         TARGET_AVX512CD_P: New Macro.
3606         TARGET_FMA_P: New Macro.
3607         TARGET_SSE4A_P: New Macro.
3608         TARGET_FMA4_P: New Macro.
3609         TARGET_XOP_P: New Macro.
3610         TARGET_LWP_P: New Macro.
3611         TARGET_ABM_P: New Macro.
3612         TARGET_BMI_P: New Macro.
3613         TARGET_BMI2_P: New Macro.
3614         TARGET_LZCNT_P: New Macro.
3615         TARGET_TBM_P: New Macro.
3616         TARGET_POPCNT_P: New Macro.
3617         TARGET_SAHF_P: New Macro.
3618         TARGET_MOVBE_P: New Macro.
3619         TARGET_CRC32_P: New Macro.
3620         TARGET_AES_P: New Macro.
3621         TARGET_PCLMUL_P: New Macro.
3622         TARGET_CMPXCHG16B_P: New Macro.
3623         TARGET_FSGSBASE_P: New Macro.
3624         TARGET_RDRND_P: New Macro.
3625         TARGET_F16C_P: New Macro.
3626         TARGET_RTM_P: New Macro.
3627         TARGET_HLE_P: New Macro.
3628         TARGET_RDSEED_P: New Macro.
3629         TARGET_PRFCHW_P: New Macro.
3630         TARGET_ADX_P: New Macro.
3631         TARGET_FXSR_P: New Macro.
3632         TARGET_XSAVE_P: New Macro.
3633         TARGET_XSAVEOPT_P: New Macro.
3634         TARGET_LP64_P: New Macro.
3635         TARGET_X32_P: New Macro.
3636         TARGET_FPMATH_DEFAULT_P: New Macro.
3637         TARGET_FLOAT_RETURNS_IN_80387_P: New Macro.
3638         * config/i386/i386.c (ix86_option_override_internal): New parameters.
3639         opts and opts_set.
3640         Change ix86_tune_string to access opts->x_ix86_tune_string.
3641         Change ix86_isa_flags to access opts->x_ix86_isa_flags.
3642         Change ix86_arch_string to access opts->x_ix86_arch_string.
3643         Change ix86_stringop_alg to access opts->x_ix86_stringop_alg.
3644         Change ix86_pmode to access opts->x_ix86_pmode.
3645         Change ix86_abi to access opts->x_ix86_abi.
3646         Change ix86_cmodel to access opts->x_ix86_cmodel.
3647         Change ix86_asm_dialect to access opts->x_ix86_asm_dialect.
3648         Change ix86_isa_flags_explicit to access
3649         opts->x_ix86_isa_flags_explicit.
3650         Change ix86_dump_tunes to access opts->x_ix86_dump_tunes.
3651         Change ix86_regparm to access opts->x_ix86_regparm.
3652         Change ix86_branch_cost to access opts->x_ix86_branch_cost.
3653         Change ix86_preferred_stack_boundary_arg to access
3654         opts->x_ix86_preferred_stack_boundary_arg.
3655         Change ix86_force_align_arg_pointer to access
3656         opts->x_ix86_force_align_arg_pointer.
3657         Change ix86_incoming_stack_boundar_arg to access
3658         opts->x_ix86_incoming_stack_boundar_arg.
3659         Change ix86_fpmath to access opts->x_ix86_fpmath.
3660         Change ix86_veclibabi_type to access opts->x_ix86_veclibabi_type.
3661         Change ix86_recip_name to access opts->x_ix86_recip_name.
3662         Change ix86_stack_protector_guard to access
3663         opts->x_ix86_stack_protector_guard.
3664         Change ix86_tune_memcpy_strategy to access
3665         opts->x_ix86_tune_memcpy_strategy.
3666         Change ix86_tune_memset_strategy to access
3667         opts->x_ix86_tune_memset_strategy.
3668         Change global_options to access opts.
3669         Change global_options_set to access opts_set.
3670         Change TARGET_64BIT to TARGET_64BIT_P (opts->...).
3671         Change TARGET_MMX to TARGET_MMX_P (opts->...).
3672         Change TARGET_3DNOW to TARGET_3DNOW_P (opts->...).
3673         Change TARGET_3DNOW_A to TARGET_3DNOW_A_P (opts->...).
3674         Change TARGET_SSE to TARGET_SSE_P (opts->...).
3675         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...).
3676         Change TARGET_SSE3 to TARGET_SSE3_P (opts->...).
3677         Change TARGET_SSSE3 to TARGET_SSSE3_P (opts->...).
3678         Change TARGET_SSE4_1 to TARGET_SSE4_1_P (opts->...).
3679         Change TARGET_SSE4_2 to TARGET_SSE4_2_P (opts->...).
3680         Change TARGET_AVX to TARGET_AVX_P (opts->...).
3681         Change TARGET_AVX2 to TARGET_AVX2_P (opts->...).
3682         Change TARGET_AVX512F to TARGET_AVX512F_P (opts->...).
3683         Change TARGET_AVX512PF to TARGET_AVX512PF_P (opts->...).
3684         Change TARGET_AVX512ER to TARGET_AVX512ER_P (opts->...).
3685         Change TARGET_AVX512CD to TARGET_AVX512CD_P (opts->...).
3686         Change TARGET_FMA to TARGET_FMA_P (opts->...).
3687         Change TARGET_SSE4A to TARGET_SSE4A_P (opts->...).
3688         Change TARGET_FMA4 to TARGET_FMA4_P (opts->...).
3689         Change TARGET_XOP to TARGET_XOP_P (opts->...).
3690         Change TARGET_LWP to TARGET_LWP_P (opts->...).
3691         Change TARGET_ABM to TARGET_ABM_P (opts->...).
3692         Change TARGET_BMI to TARGET_BMI_P (opts->...).
3693         Change TARGET_BMI2 to TARGET_BMI2_P (opts->...).
3694         Change TARGET_LZCNT to TARGET_LZCNT_P (opts->...).
3695         Change TARGET_TBM to TARGET_TBM_P (opts->...).
3696         Change TARGET_POPCNT to TARGET_POPCNT_P (opts->...).
3697         Change TARGET_SAHF to TARGET_SAHF_P (opts->...).
3698         Change TARGET_MOVBE to TARGET_MOVBE_P (opts->...).
3699         Change TARGET_CRC32 to TARGET_CRC32_P (opts->...).
3700         Change TARGET_AES to TARGET_AES_P (opts->...).
3701         Change TARGET_PCLMUL to TARGET_PCLMUL_P (opts->...).
3702         Change TARGET_CMPXCHG16B to TARGET_CMPXCHG16B_P (opts->...).
3703         Change TARGET_FSGSBASE to TARGET_FSGSBASE_P (opts->...).
3704         Change TARGET_RDRND to TARGET_RDRND_P (opts->...).
3705         Change TARGET_F16C to TARGET_F16C_P (opts->...).
3706         Change TARGET_RTM to TARGET_RTM_P (opts->...).
3707         Change TARGET_HLE to TARGET_HLE_P (opts->...).
3708         Change TARGET_RDSEED to TARGET_RDSEED_P (opts->...).
3709         Change TARGET_PRFCHW to TARGET_PRFCHW_P (opts->...).
3710         Change TARGET_ADX to TARGET_ADX_P (opts->...).
3711         Change TARGET_FXSR to TARGET_FXSR_P (opts->...).
3712         Change TARGET_XSAVE to TARGET_XSAVE_P (opts->...).
3713         Change TARGET_XSAVEOPT to TARGET_XSAVEOPT_P (opts->...).
3714         Change TARGET_LP64 to TARGET_LP64_P (opts->...).
3715         Change TARGET_X32 to TARGET_X32_P (opts->...).
3716         Change TARGET_FPMATH_DEFAULT to TARGET_FPMATH_DEFAULT_P (opts->...).
3717         Change TARGET_FLOAT_RETURNS_IN_80387 to
3718         TARGET_FLOAT_RETURNS_IN_80387_P (opts->...).
3719         (ix86_function_specific_save): New parameter. Use opts-> fields
3720         to replace global fields.
3721         (ix86_function_specific_restore): Ditto.
3722         (ix86_valid_target_attribute_inner_p): New parameters.
3723         Fix recursive call.
3724         Fix call to ix86_handle_option and set_option.
3725         (ix86_valid_target_attribute_tree): New parameters.
3726         Change global_options to access opts.
3727         Change global_options_set to access opts_set.
3728         Fix call to ix86_valid_target_attribute_inner_p.
3729         Change ix86_tune_string to access opts->x_ix86_tune_string.
3730         Change ix86_arch_string to access opts->x_ix86_arch_string.
3731         Change ix86_fpmath to access opts->x_ix86_fpmath
3732         Fix call to ix86_option_override_internal.
3733         Fix call to ix86_add_new_builtins.
3734         Fix calls to build_optimization_node and build_target_option_node.
3735         (ix86_valid_target_attribute_p): Remove access to global_options.
3736         Use new gcc_options structure func_options.
3737         Fix call to ix86_valid_target_attribute_tree.
3738         Fix call to  build_optimization_node.
3739         (get_builtin_code_for_version): Fix call to
3740         ix86_valid_target_attribute_tree.
3742 2013-10-15  David Malcolm  <dmalcolm@redhat.com>
3744         * Makefile.in (PICFLAG): New.
3745         (enable_host_shared): New.
3746         (INTERNAL_CFLAGS): Use PICFLAG.
3747         (LIBIBERTY): Use pic build of libiberty.a if configured with
3748         --enable-host-shared.
3749         * configure.ac: Add --enable-host-shared, setting up new
3750         PICFLAG variable.
3751         * configure: Regenerate.
3752         * doc/install.texi (--enable-shared): Add note contrasting it with ...
3753         (--enable-host-shared): New option.
3755 2013-10-15  Richard Biener  <rguenther@suse.de>
3757         * tree-tailcall.c (find_tail_calls): Don't use tail-call recursion
3758         for built-in functions.
3760 2013-10-15  Zhenqiang Chen  <zhenqiang.chen@arm.com>
3762         * tree-ssa-reassoc.c: Include rtl.h and tm_p.h.
3763         (optimize_range_tests_1): New function,
3764         extracted from optimize_range_tests.
3765         (optimize_range_tests_xor): Similarly.
3766         (optimize_range_tests_diff): New function.
3767         (optimize_range_tests): Use optimize_range_tests_1.
3769 2013-10-15  Cong Hou  <congh@google.com>
3771         * tree-vect-loop.c (vect_is_simple_reduction_1): Relax the
3772         requirement of the reduction pattern so that one operand of the
3773         reduction operation can come from outside of the loop.
3775 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3777         * config/arm/neon-schedgen.ml: Remove.
3778         * config/arm/cortex-a9-neon.md: Remove comment regarding
3779         neon-schedgen.ml.
3781 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3783         * config/arm/types: Remove old neon types.
3785 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3787         * config/arm/cortex-a7.md
3788         (cortex_a7_neon_type): New.
3789         (cortex_a7_neon_mul): Update for new types.
3790         (cortex_a7_neon_mla): Likewise.
3791         (cortex_a7_neon): Likewise.
3793 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3795         * config/arm/cortex-a15-neon.md
3796         (cortex_a15_neon_type): New,
3798         (cortex_a15_neon_int_1): Remove.
3799         (cortex_a15_neon_int_2): Likewise.
3800         (cortex_a15_neon_int_3): Likewise.
3801         (cortex_a15_neon_int_4): Likewise.
3802         (cortex_a15_neon_int_5): Likewise.
3803         (cortex_a15_neon_vqneg_vqabs): Likewise.
3804         (cortex_a15_neon_vmov): Likewise.
3805         (cortex_a15_neon_vaba): Likewise.
3806         (cortex_a15_neon_vaba_qqq): Likewise.
3807         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3808         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
3809         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
3810         Likewise.
3811         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3812         (cortex_a15_neon_mla_qqq_8_16): Likewise.
3813         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar): Likewise.
3814         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
3815         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
3816         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
3817         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
3818         (cortex_a15_neon_shift_1): Likewise.
3819         (cortex_a15_neon_shift_2): Likewise.
3820         (cortex_a15_neon_shift_3): Likewise.
3821         (cortex_a15_neon_vshl_ddd): Likewise.
3822         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
3823         (cortex_a15_neon_vsra_vrsra): Likewise.
3824         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
3825         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
3826         (cortex_a15_neon_bp_3cycle): Likewise.
3827         (cortex_a15_neon_ldm_2): Likewise.
3828         (cortex_a15_neon_stm_2): Likewise.
3829         (cortex_a15_neon_mcr): Likewise.
3830         (cortex_a15_neon_mrc): Likewise.
3831         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
3832         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
3833         (cortex_a15_neon_fp_vmul_ddd): Likewise.
3834         (cortex_a15_neon_fp_vmul_qqd): Likewise.
3835         (cortex_a15_neon_fp_vmla_ddd): Likewise.
3836         (cortex_a15_neon_fp_vmla_qqq): Likewise.
3837         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
3838         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
3839         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
3840         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
3841         (cortex_a15_neon_bp_simple): Likewise.
3842         (cortex_a15_neon_bp_2cycle): Likewise.
3843         (cortex_a15_neon_bp_3cycle): Likewise.
3844         (cortex_a15_neon_vld1_1_2_regs): Likewise.
3845         (cortex_a15_neon_vld1_3_4_regs): Likewise.
3846         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
3847         (cortex_a15_neon_vld2_4_regs): Likewise.
3848         (cortex_a15_neon_vld3_vld4): Likewise.
3849         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
3850         (cortex_a15_neon_vst1_3_4_regs): Likewise.
3851         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Rename to...
3852         (cortex_a15_neon_vst2_4_regs_vst3): ...This, update for new attributes.
3853         (cortex_a15_neon_vst3_vst4): Rename to...
3854         (cortex_a15_neon_vst4): This, update for new attributes.
3855         (cortex_a15_neon_vld1_vld2_lane): Update for new attributes.
3856         (cortex_a15_neon_vld3_vld4_lane): Likewise.
3857         (cortex_a15_neon_vst1_vst2_lane): Likewise.
3858         (cortex_a15_neon_vst3_vst4_lane): Likewise.
3859         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
3860         (cortex_a15_neon_ldm_2): Likewise.
3861         (cortex_a15_neon_stm_2): Likewise.
3862         (cortex_a15_neon_mcr): Likewise.
3863         (cortex_a15_neon_mcr_2_mcrr): Likewise.
3864         (cortex_a15_neon_mrc): Likewise.
3865         (cortex_a15_neon_mrrc): Likewise.
3867         (cortex_a15_neon_abd): New.
3868         (cortex_a15_neon_abd_q): Likewise.
3869         (cortex_a15_neon_aba): Likewise.
3870         (cortex_a15_neon_aba_q): Likewise.
3871         (cortex_a15_neon_acc): Likewise.
3872         (cortex_a15_neon_acc_q): Likewise.
3873         (cortex_a15_neon_arith_basic): Likewise.
3874         (cortex_a15_neon_arith_complex): Likewise.
3875         (cortex_a15_neon_multiply): Likewise.
3876         (cortex_a15_neon_multiply_q): Likewise.
3877         (cortex_a15_neon_mla): Likewise.
3878         (cortex_a15_neon_mla_q): Likewise.
3879         (cortex_a15_neon_sat_mla_long): Likewise.
3880         (cortex_a15_neon_shift_acc): Likewise.
3881         (cortex_a15_neon_shift_imm_basic): Likewise.
3882         (cortex_a15_neon_shift_imm_complex): Likewise.
3883         (cortex_a15_neon_shift_reg_basic): Likewise.
3884         (cortex_a15_neon_shift_reg_basic_q): Likewise.
3885         (cortex_a15_neon_shift_reg_complex): Likewise.
3886         (cortex_a15_neon_shift_reg_complex_q): Likewise.
3887         (cortex_a15_neon_fp_negabs): Likewise
3888         (cortex_a15_neon_fp_arith): Likewise
3889         (cortex_a15_neon_fp_arith_q): Likewise
3890         (cortex_a15_neon_fp_cvt_int): Likewise
3891         (cortex_a15_neon_fp_cvt_int_q): Likewise
3892         (cortex_a15_neon_fp_cvt_16): Likewise
3893         (cortex_a15_neon_fp_mul): Likewise
3894         (cortex_a15_neon_fp_mul_q): Likewise
3895         (cortex_a15_neon_fp_mla): Likewise
3896         (cortex_a15_neon_fp_mla_q): Likewise
3897         (cortex_a15_neon_fp_recps_rsqrte): Likewise.
3898         (cortex_a15_neon_fp_recps_rsqrte_q): Likewise.
3899         (cortex_a15_neon_bitops): Likewise.
3900         (cortex_a15_neon_bitops_q): Likewise.
3901         (cortex_a15_neon_from_gp): Likewise.
3902         (cortex_a15_neon_from_gp_q): Likewise.
3903         (cortex_a15_neon_tbl3_tbl4): Likewise.
3904         (cortex_a15_neon_zip_q): Likewise.
3905         (cortex_a15_neon_to_gp): Likewise.
3906         (cortex_a15_neon_load_a): Likewise.
3907         (cortex_a15_neon_load_b): Likewise.
3908         (cortex_a15_neon_load_c): Likewise.
3909         (cortex_a15_neon_load_d): Likewise.
3910         (cortex_a15_neon_load_e): Likewise.
3911         (cortex_a15_neon_load_f): Likewise.
3912         (cortex_a15_neon_store_a): Likewise.
3913         (cortex_a15_neon_store_b): Likewise.
3914         (cortex_a15_neon_store_c): Likewise.
3915         (cortex_a15_neon_store_d): Likewise.
3916         (cortex_a15_neon_store_e): Likewise.
3917         (cortex_a15_neon_store_f): Likewise.
3918         (cortex_a15_neon_store_g): Likewise.
3919         (cortex_a15_neon_store_h): Likewise.
3920         (cortex_a15_vfp_to_from_gp): Likewise.
3922 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3924         * config/arm/cortex-a9-neon.md (cortex_a9_neon_type): New.
3926         (cortex_a9_neon_vshl_ddd): Remove.
3927         (cortex_a9_neon_vst3_vst4): Likewise.
3928         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
3930         (cortex_a9_neon_bit_ops_q): New.
3932         (cortex_a9_neon_int_1): Use cortex_a8_neon_type.
3933         (cortex_a9_neon_int_2): Likewise.
3934         (cortex_a9_neon_int_3): Likewise.
3935         (cortex_a9_neon_int_4): Likewise.
3936         (cortex_a9_neon_int_5): Likewise.
3937         (cortex_a9_neon_vqneg_vqabs): Likewise.
3938         (cortex_a9_neon_vmov): Likewise.
3939         (cortex_a9_neon_vaba): Likewise.
3940         (cortex_a9_neon_vaba_qqq): Likewise.
3941         (cortex_a9_neon_shift_1): Likewise.
3942         (cortex_a9_neon_shift_2): Likewise.
3943         (cortex_a9_neon_shift_3): Likewise.
3944         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
3945         (cortex_a9_neon_vsra_vrsra): Likewise.
3946         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3947         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
3948         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
3949         Likewise.
3950         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3951         (cortex_a9_neon_mla_qqq_8_16): Likewise.
3952         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
3953         Likewise.
3954         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
3955         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
3956         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
3957         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
3958         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
3959         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
3960         (cortex_a9_neon_fp_vsum): Likewise.
3961         (cortex_a9_neon_fp_vmul_ddd): Likewise.
3962         (cortex_a9_neon_fp_vmul_qqd): Likewise.
3963         (cortex_a9_neon_fp_vmla_ddd): Likewise.
3964         (cortex_a9_neon_fp_vmla_qqq): Likewise.
3965         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
3966         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
3967         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
3968         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
3969         (cortex_a9_neon_bp_simple): Likewise.
3970         (cortex_a9_neon_bp_2cycle): Likewise.
3971         (cortex_a9_neon_bp_3cycle): Likewise.
3972         (cortex_a9_neon_ldr): Likewise.
3973         (cortex_a9_neon_str): Likewise.
3974         (cortex_a9_neon_vld1_1_2_regs): Likewise.
3975         (cortex_a9_neon_vld1_3_4_regs): Likewise.
3976         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
3977         (cortex_a9_neon_vld2_4_regs): Likewise.
3978         (cortex_a9_neon_vld3_vld4): Likewise.
3979         (cortex_a9_neon_vld1_vld2_lane): Likewise.
3980         (cortex_a9_neon_vld3_vld4_lane): Likewise.
3981         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
3982         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
3983         (cortex_a9_neon_vst1_3_4_regs): Likewise.
3984         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
3985         (cortex_a9_neon_vst1_vst2_lane): Likewise.
3986         (cortex_a9_neon_vst3_vst4_lane): Likewise.
3987         (cortex_a9_neon_mcr): Likewise.
3988         (cortex_a9_neon_mcr_2_mcrr): Likewise.
3989         (cortex_a9_neon_mrc): Likewise.
3990         (cortex_a9_neon_mrrc): Likewise.
3992 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3994         * config/arm/cortex-a8-neon.md (cortex_a8_neon_type): New.
3996         (cortex_a8_neon_vshl_ddd): Remove.
3997         (cortex_a8_neon_vst3_vst4): Likewise.
3998         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
4000         (cortex_a8_neon_bit_ops_q): New.
4002         (cortex_a8_neon_int_1): Use cortex_a8_neon_type.
4003         (cortex_a8_neon_int_2): Likewise..
4004         (cortex_a8_neon_int_3): Likewise.
4005         (cortex_a8_neon_int_5): Likewise.
4006         (cortex_a8_neon_vqneg_vqabs): Likewise.
4007         (cortex_a8_neon_int_4): Likewise.
4008         (cortex_a8_neon_vaba): Likewise.
4009         (cortex_a8_neon_vaba_qqq): Likewise.
4010         (cortex_a8_neon_shift_1): Likewise.
4011         (cortex_a8_neon_shift_2): Likewise.
4012         (cortex_a8_neon_shift_3): Likewise.
4013         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
4014         (cortex_a8_neon_vsra_vrsra): Likewise.
4015         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
4016         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
4017         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
4018         Likewise.
4019         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
4020         (cortex_a8_neon_mla_qqq_8_16): Likewise.
4021         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
4022         Likewise.
4023         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
4024         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
4025         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
4026         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
4027         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
4028         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
4029         (cortex_a8_neon_fp_vsum): Likewise.
4030         (cortex_a8_neon_fp_vmul_ddd): Likewise.
4031         (cortex_a8_neon_fp_vmul_qqd): Likewise.
4032         (cortex_a8_neon_fp_vmla_ddd): Likewise.
4033         (cortex_a8_neon_fp_vmla_qqq): Likewise.
4034         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
4035         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
4036         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
4037         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
4038         (cortex_a8_neon_bp_simple): Likewise.
4039         (cortex_a8_neon_bp_2cycle): Likewise.
4040         (cortex_a8_neon_bp_3cycle): Likewise.
4041         (cortex_a8_neon_ldr): Likewise.
4042         (cortex_a8_neon_str): Likewise.
4043         (cortex_a8_neon_vld1_1_2_regs): Likewise.
4044         (cortex_a8_neon_vld1_3_4_regs): Likewise.
4045         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
4046         (cortex_a8_neon_vld2_4_regs): Likewise.
4047         (cortex_a8_neon_vld3_vld4): Likewise.
4048         (cortex_a8_neon_vld1_vld2_lane): Likewise.
4049         (cortex_a8_neon_vld3_vld4_lane): Likewise.
4050         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
4051         (cortex_a8_neon_vst1_3_4_regs): Likewise.
4052         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
4053         (cortex_a8_neon_vst1_vst2_lane): Likewise.
4054         (cortex_a8_neon_vst3_vst4_lane): Likewise.
4055         (cortex_a8_neon_mcr): Likewise.
4056         (cortex_a8_neon_mcr_2_mcrr): Likewise.
4057         (cortex_a8_neon_mrc): Likewise.
4058         (cortex_a8_neon_mrrc): Likewise.
4060 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
4062         * config/aarch64/iterators.md (Vetype): Add SF and DF modes.
4063         (fp): New.
4064         * config/aarch64/aarch64-simd.md (neon_type): Remove.
4065         (aarch64_simd_dup<mode>): Add "type" attribute.
4066         (aarch64_dup_lane<mode>): Likewise.
4067         (aarch64_dup_lane_<vswap_width_name><mode>): Likewise.
4068         (*aarch64_simd_mov<mode>): Likewise.
4069         (aarch64_simd_mov_from_<mode>low): Likewise.
4070         (aarch64_simd_mov_from_<mode>high): Likewise.
4071         (orn<mode>3): Likewise.
4072         (bic<mode>3): Likewise.
4073         (add<mode>3): Likewise.
4074         (sub<mode>3): Likewise.
4075         (mul<mode>3): Likewise.
4076         (*aarch64_mul3_elt<mode>): Likewise.
4077         (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
4078         (*aarch64_mul3_elt_to_128df): Likewise.
4079         (*aarch64_mul3_elt_to_64v2df): Likewise.
4080         (neg<mode>2): Likewise.
4081         (abs<mode>2): Likewise.
4082         (abd<mode>_3): Likewise.
4083         (aba<mode>_3): Likewise.
4084         (fabd<mode>_3): Likewise.
4085         (*fabd_scalar<mode>3): Likewise.
4086         (and<mode>3): Likewise.
4087         (ior<mode>3): Likewise.
4088         (xor<mode>3): Likewise.
4089         (one_cmpl<mode>2): Likewise.
4090         (aarch64_simd_vec_set<mode>): Likewise.
4091         (aarch64_simd_lshr<mode>): Likewise.
4092         (aarch64_simd_ashr<mode>): Likewise.
4093         (aarch64_simd_imm_shl<mode>): Likewise.
4094         (aarch64_simd_reg_sshl<mode): Likewise.
4095         (aarch64_simd_reg_shl<mode>_unsigned): Likewise.
4096         (aarch64_simd_reg_shl<mode>_signed): Likewise.
4097         (aarch64_simd_vec_setv2di): Likewise.
4098         (aarch64_simd_vec_set<mode>): Likewise.
4099         (aarch64_mla<mode>): Likewise.
4100         (*aarch64_mla_elt<mode>): Likewise.
4101         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4102         (aarch64_mls<mode>): Likewise.
4103         (*aarch64_mls_elt<mode>): Likewise.
4104         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4105         (<su><maxmin><mode>3): Likewise.
4106         (move_lo_quad_<mode>): Likewise.
4107         (aarch64_simd_move_hi_quad_<mode>): Likewise.
4108         (aarch64_simd_vec_pack_trunc_<mode>): Likewise.
4109         (vec_pack_trunc_<mode>): Likewise.
4110         (aarch64_simd_vec_unpack<su>_lo_<mode>): Likewise.
4111         (aarch64_simd_vec_unpack<su>_hi_<mode>): Likewise.
4112         (*aarch64_<su>mlal_lo<mode>): Likewise.
4113         (*aarch64_<su>mlal_hi<mode>): Likewise.
4114         (*aarch64_<su>mlsl_lo<mode>): Likewise.
4115         (*aarch64_<su>mlsl_hi<mode>): Likewise.
4116         (*aarch64_<su>mlal<mode>): Likewise.
4117         (*aarch64_<su>mlsl<mode>): Likewise.
4118         (aarch64_simd_vec_<su>mult_lo_<mode>): Likewise.
4119         (aarch64_simd_vec_<su>mult_hi_<mode>): Likewise.
4120         (add<mode>3): Likewise.
4121         (sub<mode>3): Likewise.
4122         (mul<mode>3): Likewise.
4123         (div<mode>3): Likewise.
4124         (neg<mode>2): Likewise.
4125         (abs<mode>2): Likewise.
4126         (fma<mode>4): Likewise.
4127         (*aarch64_fma4_elt<mode>): Likewise.
4128         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4129         (*aarch64_fma4_elt_to_128df): Likewise.
4130         (*aarch64_fma4_elt_to_64v2df): Likewise.
4131         (fnma<mode>4): Likewise.
4132         (*aarch64_fnma4_elt<mode>): Likewise.
4133         (*aarch64_fnma4_elt_<vswap_width_name><mode>
4134         (*aarch64_fnma4_elt_to_128df): Likewise.
4135         (*aarch64_fnma4_elt_to_64v2df): Likewise.
4136         (<frint_pattern><mode>2): Likewise.
4137         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): Likewise.
4138         (<optab><fcvt_target><VDQF:VDQF:mode>2): Likewise.
4139         (vec_unpacks_lo_v4sf): Likewise.
4140         (aarch64_float_extend_lo_v2df): Likewise.
4141         (vec_unpacks_hi_v4sf): Likewise.
4142         (aarch64_float_truncate_lo_v2sf): Likewise.
4143         (aarch64_float_truncate_hi_v4sf): Likewise.
4144         (aarch64_vmls<mode>): Likewise.
4145         (<su><maxmin><mode>3): Likewise.
4146         (<maxmin_uns><mode>3): Likewise.
4147         (reduc_<sur>plus_<mode>): Likewise.
4148         (reduc_<sur>plus_v2di): Likewise.
4149         (reduc_<sur>plus_v2si): Likewise.
4150         (reduc_<sur>plus_<mode>): Likewise.
4151         (aarch64_addpv4sf): Likewise.
4152         (clz<mode>2): Likewise.
4153         (reduc_<maxmin_uns>_<mode>): Likewise.
4154         (reduc_<maxmin_uns>_v2di): Likewise.
4155         (reduc_<maxmin_uns>_v2si): Likewise.
4156         (reduc_<maxmin_uns>_<mode>): Likewise.
4157         (reduc_<maxmin_uns>_v4sf): Likewise.
4158         (aarch64_simd_bsl<mode>_internal): Likewise.
4159         (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
4160         (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
4161         (aarch64_get_lane<mode>): Likewise.
4162         (*aarch64_combinez<mode>): Likewise.
4163         (aarch64_combine<mode>): Likewise.
4164         (aarch64_simd_combine<mode>): Likewise.
4165         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): Likewise.
4166         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): Likewise.
4167         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>): Likewise.
4168         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Likewise.
4169         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Likewise.
4170         (aarch64_<sur>h<addsub><mode>): Likewise.
4171         (aarch64_<sur><addsub>hn<mode>): Likewise.
4172         (aarch64_<sur><addsub>hn2<mode>): Likewise.
4173         (aarch64_pmul<mode>): Likewise.
4174         (aarch64_<su_optab><optab><mode>): Likewise.
4175         (aarch64_<sur>qadd<mode>): Likewise.
4176         (aarch64_sqmovun<mode>): Likewise.
4177         (aarch64_<sur>qmovn<mode>): Likewise.
4178         (aarch64_s<optab><mode>): Likewise.
4179         (aarch64_sq<r>dmulh<mode>): Likewise.
4180         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4181         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4182         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4183         (aarch64_sqdml<SBINQOPS:as>l<mode>): Likewise.
4184         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4185         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4186         (aarch64_sqdml<SBINQOPS:as>l_n<mode>): Likewise.
4187         (aarch64_sqdml<SBINQOPS:as>l2<mode>_internal): Likewise.
4188         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4189         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
4190         (aarch64_sqdmull<mode>): Likewise.
4191         (aarch64_sqdmull_lane<mode>_internal): Likewise.
4192         (aarch64_sqdmull_n<mode>): Likewise.
4193         (aarch64_sqdmull2<mode>_internal): Likewise.
4194         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4195         (aarch64_sqdmull2_n<mode>_internal): Likewise.
4196         (aarch64_<sur>shl<mode>): Likewise.
4197         (aarch64_<sur>q<r>shl<mode>
4198         (aarch64_<sur>shll_n<mode>): Likewise.
4199         (aarch64_<sur>shll2_n<mode>): Likewise.
4200         (aarch64_<sur>shr_n<mode>): Likewise.
4201         (aarch64_<sur>sra_n<mode>): Likewise.
4202         (aarch64_<sur>s<lr>i_n<mode>): Likewise.
4203         (aarch64_<sur>qshl<u>_n<mode>): Likewise.
4204         (aarch64_<sur>q<r>shr<u>n_n<mode>): Likewise.
4205         (aarch64_cm<optab><mode>): Likewise.
4206         (aarch64_cm<optab>di): Likewise.
4207         (aarch64_cm<optab><mode>): Likewise.
4208         (aarch64_cm<optab>di): Likewise.
4209         (aarch64_cmtst<mode>): Likewise.
4210         (aarch64_cmtstdi): Likewise.
4211         (aarch64_cm<optab><mode>): Likewise.
4212         (*aarch64_fac<optab><mode>): Likewise.
4213         (aarch64_addp<mode>): Likewise.
4214         (aarch64_addpdi): Likewise.
4215         (sqrt<mode>2): Likewise.
4216         (vec_load_lanesoi<mode>): Likewise.
4217         (vec_store_lanesoi<mode>): Likewise.
4218         (vec_load_lanesci<mode>): Likewise.
4219         (vec_store_lanesci<mode>): Likewise.
4220         (vec_load_lanesxi<mode>): Likewise.
4221         (vec_store_lanesxi<mode>): Likewise.
4222         (*aarch64_mov<mode>): Likewise.
4223         (aarch64_ld2<mode>_dreg): Likewise.
4224         (aarch64_ld2<mode>_dreg): Likewise.
4225         (aarch64_ld3<mode>_dreg): Likewise.
4226         (aarch64_ld3<mode>_dreg): Likewise.
4227         (aarch64_ld4<mode>_dreg): Likewise.
4228         (aarch64_ld4<mode>_dreg): Likewise.
4229         (aarch64_tbl1<mode>): Likewise.
4230         (aarch64_tbl2v16qi): Likewise.
4231         (aarch64_combinev16qi): Likewise.
4232         (aarch64_<PERMUTE:perm_insn><PERMUTE:perm_hilo><mode>): Likewise.
4233         (aarch64_st2<mode>_dreg): Likewise.
4234         (aarch64_st2<mode>_dreg): Likewise.
4235         (aarch64_st3<mode>_dreg): Likewise.
4236         (aarch64_st3<mode>_dreg): Likewise.
4237         (aarch64_st4<mode>_dreg): Likewise.
4238         (aarch64_st4<mode>_dreg): Likewise.
4239         (*aarch64_simd_ld1r<mode>): Likewise.
4240         (aarch64_frecpe<mode>): Likewise.
4241         (aarch64_frecp<FRECP:frecp_suffix><mode>): Likewise.
4242         (aarch64_frecps<mode>): Likewise.
4244 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
4246         * config/arm/iterators.md (V_elem_ch): New.
4247         (q): Likewise.
4248         (VQH_type): Likewise.
4249         * config/arm/arm.md (is_neon_type): New.
4250         (conds): Use is_neon_type.
4251         (anddi3_insn): Update type attribute.
4252         (xordi3_insn): Likewise.
4253         (one_cmpldi2): Likewise.
4254         * config/arm/vfp.md (movhf_vfp_neon): Update type attribute.
4255         * config/arm/neon.md (neon_mov): Update type attribute.
4256         (*movmisalign<mode>_neon_store): Likewise.
4257         (*movmisalign<mode>_neon_load): Likewise.
4258         (vec_set<mode>_internal): Likewise.
4259         (vec_set<mode>_internal): Likewise.
4260         (vec_setv2di_internal): Likewise.
4261         (vec_extract<mode>): Likewise.
4262         (vec_extract<mode>): Likewise.
4263         (vec_extractv2di): Likewise.
4264         (*add<mode>3_neon): Likewise.
4265         (adddi3_neon): Likewise.
4266         (*sub<mode>3_neon): Likewise.
4267         (subdi3_neon): Likewise.
4268         (fma<VCVTF:mode>4): Likewise.
4269         (fma<VCVTF:mode>4_intrinsic): Likewise.
4270         (*fmsub<VCVTF:mode>4): Likewise.
4271         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
4272         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
4273         (ior<mode>3): Likewise.
4274         (and<mode>3): Likewise.
4275         (orn<mode>3_neon): Likewise.
4276         (orndi3_neon): Likewise.
4277         (bic<mode>3_neon): Likewise.
4278         (bicdi3_neon): Likewise.
4279         (xor<mode>3): Likewise.
4280         (one_cmpl<mode>2): Likewise.
4281         (abs<mode>2): Likewise.
4282         (neg<mode>2): Likewise.
4283         (negdi2_neon): Likewise.
4284         (*umin<mode>3_neon): Likewise.
4285         (*umax<mode>3_neon): Likewise.
4286         (*smin<mode>3_neon): Likewise.
4287         (*smax<mode>3_neon): Likewise.
4288         (vashl<mode>3): Likewise.
4289         (vashr<mode>3_imm): Likewise.
4290         (vlshr<mode>3_imm): Likewise.
4291         (ashl<mode>3_signed): Likewise.
4292         (ashl<mode>3_unsigned): Likewise.
4293         (neon_load_count): Likewise.
4294         (ashldi3_neon_noclobber): Likewise.
4295         (ashldi3_neon): Likewise.
4296         (signed_shift_di3_neon): Likewise.
4297         (unsigned_shift_di3_neon): Likewise.
4298         (ashrdi3_neon_imm_noclobber): Likewise.
4299         (lshrdi3_neon_imm_noclobber): Likewise.
4300         (<shift>di3_neon): Likewise.
4301         (widen_ssum<mode>3): Likewise.
4302         (widen_usum<mode>3): Likewise.
4303         (quad_halves_<code>v4si): Likewise.
4304         (quad_halves_<code>v4sf): Likewise.
4305         (quad_halves_<code>v8hi): Likewise.
4306         (quad_halves_<code>v16qi): Likewise.
4307         (reduc_splus_v2di): Likewise.
4308         (neon_vpadd_internal<mode>): Likewise.
4309         (neon_vpsmin<mode>): Likewise.
4310         (neon_vpsmax<mode>): Likewise.
4311         (neon_vpumin<mode>): Likewise.
4312         (neon_vpumax<mode>): Likewise.
4313         (*ss_add<mode>_neon): Likewise.
4314         (*us_add<mode>_neon): Likewise.
4315         (*ss_sub<mode>_neon): Likewise.
4316         (*us_sub<mode>_neon): Likewise.
4317         (neon_vadd<mode>_unspec): Likewise.
4318         (neon_vaddl<mode>): Likewise.
4319         (neon_vaddw<mode>): Likewise.
4320         (neon_vhadd<mode>): Likewise.
4321         (neon_vqadd<mode>): Likewise.
4322         (neon_vaddhn<mode>): Likewise.
4323         (neon_vmul<mode>): Likewise.
4324         (neon_vfms<VCVTF:mode>): Likewise.
4325         (neon_vmlal<mode>): Likewise.
4326         (neon_vmls<mode>): Likewise.
4327         (neon_vmlsl<mode>): Likewise.
4328         (neon_vqdmulh<mode>): Likewise.
4329         (neon_vqdmlal<mode>): Likewise.
4330         (neon_vqdmlsl<mode>): Likewise.
4331         (neon_vmull<mode>): Likewise.
4332         (neon_vqdmull<mode>): Likewise.
4333         (neon_vsub<mode>_unspec): Likewise.
4334         (neon_vsubl<mode>): Likewise.
4335         (neon_vsubw<mode>): Likewise.
4336         (neon_vqsub<mode>): Likewise.
4337         (neon_vhsub<mode>): Likewise.
4338         (neon_vsubhn<mode>): Likewise.
4339         (neon_vceq<mode>): Likewise.
4340         (neon_vcge<mode>): Likewise.
4341         (neon_vcgeu<mode>): Likewise.
4342         (neon_vcgt<mode>): Likewise.
4343         (neon_vcgtu<mode>): Likewise.
4344         (neon_vcle<mode>): Likewise.
4345         (neon_vclt<mode>): Likewise.
4346         (neon_vcage<mode>): Likewise.
4347         (neon_vcagt<mode>): Likewise.
4348         (neon_vtst<mode>): Likewise.
4349         (neon_vabd<mode>): Likewise.
4350         (neon_vabdl<mode>): Likewise.
4351         (neon_vaba<mode>): Likewise.
4352         (neon_vabal<mode>): Likewise.
4353         (neon_vmax<mode>): Likewise.
4354         (neon_vmin<mode>): Likewise.
4355         (neon_vpaddl<mode>): Likewise.
4356         (neon_vpadal<mode>): Likewise.
4357         (neon_vpmax<mode>): Likewise.
4358         (neon_vpmin<mode>): Likewise.
4359         (neon_vrecps<mode>): Likewise.
4360         (neon_vrsqrts<mode>): Likewise.
4361         (neon_vqabs<mode>): Likewise.
4362         (neon_vqneg<mode>): Likewise.
4363         (neon_vcls<mode>): Likewise.
4364         (clz<mode>2): Likewise.
4365         (popcount<mode>2): Likewise.
4366         (neon_vrecpe<mode>): Likewise.
4367         (neon_vrsqrte<mode>): Likewise.
4368         (neon_vget_lane<mode>_sext_internal): Likewise.
4369         (neon_vget_lane<mode>_zext_internal): Likewise.
4370         (neon_vdup_n<mode>): Likewise.
4371         (neon_vdup_n<mode>): Likewise.
4372         (neon_vdup_nv2di): Likewise.
4373         (neon_vdup_lane<mode>_interal): Likewise.
4374         (*neon_vswp<mode>): Likewise.
4375         (neon_vcombine<mode>): Likewise.
4376         (float<mode><V_cvtto>2): Likewise.
4377         (floatuns<mode><V_cvtto>2): Likewise.
4378         (fix_trunc<mode><V_cvtto>2): Likewise.
4379         (fixuns_trunc<mode><V_cvtto>2
4380         (neon_vcvt<mode>): Likewise.
4381         (neon_vcvt<mode>): Likewise.
4382         (neon_vcvtv4sfv4hf): Likewise.
4383         (neon_vcvtv4hfv4sf): Likewise.
4384         (neon_vcvt_n<mode>): Likewise.
4385         (neon_vcvt_n<mode>): Likewise.
4386         (neon_vmovn<mode>): Likewise.
4387         (neon_vqmovn<mode>): Likewise.
4388         (neon_vqmovun<mode>): Likewise.
4389         (neon_vmovl<mode>): Likewise.
4390         (neon_vmul_lane<mode>): Likewise.
4391         (neon_vmul_lane<mode>): Likewise.
4392         (neon_vmull_lane<mode>): Likewise.
4393         (neon_vqdmull_lane<mode>): Likewise.
4394         (neon_vqdmulh_lane<mode>): Likewise.
4395         (neon_vqdmulh_lane<mode>): Likewise.
4396         (neon_vmla_lane<mode>): Likewise.
4397         (neon_vmla_lane<mode>): Likewise.
4398         (neon_vmlal_lane<mode>): Likewise.
4399         (neon_vqdmlal_lane<mode>): Likewise.
4400         (neon_vmls_lane<mode>): Likewise.
4401         (neon_vmls_lane<mode>): Likewise.
4402         (neon_vmlsl_lane<mode>): Likewise.
4403         (neon_vqdmlsl_lane<mode>): Likewise.
4404         (neon_vext<mode>): Likewise.
4405         (neon_vrev64<mode>): Likewise.
4406         (neon_vrev32<mode>): Likewise.
4407         (neon_vrev16<mode>): Likewise.
4408         (neon_vbsl<mode>_internal): Likewise.
4409         (neon_vshl<mode>): Likewise.
4410         (neon_vqshl<mode>): Likewise.
4411         (neon_vshr_n<mode>): Likewise.
4412         (neon_vshrn_n<mode>): Likewise.
4413         (neon_vqshrn_n<mode>): Likewise.
4414         (neon_vqshrun_n<mode>): Likewise.
4415         (neon_vshl_n<mode>): Likewise.
4416         (neon_vqshl_n<mode>): Likewise.
4417         (neon_vqshlu_n<mode>): Likewise.
4418         (neon_vshll_n<mode>): Likewise.
4419         (neon_vsra_n<mode>): Likewise.
4420         (neon_vsri_n<mode>): Likewise.
4421         (neon_vsli_n<mode>): Likewise.
4422         (neon_vtbl1v8qi): Likewise.
4423         (neon_vtbl2v8qi): Likewise.
4424         (neon_vtbl3v8qi): Likewise.
4425         (neon_vtbl4v8qi): Likewise.
4426         (neon_vtbl1v16qi): Likewise.
4427         (neon_vtbl2v16qi): Likewise.
4428         (neon_vcombinev16qi): Likewise.
4429         (neon_vtbx1v8qi): Likewise.
4430         (neon_vtbx2v8qi): Likewise.
4431         (neon_vtbx3v8qi): Likewise.
4432         (neon_vtbx4v8qi): Likewise.
4433         (*neon_vtrn<mode>_insn): Likewise.
4434         (*neon_vzip<mode>_insn): Likewise.
4435         (*neon_vuzp<mode>_insn): Likewise.
4436         (neon_vld1<mode>): Likewise.
4437         (neon_vld1_lane<mode>): Likewise.
4438         (neon_vld1_lane<mode>): Likewise.
4439         (neon_vld1_dup<mode>): Likewise.
4440         (neon_vld1_dup<mode>): Likewise.
4441         (neon_vld1_dupv2di): Likewise.
4442         (neon_vst1<mode>): Likewise.
4443         (neon_vst1_lane<mode>): Likewise.
4444         (neon_vst1_lane<mode>): Likewise.
4445         (neon_vld2<mode>): Likewise.
4446         (neon_vld2<mode>): Likewise.
4447         (neon_vld2_lane<mode>): Likewise.
4448         (neon_vld2_lane<mode>): Likewise.
4449         (neon_vld2_dup<mode>): Likewise.
4450         (neon_vst2<mode>): Likewise.
4451         (neon_vst2<mode>): Likewise.
4452         (neon_vst2_lane<mode>): Likewise.
4453         (neon_vst2_lane<mode>): Likewise.
4454         (neon_vld3<mode>): Likewise.
4455         (neon_vld3qa<mode>): Likewise.
4456         (neon_vld3qb<mode>): Likewise.
4457         (neon_vld3_lane<mode>): Likewise.
4458         (neon_vld3_lane<mode>): Likewise.
4459         (neon_vld3_dup<mode>): Likewise.
4460         (neon_vst3<mode>): Likewise.
4461         (neon_vst3qa<mode>): Likewise.
4462         (neon_vst3qb<mode>): Likewise.
4463         (neon_vst3_lane<mode>): Likewise.
4464         (neon_vst3_lane<mode>): Likewise.
4465         (neon_vld4<mode>): Likewise.
4466         (neon_vld4qa<mode>): Likewise.
4467         (neon_vld4qb<mode>): Likewise.
4468         (neon_vld4_lane<mode>): Likewise.
4469         (neon_vld4_lane<mode>): Likewise.
4470         (neon_vld4_dup<mode>): Likewise.
4471         (neon_vst4<mode>): Likewise.
4472         (neon_vst4qa<mode>): Likewise.
4473         (neon_vst4qb<mode>): Likewise.
4474         (neon_vst4_lane<mode>): Likewise.
4475         (neon_vst4_lane<mode>): Likewise.
4476         (neon_vec_unpack<US>_lo_<mode>): Likewise.
4477         (neon_vec_unpack<US>_hi_<mode>): Likewise.
4478         (neon_vec_<US>mult_lo_<mode>): Likewise.
4479         (neon_vec_<US>mult_hi_<mode>): Likewise.
4480         (neon_vec_<US>shiftl_<mode>): Likewise.
4481         (neon_unpack<US>_<mode>): Likewise.
4482         (neon_vec_<US>mult_<mode>): Likewise.
4483         (vec_pack_trunc_<mode>): Likewise.
4484         (neon_vec_pack_trunc_<mode>): Likewise.
4485         (neon_vabd<mode>_2): Likewise.
4486         (neon_vabd<mode>_3): Likewise.
4488 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
4490         * config/aarch64/aarch64.md (movtf_aarch64): Update type attribute.
4491         (load_pair): Update type attribute.
4492         (store_pair): Update type attribute.
4493         * config/aarch64/iterators.md (q): New.
4495 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
4497         * config/arm/types.md: Add new types for Neon insns.
4499 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4500             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4501             Sergey Lega  <sergey.s.lega@intel.com>
4502             Anna Tikhonova  <anna.tikhonova@intel.com>
4503             Ilya Tocar  <ilya.tocar@intel.com>
4504             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4505             Ilya Verbin  <ilya.verbin@intel.com>
4506             Kirill Yukhin  <kirill.yukhin@intel.com>
4507             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4509         * config/i386/sse.md (unspec): Add UNSPEC_RCP14, UNSPEC_RSQRT14,
4510         UNSPEC_FIXUPIMM, UNSPEC_SCALEF, UNSPEC_GETEXP, UNSPEC_GETMANT,
4511         UNSPEC_EXP2, UNSPEC_RCP28, UNSPEC_RSQRT28.
4512         (rcp14<mode>): New.
4513         (srcp14<mode>): Ditto.
4514         (rsqrt14<mode>): Ditto.
4515         (rsqrt14<mode>): Ditto.
4516         (avx512f_vmscalef<mode>): Ditto.
4517         (avx512f_scalef<mode>): Ditto.
4518         (avx512f_getexp<mode>): Ditto.
4519         (avx512f_sgetexp<mode>): Ditto.
4520         (avx512f_fixupimm<mode>): Ditto.
4521         (avx512f_sfixupimm<mode>): Ditto.
4522         (avx512f_rndscale<mode>): Ditto.
4523         (*avx512er_exp2<mode>): Ditto.
4524         (*avx512er_rcp28<mode>): Ditto.
4525         (avx512er_rsqrt28<mode>): Ditto.
4526         (avx512f_getmant<mode>): Ditto.
4527         (avx512f_getmant<mode>): Ditto.
4528         (avx512f_rndscale<mode>): Fix formatting.
4530 2013-10-15  Martin Jambor  <mjambor@suse.cz>
4532         * ipa-utils.h (ipa_edge_within_scc): Declare.
4533         * ipa-cp.c (edge_within_scc): Moved...
4534         * ipa-utils.c (ipa_edge_within_scc): ...here.  Updated all callers.
4536 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4537             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4538             Sergey Lega  <sergey.s.lega@intel.com>
4539             Anna Tikhonova  <anna.tikhonova@intel.com>
4540             Ilya Tocar  <ilya.tocar@intel.com>
4541             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4542             Ilya Verbin  <ilya.verbin@intel.com>
4543             Kirill Yukhin  <kirill.yukhin@intel.com>
4544             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4546         * config/i386/predicates.md (const_8_to_15_operand): New.
4547         (const_16_to_31_operand): Ditto.
4548         * config/i386/sse.md (V8FI): New.
4549         (V16FI): Ditto.
4550         (reduc_splus_v8df): Ditto.
4551         (reduc_splus_v16sf): Ditto.
4552         (avx512f_vextract<shuffletype>32x4_1): Ditto.
4553         (vec_extract_hi_<mode>): Ditto.
4554         (avx512f_vinsert<shuffletype>32x4_1): Ditto.
4555         (vec_set_lo_<mode>): Ditto.
4556         (vec_set_hi_<mode>): Ditto.
4557         (avx512f_shuf_<shuffletype>64x2_1): Ditto.
4558         (avx512f_shuf_<shuffletype>32x4_1): Ditto.
4559         (avx512f_pshufd_1): Ditto.
4560         (avx512f_broadcast<mode>): Ditto.
4561         (avx512f_broadcast<mode>): Ditto.
4562         (define_split): Split vec_extract_lo into move.
4563         (ssequartermode): Ditto.
4564         (ssedoublemode): Extened with wider modes.
4565         (vec_extract_lo_<mode>): Ditto.
4567 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4568             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4569             Sergey Lega  <sergey.s.lega@intel.com>
4570             Anna Tikhonova  <anna.tikhonova@intel.com>
4571             Ilya Tocar  <ilya.tocar@intel.com>
4572             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4573             Ilya Verbin  <ilya.verbin@intel.com>
4574             Kirill Yukhin  <kirill.yukhin@intel.com>
4575             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4577         * config/i386/predicates.md (register_or_constm1_operand): New.
4578         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_PCMP, UNSPEC_TESTM,
4579         UNSPEC_TESTNM, UNSPEC_VTERNLOG, UNSPEC_ALIGN, UNSPEC_CONFLICT,
4580         UNSPEC_MASKED_EQ, UNSPEC_MASKED_GT, UNSPEC_GATHER_PREFETCH,
4581         UNSPEC_SCATTER_PREFETCH
4582         (VI48_512): New.
4583         (avx512f_ucmp<mode>3): Ditto.
4584         (avx512f_vternlog<mode>): Ditto.
4585         (avx512f_align<mode>): Ditto.
4586         (<shift_insn><mode>3): Ditto.
4587         (avx512f_<rotate>v<mode>): Ditto.
4588         (avx512f_<rotate><mode>): Ditto.
4589         (avx512f_eq<mode>3): Ditto.
4590         (avx512f_eq<mode>3_1): Ditto.
4591         (avx512f_gt<mode>3): Ditto.
4592         (avx512f_testm<mode>3): Ditto.
4593         (avx512f_testnm<mode>3): Ditto.
4594         (avx512pf_gatherpf<mode>): Ditto.
4595         (*avx512pf_gatherpf<mode>_mask): Ditto.
4596         (*avx512pf_gatherpf<mode>): Ditto.
4597         (avx512pf_scatterpf<mode>): Ditto.
4598         (*avx512pf_scatterpf<mode>_mask): Ditto.
4599         (*avx512pf_scatterpf<mode>): Ditto.
4600         (avx512f_vec_dup_gpr<mode>): Ditto.
4601         (clz<mode>2): Ditto.
4602         (conflict<mode>): Ditto.
4603         (REDUC_SMINMAX_MODE): Extened with wider modes.
4604         (reduc_<code>_<mode>): Ditto.
4605         (vlshr<mode>3): Ditto.
4606         (vashl<mode>3): Ditto.
4608 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4609             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4610             Sergey Lega  <sergey.s.lega@intel.com>
4611             Anna Tikhonova  <anna.tikhonova@intel.com>
4612             Ilya Tocar  <ilya.tocar@intel.com>
4613             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4614             Ilya Verbin  <ilya.verbin@intel.com>
4615             Kirill Yukhin  <kirill.yukhin@intel.com>
4616             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4618         * config/i386/sse.md (unspec): Added UNSPEC_VPERMI2, UNSPEC_VPERMT2,
4619         UNSPEC_SCATTER.
4620         (VI48F_512): New.
4621         (avx512fmaskmode): Ditto.
4622         (bcstscalarsuff): Ditto.
4623         (avx512f_blendm<mode>): Ditto.
4624         (cmp_imm_predicate): Ditto.
4625         (avx512f_cmp<mode>3): Ditto.
4626         (avx512f_vec_dup<mode>): Ditto.
4627         (avx512f_vec_dup_mem<mode>): Ditto.
4628         (avx512f_vpermi2var<mode>3): Ditto.
4629         (avx512f_vpermt2var<mode>3): Ditto.
4630         (vec_init<mode>): Ditto.
4631         (avx512f_gathersi<mode>): Ditto.
4632         (*avx512f_gathersi<mode>): Ditto.
4633         (*avx512f_gathersi<mode>_2): Ditto.
4634         (avx512f_gatherdi<mode>): Ditto.
4635         (*avx512f_gatherdi<mode>): Ditto.
4636         (*avx512f_gatherdi<mode>_2): Ditto.
4637         (avx512f_scattersi<mode>): Ditto.
4638         (*avx512f_scattersi<mode>): Ditto.
4639         (avx512f_scatterdi<mode>): Ditto.
4640         (*avx512f_scatterdi<mode>): Ditto.
4641         (sseintprefix): Extened with wider modes.
4642         (VEC_GATHER_IDXSI): Ditto.
4643         (VEC_GATHER_IDXDI): Ditto.
4644         (VEC_GATHER_SRCDI): Ditto.
4646 2013-10-15  Matthew Gretton-Dann  <matthew.gretton-dann@arm.com>
4647             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4649         * config/arm/t-aprofile: New file.
4650         * config.gcc: Handle --with-multilib-list option.
4652 2013-10-15  Bernd Schmidt  <bernds@codesourcery.com>
4654         * reload1.c (reloads_unique_chain_p): Ensure that r1 is
4655         the input for r2.
4657 2013-10-15  Richard Biener  <rguenther@suse.de>
4659         * tree-loop-distribution.c (build_empty_rdg): Inline into
4660         single user.
4661         (rdg_flag_vertex): Inline into single user.
4662         (rdg_flag_vertex_and_dependent): Likewise.
4663         (build_rdg_partition_for_vertex): Remove processed bitmap.
4664         (rdg_build_partitions): Simplify.
4666 2013-10-15  Richard Biener  <rguenther@suse.de>
4668         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1):
4669         Restructure forwarding through conversions and copies to
4670         avoid performing copy-propagation the wrong way.  Adjust
4671         recursion invocations.
4672         (forward_propagate_addr_expr): Add argument stating if we
4673         are recursing from a single-use.
4674         (ssa_forward_propagate_and_combine): Adjust.
4676 2013-10-14  David Malcolm  <dmalcolm@redhat.com>
4678         * dumpfile.h (gcc::dump_manager): New class, to hold state
4679         relating to dumpfile management.
4680         (get_dump_file_name): Remove in favor of method of dump_manager.
4681         (dump_initialized_p): Likewise.
4682         (dump_start): Likewise.
4683         (dump_finish): Likewise.
4684         (dump_switch_p): Likewise.
4685         (dump_register): Likewise.
4686         (get_dump_file_info): Likewise.
4687         * context.c (gcc::context::context): Construct the dump_manager
4688         instance.
4689         * context.h (gcc::context::get_dumps): New.
4690         (gcc::context::m_dumps): New.
4691         * coverage.c (coverage_init): Port to dump_manager API.
4692         * dumpfile.c (extra_dump_files): Convert to field of gcc::dump_manager.
4693         (extra_dump_files_in_use): Likewise.
4694         (extra_dump_files_alloced): Likewise.
4695         (gcc::dump_manager::dump_manager): New.
4696         (dump_register): Convert to...
4697         (gcc::dump_manager::dump_register): ...method, replacing
4698         function-static next_dump with m_next_dump field.
4699         (get_dump_file_info): Convert to...
4700         (gcc::dump_manager::get_dump_file_info): ...method.
4701         (get_dump_file_name): Convert to...
4702         (gcc::dump_manager::get_dump_file_name): ...method.
4703         (dump_start): Convert to...
4704         (gcc::dump_manager::dump_start): ...method.
4705         (dump_finish): Convert to...
4706         (gcc::dump_manager::dump_finish): ...method.
4707         (dump_begin): Replace body with...
4708         (gcc::dump_manager::dump_begin): ...new method.
4709         (dump_phase_enabled_p): Convert to...
4710         (gcc::dump_manager::dump_phase_enabled_p): ...method.
4711         (dump_phase_enabled_p): Convert to...
4712         (gcc::dump_manager::dump_phase_enabled_p): ...method.
4713         (dump_initialized_p):  Convert to...
4714         (gcc::dump_manager::dump_initialized_p): ...method.
4715         (dump_flag_name): Replace body with...
4716         (gcc::dump_manager::dump_flag_name): ...new method.
4717         (dump_enable_all): Convert to...
4718         (gcc::dump_manager::dump_enable_all): ...new method.
4719         (opt_info_enable_passes): Convert to...
4720         (gcc::dump_manager::opt_info_enable_passes): ...new method.
4721         (dump_switch_p_1): Convert to...
4722         (gcc::dump_manager::dump_switch_p_1): ...new method.
4723         (dump_switch_p):  Convert to...
4724         (gcc::dump_manager::dump_switch_p): ...new method.
4725         (opt_info_switch_p): Port to dump_manager API.
4726         (enable_rtl_dump_file): Likewise.
4727         * opts-global.c (handle_common_deferred_options): Port to new
4728         dump_manager API.
4729         * passes.c (pass_manager::finish_optimization_passes): Likewise.
4730         (pass_manager::register_one_dump_file): Likewise.
4731         (pass_manager::register_pass): Likewise.
4732         (pass_init_dump_file): Likewise.
4733         (pass_fini_dump_file): Likewise.
4734         * statistics.c (statistics_early_init): Likewise.
4736 2013-10-14  Richard Biener  <rguenther@suse.de>
4738         * gimple.c (gimple_canonical_types, canonical_type_hash_cache,
4739         iterative_hash_canonical_type, gimple_canonical_type_hash,
4740         gimple_canonical_types_compatible_p, gimple_canonical_type_eq,
4741         gimple_register_canonical_type, print_gimple_types_stats,
4742         free_gimple_type_tables): Move to lto/lto.c
4743         (gt-gimple.h): Do not include.
4744         * gimple.h (gimple_register_canonical_type,
4745         print_gimple_types_stats, free_gimple_type_tables): Remove.
4746         * Makefile.in (GTFILES): Remove gimple.c.
4748 2013-10-14  Travis Snoozy  <quandary@remstate.com>
4750         PR target/58716
4751         * config/msp430/msp430.c (msp430_option_override): Correct thinko
4752         scanning for msp430x targets.
4754 2013-10-14  Eric Botcazou  <ebotcazou@adacore.com>
4756         PR bootstrap/58509
4757         * config/sparc/sparc-protos.h (widen_mem_for_ldd_peep): Declare.
4758         (registers_ok_for_ldd_peep): Move around.
4759         * config/sparc/sparc.c (widen_mem_for_ldd_peep): New.
4760         * config/sparc/sparc.md (widening peepholes): Use it.
4762 2013-10-14  Richard Biener  <rguenther@suse.de>
4764         PR middle-end/58712
4765         PR middle-end/55358
4766         * gimple.c (iterative_hash_canonical_type): Make sure to
4767         record the hash into the correct hashtable slot.
4769 2013-10-13  Eric Botcazou  <ebotcazou@adacore.com>
4771         PR rtl-optimization/58662
4772         * combine.c (try_combine): Take into account death nodes on I2 when
4773         splitting a PARALLEL of two independent SETs.  Fix dump message.
4775 2013-10-12  Oleg Endo  <olegendo@gcc.gnu.org>
4777         PR target/51244
4778         * config/sh/sh_treg_combine.cc: New SH specific RTL pass.
4779         * config.gcc (SH extra_objs): Add sh_ifcvt.o.
4780         * config/sh/t-sh (sh_treg_combine.o): New entry.
4781         * config/sh/sh.c (sh_fixed_condition_code_regs): New function that
4782         implements the target hook TARGET_FIXED_CONDITION_CODE_REGS.
4783         (register_sh_passes): New function.  Register sh_treg_combine pass.
4784         (sh_option_override): Invoke it.
4785         (sh_canonicalize_comparison): Handle op0_preserve_value.
4786         * sh.md (*cbranch_t"): Do not try to optimize missed test and branch
4787         opportunities.  Canonicalize branch condition.
4788         (nott): Allow only if pseudos can be created for non-SH2A.
4790 2013-10-12  H.J. Lu  <hongjiu.lu@intel.com>
4792         PR target/58690
4793         * config/i386/i386.c (ix86_copy_addr_to_reg): New function.
4794         (ix86_expand_movmem): Replace copy_addr_to_reg with
4795         ix86_copy_addr_to_reg.
4796         (ix86_expand_setmem): Likewise.
4798 2013-10-12  Alexander Monakov  <amonakov@ispras.ru>
4800         * config/i386/i386.c (ix86_expand_sse_compare_and_jump): Use mode
4801         provided by ix86_fp_compare_mode instead of CCFPUmode.
4803 2013-10-12  James Greenhalgh  <james.greenhalgh@arm.com>
4805         * config/aarch64/arm_neon.h
4806         (vtbx<1,3>_<psu>8): Fix register constriants.
4808 2013-10-11  Jeff Law  <law@redhat.com>
4810         PR tree-optimization/58640
4811         * tree-ssa-threadupdate.c (mark_threaded_blocks): Truncate jump
4812         threading paths that cross over two loop entry points.
4814 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4816         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): Generalize to
4817         handle vector float as well.
4818         (*vsx_le_perm_load_v4si): Likewise.
4819         (*vsx_le_perm_store_v2di): Likewise.
4820         (*vsx_le_perm_store_v4si): Likewise.
4822 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4824         * config/rs6000/vector.md (vec_realign_load<mode>): Generate vperm
4825         directly to circumvent subtract from splat{31} workaround.
4826         * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_le): New
4827         prototype.
4828         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): New.
4829         * config/rs6000/altivec.md (define_c_enum "unspec"): Add
4830         UNSPEC_VPERM_X and UNSPEC_VPERM_UNS_X.
4831         (altivec_vperm_<mode>): Convert to define_insn_and_split to
4832         separate big and little endian logic.
4833         (*altivec_vperm_<mode>_internal): New define_insn.
4834         (altivec_vperm_<mode>_uns): Convert to define_insn_and_split to
4835         separate big and little endian logic.
4836         (*altivec_vperm_<mode>_uns_internal): New define_insn.
4837         (vec_permv16qi): Add little endian logic.
4839 2013-10-11  Marc Glisse  <marc.glisse@inria.fr>
4841         * doc/extend.texi (returns_nonnull): Remove arguments.
4843 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4844             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4845             Sergey Lega  <sergey.s.lega@intel.com>
4846             Anna Tikhonova  <anna.tikhonova@intel.com>
4847             Ilya Tocar  <ilya.tocar@intel.com>
4848             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4849             Ilya Verbin  <ilya.verbin@intel.com>
4850             Kirill Yukhin  <kirill.yukhin@intel.com>
4851             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4853         * config/i386/sse.md (VI48F_256_512): New.
4854         (avx2_permvar<mode>): Change to ...
4855         (<avx2_avx512f>_permvar<mode>): This.
4857 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4858             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4859             Sergey Lega  <sergey.s.lega@intel.com>
4860             Anna Tikhonova  <anna.tikhonova@intel.com>
4861             Ilya Tocar  <ilya.tocar@intel.com>
4862             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4863             Ilya Verbin  <ilya.verbin@intel.com>
4864             Kirill Yukhin  <kirill.yukhin@intel.com>
4865             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4867         * config/i386/i386.c (bdesc_args): Change corresponding pattern for
4868         __builtin_ia32_cvtps2dq, __builtin_ia32_cvtps2dq256.
4869         * config/i386/sse.md (VI4_AVX): New.
4870         (sf2simodelower): Ditto.
4871         (sse2_cvtps2dq): Change to ...
4872         (<sse2_avx_avx512f>_fix_notrunc<sf2simodelower><mode>): This.
4874 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4875             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4876             Sergey Lega  <sergey.s.lega@intel.com>
4877             Anna Tikhonova  <anna.tikhonova@intel.com>
4878             Ilya Tocar  <ilya.tocar@intel.com>
4879             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4880             Ilya Verbin  <ilya.verbin@intel.com>
4881             Kirill Yukhin  <kirill.yukhin@intel.com>
4882             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4884         * config/i386/sse.md (V_512): New.
4885         (VI_512): Ditto.
4886         (vcond<V_512:mode><VF_512:mode>): Ditto.
4887         (vcond<V_512:mode><VI_512:mode>): Ditto.
4888         (vcondu<V_512:mode><VI_512:mode>): Ditto.
4890 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4891             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4892             Sergey Lega  <sergey.s.lega@intel.com>
4893             Anna Tikhonova  <anna.tikhonova@intel.com>
4894             Ilya Tocar  <ilya.tocar@intel.com>
4895             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4896             Ilya Verbin  <ilya.verbin@intel.com>
4897             Kirill Yukhin  <kirill.yukhin@intel.com>
4898             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4900         * config/i386/i386.c (ix86_rtx_costs): Enable fma for TARGET_AVX512F.
4901         * config/i386/sse.md (FMAMODEM): Changed modes and conditions.
4902         (FMAMODE): Ditto.
4903         (fma<mode>4): Removed condition.
4904         (fms<mode>4): Ditto.
4905         (fnma<mode>4): Ditto.
4906         (fnms<mode>4): Ditto.
4907         (fma4i_fmadd_<mode>): Ditto.
4908         (*fma_fmadd_<mode>): Ditto.
4909         (*fma_fmsub_<mode>): Ditto.
4910         (*fma_fnmadd_<mode>): Ditto.
4911         (*fma_fnmsub_<mode>): Ditto.
4912         (fmaddsub_<mode>): Allow for TARGET_AVX512F.
4913         (*fma_fmaddsub_<mode>): Ditto.
4914         (*fma_fmsubadd_<mode>): Ditto.
4915         (*fmai_fmadd_<mode>): Ditto.
4916         (*fmai_fmsub_<mode>): Ditto.
4917         (*fmai_fnmadd_<mode>): Ditto.
4918         (*fmai_fnmsub_<mode>): Ditto.
4920 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4921             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4922             Sergey Lega  <sergey.s.lega@intel.com>
4923             Anna Tikhonova  <anna.tikhonova@intel.com>
4924             Ilya Tocar  <ilya.tocar@intel.com>
4925             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4926             Ilya Verbin  <ilya.verbin@intel.com>
4927             Kirill Yukhin  <kirill.yukhin@intel.com>
4928             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4930         * config/i386/sse.md (VI248_AVX2_8_AVX512F): New.
4931         (VI124_256): Changed to ...
4932         (VI124_256_48_512): This.
4933         (ssepackmode): Extended with wider modes.
4934         (<code><mode>3): Changed iterator.
4935         (*avx2_<code><mode>3): Ditto.
4936         (vec_pack_trunc_<mode>): Ditto.
4938 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4939             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4940             Sergey Lega  <sergey.s.lega@intel.com>
4941             Anna Tikhonova  <anna.tikhonova@intel.com>
4942             Ilya Tocar  <ilya.tocar@intel.com>
4943             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4944             Ilya Verbin  <ilya.verbin@intel.com>
4945             Kirill Yukhin  <kirill.yukhin@intel.com>
4946             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4948         * config/i386/sse.md (VI124_AVX2_48_AVX512F): New.
4949         (VI8F_256_512): Ditto.
4950         (abs<mode>2): Changed iterator.
4951         (avx2_perm<mode>): Changed to ...
4952         (<avx2_avx512f>_perm<mode>): This.
4953         (avx2_perm<mode>_1): Changed to ...
4954         (<avx2_avx512f>_perm<mode>_1): This.
4956 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4957             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4958             Sergey Lega  <sergey.s.lega@intel.com>
4959             Anna Tikhonova  <anna.tikhonova@intel.com>
4960             Ilya Tocar  <ilya.tocar@intel.com>
4961             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4962             Ilya Verbin  <ilya.verbin@intel.com>
4963             Kirill Yukhin  <kirill.yukhin@intel.com>
4964             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4966         * config/i386/sse.md (VI48_AVX512F): New.
4967         (VI48_AVX2): Changed to ...
4968         (VI48_AVX2_48_AVX512F): This.
4969         (avx2_ashrv<mode>): Changed to ...
4970         (<avx2_avx512f>_ashrv<mode>): This.
4971         (avx2_<shift_insn>v<mode>): Changed to ...
4972         (<avx2_avx512f>_<shift_insn>v<mode>): This.
4974 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4975             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4976             Sergey Lega  <sergey.s.lega@intel.com>
4977             Anna Tikhonova  <anna.tikhonova@intel.com>
4978             Ilya Tocar  <ilya.tocar@intel.com>
4979             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4980             Ilya Verbin  <ilya.verbin@intel.com>
4981             Kirill Yukhin  <kirill.yukhin@intel.com>
4982             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4984         * config/i386/sse.md (VI4_AVX512F): New.
4985         (VI8_AVX2_AVX512F): Ditto.
4986         (mul<mode>3): Extended with wider modes.
4987         (*<sse4_1_avx2>_mul<mode>3): Ditto.
4988         (mul<mode>3): Ditto.
4989         (vec_widen_<s>mult_odd_<mode>): Ditto.
4991 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4992             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4993             Sergey Lega  <sergey.s.lega@intel.com>
4994             Anna Tikhonova  <anna.tikhonova@intel.com>
4995             Ilya Tocar  <ilya.tocar@intel.com>
4996             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4997             Ilya Verbin  <ilya.verbin@intel.com>
4998             Kirill Yukhin  <kirill.yukhin@intel.com>
4999             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
5001         * config/i386/sse.md (VI2_AVX512F): New.
5002         (VI124_AVX512F): Ditto.
5003         (sseunpackmode): Extended with wider modes.
5004         (sseunpackfltmode): Ditto.
5005         (vec_unpacks_float_hi_<mode>): Ditto.
5006         (vec_unpacks_float_lo_<mode>): Ditto.
5007         (vec_unpacku_float_hi_<mode>): Ditto.
5008         (vec_unpacku_float_lo_<mode>): Ditto.
5009         (vec_unpacks_lo_<mode>): Ditto.
5010         (vec_unpacks_hi_<mode>): Ditto.
5011         (vec_unpacku_lo_<mode>): Ditto.
5012         (vec_unpacku_hi_<mode>): Ditto.
5014 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5015             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
5016             Sergey Lega  <sergey.s.lega@intel.com>
5017             Anna Tikhonova  <anna.tikhonova@intel.com>
5018             Ilya Tocar  <ilya.tocar@intel.com>
5019             Andrey Turetskiy  <andrey.turetskiy@intel.com>
5020             Ilya Verbin  <ilya.verbin@intel.com>
5021             Kirill Yukhin  <kirill.yukhin@intel.com>
5022             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
5024         * config/i386/i386.md (multdiv): New.
5025         (multdiv_mnemonic): Ditto.
5026         * config/i386/sse.md (<sse>_vmmul<mode>3): Changed to...
5027         (<sse>_vm<multdiv_mnemonic><mode>3): This.
5028         (<sse>_vmdiv<mode>3): Removed.
5030 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5031             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
5032             Sergey Lega  <sergey.s.lega@intel.com>
5033             Anna Tikhonova  <anna.tikhonova@intel.com>
5034             Ilya Tocar  <ilya.tocar@intel.com>
5035             Andrey Turetskiy  <andrey.turetskiy@intel.com>
5036             Ilya Verbin  <ilya.verbin@intel.com>
5037             Kirill Yukhin  <kirill.yukhin@intel.com>
5038             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
5040         * config/i386/sse.md (V): Extended with wider modes.
5041         (VF2): Ditto.
5042         (ssehalfvecmode): Ditto.
5043         (i128): Ditto.
5044         (ssepackfltmode): Ditto.
5045         (avx_vec_concat<mode>): Ditto.
5046         (V_256_512): New iterator.
5047         (VF2_512_256): Ditto.
5048         (si2dfmode): New attribute.
5049         (si2dfmodelower): Ditto.
5050         (sf2dfmode): Ditto.
5051         (concat_tg_mode): Ditto.
5052         (floatv4siv4df2): Changed to ...
5053         (float<si2dfmodelower><mode>2): This.
5054         (avx_cvtps2pd256): Changed to ...
5055         (<sse2_avx_avx512f>_cvtps2pd<avxsizesuffix>): This.
5056         (vec_pack_trunc_v4df): Changed to ...
5057         (vec_pack_trunc_<mode>): This.
5058         (avx_vpermil<mode>): Changed to ...
5059         (<sse2_avx_avx512f>_vpermil<mode>): This.
5060         (<fixsuffix>fix_truncv8dfv8si2): New.
5061         (vec_pack_sfix_trunc_v8df): Ditto.
5062         (avx512f_rndscale<mode>): Ditto.
5063         (avx512f_roundpd512): Ditto.
5064         (vec_pack_ufix_trunc_<mode>): Updated iterator.
5066 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5067             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
5068             Sergey Lega  <sergey.s.lega@intel.com>
5069             Anna Tikhonova  <anna.tikhonova@intel.com>
5070             Ilya Tocar  <ilya.tocar@intel.com>
5071             Andrey Turetskiy  <andrey.turetskiy@intel.com>
5072             Ilya Verbin  <ilya.verbin@intel.com>
5073             Kirill Yukhin  <kirill.yukhin@intel.com>
5074             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
5076         * config/i386/i386.md (any_fix): New iterator.
5077         (fixsuffix): New attribute.
5078         * config/i386/sse.md (VF1): Extened with wider modes.
5079         (VI): Ditto.
5080         (VI_AVX2): Ditto.
5081         (VI8): Ditto.
5082         (sseintvecmodelower): Ditto.
5083         (ssescalarmode): Ditto.
5084         (ssescalarnum): Ditto.
5085         (VF1_128_256): New.
5086         (ssexmmmode): Ditto.
5087         (<fixsuffix>fix_truncv16sfv16si2): Ditto.
5088         (<sse>_rcp<mode>2): Change iterator.
5089         (rsqrt<mode>2): Ditto.
5090         (<sse>_rsqrt<mode>2): Ditto.
5091         (avx2_vec_dup<mode>): Ditto.
5092         (<sse4_1>_round<ssemodesuffix>_sfix<avxsizesuffix>): Ditto.
5093         (round<mode>2_sfix): Ditto.
5094         (avx2_pbroadcast<mode>): Ditto.
5095         (*andnot<mode>3): Handle XI mode.
5096         (*<code><mode>3): Ditto.
5097         (AVXTOSSEMODE): Removed.
5098         (avx_vpermil<mode>): Changed to ...
5099         (<sse2_avx_avx512f>_vpermil<mode>): This.
5101 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5102             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
5103             Sergey Lega  <sergey.s.lega@intel.com>
5104             Anna Tikhonova  <anna.tikhonova@intel.com>
5105             Ilya Tocar  <ilya.tocar@intel.com>
5106             Andrey Turetskiy  <andrey.turetskiy@intel.com>
5107             Ilya Verbin  <ilya.verbin@intel.com>
5108             Kirill Yukhin  <kirill.yukhin@intel.com>
5109             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
5111         * config/i386/sse.md (<sse2>_movnt<mode>): Update constraint to "v".
5112         (<sse>_comi): Ditto.
5113         (<sse>_ucomi): Ditto.
5114         (sse_cvtss2siq_2): Ditto.
5115         (sse2_cvtsd2si): Ditto.
5116         (sse2_cvtsd2siq): Ditto.
5117         (sse2_cvttsd2si): Ditto.
5118         (sse2_cvttsd2siq): Ditto.
5119         (<shift_insn><mode>3): Ditto.
5120         (sse2_cvtsi2sdq): Update constraint and prefix.
5121         (sse_cvtsi2ss): Update prefix.
5122         (sse_cvtsi2ssq): Ditto.
5124 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
5126         * tree-vrp.c (infer_nonnull_range): Use is_gimple_call,
5127         ignore internal calls.
5129 2013-10-11  Richard Biener  <rguenther@suse.de>
5131         * tree-pretty-print.c (dump_generic_node): Allow to dump both (D)
5132         and (ab) for SSA_NAMEs.  Mark INTEGER_CSTs with (OVF) if
5133         TREE_OVERFLOW is set.
5135 2013-10-11  Thomas Schwinge  <thomas@codesourcery.com>
5137         * tree.h (OMP_CLAUSE_CODE): Remove duplicate definition.
5139         * gimple.c: GIMPLE statements have subcodes, not sub-codes.
5140         * gimple.h: Likewise.
5142         * doc/generic.texi (OpenMP): OMP_CLAUSE_* are subcodes, not sub-codes.
5144         * doc/generic.texi (Adding new DECL node types): Explain *_CHECK
5145         macros.
5147         * doc/gimple.texi (is_gimple_omp): Move into the correct section.
5149         * acinclude.m4 (gcc_GAS_FLAGS): Add more gcc_cv_as_flags overrides.
5150         * configure: Regenerate.
5152 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
5154         * tree-pretty-print.c (dump_omp_clause): Handle OMP_CLAUSE__LOOPTEMP_
5155         and new OpenMP 4.0 clauses, handle UDR OMP_CLAUSE_REDUCTION,
5156         formatting fixes, use pp_colon instead of pp_character (..., ':'),
5157         similarly pp_right_paren.
5158         (dump_generic_node): Handle OMP_DISTRIBUTE, OMP_TEAMS,
5159         OMP_TARGET_DATA, OMP_TARGET, OMP_TARGET_UPDATE, OMP_TASKGROUP,
5160         allow OMP_FOR_INIT to be NULL, handle OMP_ATOMIC_SEQ_CST.
5161         * tree.c (omp_clause_num_ops, omp_clause_code_name): Add OpenMP 4.0
5162         clauses.
5163         (omp_declare_simd_clauses_equal,
5164         omp_remove_redundant_declare_simd_attrs): New functions.
5165         (attribute_value_equal): Use omp_declare_simd_clauses_equal.
5166         (walk_tree_1): Handle new OpenMP 4.0 clauses.
5167         * tree.h (OMP_LOOP_CHECK): Define.
5168         (OMP_FOR_BODY, OMP_FOR_CLAUSES, OMP_FOR_INIT, OMP_FOR_COND,
5169         OMP_FOR_INCR, OMP_FOR_PRE_BODY): Use it.
5170         (OMP_TASKGROUP_BODY, OMP_TEAMS_BODY, OMP_TEAMS_CLAUSES,
5171         OMP_TARGET_DATA_BODY, OMP_TARGET_DATA_CLAUSES, OMP_TARGET_BODY,
5172         OMP_TARGET_CLAUSES, OMP_TARGET_UPDATE_CLAUSES, OMP_CLAUSE_SIZE,
5173         OMP_ATOMIC_SEQ_CST, OMP_CLAUSE_DEPEND_KIND, OMP_CLAUSE_MAP_KIND,
5174         OMP_CLAUSE_MAP_ZERO_BIAS_ARRAY_SECTION, OMP_CLAUSE_PROC_BIND_KIND,
5175         OMP_CLAUSE_REDUCTION_OMP_ORIG_REF, OMP_CLAUSE_ALIGNED_ALIGNMENT,
5176         OMP_CLAUSE_NUM_TEAMS_EXPR, OMP_CLAUSE_THREAD_LIMIT_EXPR,
5177         OMP_CLAUSE_DEVICE_ID, OMP_CLAUSE_DIST_SCHEDULE_CHUNK_EXPR,
5178         OMP_CLAUSE_SIMDLEN_EXPR): Define.
5179         (OMP_CLAUSE_DECL): Change range up to OMP_CLAUSE__LOOPTEMP_.
5180         (omp_remove_redundant_declare_simd_attrs): New prototype.
5181         * gimple.def (GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET,
5182         GIMPLE_OMP_TEAMS): New codes.
5183         (GIMPLE_OMP_RETURN): Use GSS_OMP_ATOMIC_STORE instead of GSS_BASE.
5184         * omp-low.c (struct omp_context): Add cancel_label and cancellable
5185         fields.
5186         (target_nesting_level): New variable.
5187         (extract_omp_for_data): Handle GF_OMP_FOR_KIND_DISTRIBUTE and
5188         OMP_CLAUSE_DIST_SCHEDULE.  Don't fallback to library implementation
5189         for collapse > 1 static schedule unless ordered.
5190         (get_ws_args_for): Add par_stmt argument.  Handle combined loops.
5191         (determine_parallel_type): Adjust get_ws_args_for caller.
5192         (install_var_field): Handle mask & 4 for double indirection.
5193         (scan_sharing_clauses): Ignore shared clause on teams construct.
5194         Handle OMP_CLAUSE__LOOPTEMP_ and new OpenMP 4.0 clauses.
5195         (create_omp_child_function): If inside target or declare target
5196         constructs, set "omp declare target" attribute on the child function.
5197         (find_combined_for): New function.
5198         (scan_omp_parallel): Handle combined loops.
5199         (scan_omp_target, scan_omp_teams): New functions.
5200         (check_omp_nesting_restrictions): Check new OpenMP 4.0 nesting
5201         restrictions and set ctx->cancellable for cancellable constructs.
5202         (scan_omp_1_stmt): Call check_omp_nesting_restrictions also on
5203         selected builtin calls.  Handle GIMPLE_OMP_TASKGROUP,
5204         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS.
5205         (build_omp_barrier): Add lhs argument, return gimple rather than tree.
5206         (omp_clause_aligned_alignment): New function.
5207         (lower_rec_simd_input_clauses): Only call SET_DECL_VALUE_EXPR on decls.
5208         (lower_rec_input_clauses): Add FD argument.  Ignore shared clauses
5209         on teams constructs.  Handle user defined reductions and new
5210         OpenMP 4.0 clauses.
5211         (lower_reduction_clauses): Don't set placeholder to address of ref
5212         if it has already the right type.
5213         (lower_send_clauses): Handle OMP_CLAUSE__LOOPTEMP_.
5214         (expand_parallel_call): Use the new non-_start suffixed builtins,
5215         handle OMP_CLAUSE_PROC_BIND, don't call the outlined function
5216         and GOMP_parallel_end after the call.
5217         (expand_task_call): Handle OMP_CLAUSE_DEPEND.
5218         (expand_omp_for_init_counts): Handle combined loops.
5219         (expand_omp_for_init_vars): Add inner_stmt argument, handle combined
5220         loops.
5221         (expand_omp_for_generic): Likewise.  Use GOMP_loop_end_cancel at the
5222         end of cancellable loops.
5223         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
5224         Likewise.  Handle collapse > 1 loops.
5225         (expand_omp_simd): Handle combined loops.
5226         (expand_omp_for): Add inner_stmt argument, adjust callers of
5227         expand_omp_for* functions, use expand_omp_for_static*chunk even
5228         for collapse > 1 unless ordered.
5229         (expand_omp_sections): Use GOMP_sections_end_cancel at the end
5230         of cancellable sections.
5231         (expand_omp_single): Remove need_barrier variable, just rely on
5232         gimple_omp_return_nowait_p.  Adjust build_omp_barrier caller.
5233         (expand_omp_synch): Allow GIMPLE_OMP_TASKGROUP and GIMPLE_OMP_TEAMS.
5234         (expand_omp_atomic_load, expand_omp_atomic_store,
5235         expand_omp_atomic_fetch_op): Handle gimple_omp_atomic_seq_cst_p.
5236         (expand_omp_target): New function.
5237         (expand_omp): Handle combined loops.  Handle GIMPLE_OMP_TASKGROUP,
5238         GIMPLE_OMP_TEAMS, GIMPLE_OMP_TARGET.
5239         (build_omp_regions_1): Immediately close region for
5240         GF_OMP_TARGET_KIND_UPDATE.
5241         (maybe_add_implicit_barrier_cancel): New function.
5242         (lower_omp_sections): Adjust lower_rec_input_clauses caller.  Handle
5243         cancellation.
5244         (lower_omp_single): Likewise.  Add clobber after the barrier.
5245         (lower_omp_taskgroup): New function.
5246         (lower_omp_for): Handle combined loops.  Adjust
5247         lower_rec_input_clauses caller.  Handle cancellation.
5248         (lower_depend_clauses): New function.
5249         (lower_omp_taskreg): Lower depend clauses.  Adjust
5250         lower_rec_input_clauses caller.  Add clobber after the call.  Handle
5251         cancellation.
5252         (lower_omp_target, lower_omp_teams): New functions.
5253         (lower_omp_1): Handle cancellation.  Handle GIMPLE_OMP_TASKGROUP,
5254         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GOMP_barrier, GOMP_cancel
5255         and GOMP_cancellation_point calls.
5256         (lower_omp): Fold stmts inside of target region.
5257         (diagnose_sb_1, diagnose_sb_2): Handle GIMPLE_OMP_TASKGROUP,
5258         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
5259         * builtin-types.def (DEF_FUNCTION_TYPE_8): Document.
5260         (BT_FN_VOID_OMPFN_PTR_UINT,
5261         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG,
5262         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG,
5263         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT): Remove.
5264         (BT_FN_VOID_OMPFN_PTR_UINT_UINT_UINT,
5265         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_UINT,
5266         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG_UINT,
5267         BT_FN_BOOL_INT, BT_FN_BOOL_INT_BOOL, BT_FN_VOID_UINT_UINT,
5268         BT_FN_VOID_INT_PTR_SIZE_PTR_PTR_PTR,
5269         BT_FN_VOID_INT_OMPFN_PTR_SIZE_PTR_PTR_PTR,
5270         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR): New.
5271         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1,
5272         call_may_clobber_ref_p_1): Handle BUILT_IN_GOMP_BARRIER_CANCEL,
5273         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_LOOP_END_CANCEL,
5274         BUILT_IN_GOMP_SECTIONS_END_CANCEL.  Don't handle
5275         BUILT_IN_GOMP_PARALLEL_END.
5276         * gimple-low.c (lower_stmt): Handle GIMPLE_OMP_TASKGROUP,
5277         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
5278         * gimple-pretty-print.c (dump_gimple_omp_for): Handle
5279         GF_OMP_FOR_KIND_DISTRIBUTE.
5280         (dump_gimple_omp_target, dump_gimple_omp_teams): New functions.
5281         (dump_gimple_omp_block): Handle GIMPLE_OMP_TASKGROUP.
5282         (dump_gimple_omp_return): Print lhs if it has any.
5283         (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Handle
5284         gimple_omp_atomic_seq_cst_p.
5285         (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET
5286         and GIMPLE_OMP_TEAMS.
5287         * langhooks.c (lhd_omp_mappable_type): New function.
5288         * tree-vectorizer.c (struct simd_array_to_simduid): Fix up comment.
5289         * langhooks.h (struct lang_hooks_for_types): Add omp_mappable_type
5290         hook.
5291         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP,
5292         GOVD_ALIGNED and GOVD_MAP_TO_ONLY.
5293         (enum omp_region_type): Add ORT_TEAMS, ORT_TARGET_DATA and ORT_TARGET.
5294         (struct gimplify_omp_ctx): Add combined_loop field.
5295         (gimplify_call_expr, gimplify_modify_expr): Don't call fold_stmt
5296         on stmts inside of target region.
5297         (is_gimple_stmt): Return true for OMP_DISTRIBUTE and OMP_TASKGROUP.
5298         (omp_firstprivatize_variable): Handle GOVD_MAP, GOVD_ALIGNED,
5299         ORT_TARGET and ORT_TARGET_DATA.
5300         (omp_add_variable): Avoid checks on readding var for GOVD_ALIGNED.
5301         Handle GOVD_MAP.
5302         (omp_notice_threadprivate_variable): Complain about threadprivate
5303         variables in target region.
5304         (omp_notice_variable): Complain about vars with non-mappable type
5305         in target region.  Handle ORT_TEAMS, ORT_TARGET and ORT_TARGET_DATA.
5306         (omp_check_private): Ignore ORT_TARGET* regions.
5307         (gimplify_scan_omp_clauses, gimplify_adjust_omp_clauses_1,
5308         gimplify_adjust_omp_clauses): Handle new OpenMP 4.0 clauses.
5309         (find_combined_omp_for): New function.
5310         (gimplify_omp_for): Handle gimplification of combined loops.
5311         (gimplify_omp_workshare): Gimplify also OMP_TARGET, OMP_TARGET_DATA,
5312         OMP_TEAMS.
5313         (gimplify_omp_target_update): New function.
5314         (gimplify_omp_atomic): Handle OMP_ATOMIC_SEQ_CST.
5315         (gimplify_expr): Handle OMP_DISTRIBUTE, OMP_TARGET, OMP_TARGET_DATA,
5316         OMP_TARGET_UPDATE, OMP_TEAMS, OMP_TASKGROUP.
5317         (gimplify_body): If fndecl has "omp declare target" attribute, add
5318         implicit ORT_TARGET context around it.
5319         * tree.def (OMP_DISTRIBUTE, OMP_TEAMS, OMP_TARGET_DATA, OMP_TARGET,
5320         OMP_TASKGROUP, OMP_TARGET_UPDATE): New tree codes.
5321         * tree-nested.c (convert_nonlocal_reference_stmt,
5322         convert_local_reference_stmt, convert_gimple_call): Handle
5323         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
5324         * omp-builtins.def (BUILT_IN_GOMP_TASK): Use
5325         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR
5326         instead of BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT.
5327         (BUILT_IN_GOMP_TARGET, BUILT_IN_GOMP_TARGET_DATA,
5328         BUILT_IN_GOMP_TARGET_END_DATA, BUILT_IN_GOMP_TARGET_UPDATE,
5329         BUILT_IN_GOMP_TEAMS, BUILT_IN_BARRIER_CANCEL,
5330         BUILT_IN_GOMP_LOOP_END_CANCEL,
5331         BUILT_IN_GOMP_SECTIONS_END_CANCEL, BUILT_IN_OMP_GET_TEAM_NUM,
5332         BUILT_IN_OMP_GET_NUM_TEAMS, BUILT_IN_GOMP_TASKGROUP_START,
5333         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_PARALLEL_LOOP_STATIC,
5334         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC,
5335         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED,
5336         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME, BUILT_IN_GOMP_PARALLEL,
5337         BUILT_IN_GOMP_PARALLEL_SECTIONS, BUILT_IN_GOMP_CANCEL,
5338         BUILT_IN_GOMP_CANCELLATION_POINT): New built-ins.
5339         (BUILT_IN_GOMP_PARALLEL_LOOP_STATIC_START,
5340         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC_START,
5341         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED_START,
5342         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME_START,
5343         BUILT_IN_GOMP_PARALLEL_START, BUILT_IN_GOMP_PARALLEL_END,
5344         BUILT_IN_GOMP_PARALLEL_SECTIONS_START): Remove.
5345         * tree-inline.c (remap_gimple_stmt, estimate_num_insns):
5346         Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
5347         * gimple.c (gimple_build_omp_taskgroup, gimple_build_omp_target,
5348         gimple_build_omp_teams): New functions.
5349         (walk_gimple_op): Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and
5350         GIMPLE_OMP_TASKGROUP.  Walk optional lhs on GIMPLE_OMP_RETURN.
5351         (walk_gimple_stmt, gimple_copy): Handle GIMPLE_OMP_TARGET,
5352         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
5353         * gimple.h (enum gf_mask): GF_OMP_FOR_KIND_DISTRIBUTE,
5354         GF_OMP_FOR_COMBINED, GF_OMP_FOR_COMBINED_INTO,
5355         GF_OMP_TARGET_KIND_MASK, GF_OMP_TARGET_KIND_REGION,
5356         GF_OMP_TARGET_KIND_DATA, GF_OMP_TARGET_KIND_UPDATE,
5357         GF_OMP_ATOMIC_SEQ_CST): New.
5358         (gimple_build_omp_taskgroup, gimple_build_omp_target,
5359         gimple_build_omp_teams): New prototypes.
5360         (gimple_has_substatements): Handle GIMPLE_OMP_TARGET,
5361         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
5362         (gimple_omp_subcode): Use GIMPLE_OMP_TEAMS instead of
5363         GIMPLE_OMP_SINGLE as end of range.
5364         (gimple_omp_return_set_lhs, gimple_omp_return_lhs,
5365         gimple_omp_return_lhs_ptr, gimple_omp_atomic_seq_cst_p,
5366         gimple_omp_atomic_set_seq_cst, gimple_omp_for_combined_p,
5367         gimple_omp_for_set_combined_p, gimple_omp_for_combined_into_p,
5368         gimple_omp_for_set_combined_into_p, gimple_omp_target_clauses,
5369         gimple_omp_target_clauses_ptr, gimple_omp_target_set_clauses,
5370         gimple_omp_target_kind, gimple_omp_target_set_kind,
5371         gimple_omp_target_child_fn, gimple_omp_target_child_fn_ptr,
5372         gimple_omp_target_set_child_fn, gimple_omp_target_data_arg,
5373         gimple_omp_target_data_arg_ptr, gimple_omp_target_set_data_arg,
5374         gimple_omp_teams_clauses, gimple_omp_teams_clauses_ptr,
5375         gimple_omp_teams_set_clauses): New inlines.
5376         (CASE_GIMPLE_OMP): Add GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS
5377         and GIMPLE_OMP_TASKGROUP.
5378         * tree-core.h (enum omp_clause_code): Add new OpenMP 4.0 clause codes.
5379         (enum omp_clause_depend_kind, enum omp_clause_map_kind,
5380         enum omp_clause_proc_bind_kind): New.
5381         (union omp_clause_subcode): Add depend_kind, map_kind and
5382         proc_bind_kind fields.
5383         * tree-cfg.c (make_edges): Handle GIMPLE_OMP_TARGET,
5384         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
5385         * langhooks-def.h (lhd_omp_mappable_type): New prototype.
5386         (LANG_HOOKS_OMP_MAPPABLE_TYPE): Define.
5387         (LANG_HOOKS_FOR_TYPES_INITIALIZER): Add it.
5389 2013-10-10  Teresa Johnson  <tejohnson@google.com>
5391         * predict.c (tree_estimate_probability): Add new parameter
5392         for estimate_bb_frequencies.
5393         (estimate_bb_frequencies): Add new parameter to force estimation.
5394         (rebuild_frequencies): When max frequency in function is small,
5395         recompute counts from frequencies.
5396         * predict.h (estimate_bb_frequencies): New parameter.
5398 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
5400         * ipa-inline.c (ipa_inline): Fix leak of "order" when
5401         optimizations are disabled.
5403 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
5405         * coverage.c (coverage_finish): Fix leak of da_file_name.
5407 2013-10-10  Jan Hubicka  <jh@suse.cz>
5409         * config/i386/x86-tune.def: Enable X86_TUNE_SSE_TYPELESS_STORES
5410         for generic, enable X86_TUNE_SSE_LOAD0_BY_PXOR for Bulldozer,
5411         Bobcat and generic.
5413 2013-10-10  Jakub Jelinek  <jakub@redhat.com>
5415         PR middle-end/58670
5416         * stmt.c (expand_asm_operands): Add FALLTHRU_BB argument,
5417         if any labels are in FALLTHRU_BB, use a special label emitted
5418         immediately after the asm goto insn rather than label_rtx
5419         of the LABEL_DECL.
5420         (expand_asm_stmt): Adjust caller.
5421         * cfgrtl.c (commit_one_edge_insertion): Force splitting of
5422         edge if the last insn in predecessor is a jump with single successor,
5423         but it isn't simplejump_p.
5425 2013-10-10  Richard Biener  <rguenther@suse.de>
5427         PR tree-optimization/58656
5428         * tree-ssa-pre.c (phi_translate): Do not cache failed translations.
5430 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
5432         * gimplify.c: Include expr.h and tm_p.h for targets with special
5433         va-arg padding requirements.
5435 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
5437         * tree-flow.h: Move some prototypes to gimple.h.
5438         (gimple_fold_indirect_ref): Move prototype to gimple-fold.h.
5439         * gimple.h: Relocate some prototypes from tree-flow.h
5440         * builtins.c (std_gimplify_va_arg_expr, build_va_arg_indirect_ref):
5441         Move to gimplify.c.
5442         * gimplify.c (gimple_fold_indirect_ref): Move to gimple-fold.c.
5443         (build_va_arg_indirect_ref): Relocate and make static.
5444         (std_gimplify_va_arg_expr): Relocate here.
5445         * gimple-fold.c (gimple_fold_indirect_ref): Relocate here.
5446         * gimple-fold.h (gimple_fold_indirect_ref): Add prototype.
5448 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5450         * doc/md.texi: Document the mnemonic attribute.
5452 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5454         PR target/57377
5455         * gensupport.c (gen_mnemonic_attr): Handle (set (attr x) y) and
5456         (set_attr_alternative x ...) when searching for user defined
5457         mnemonic attribute.
5459 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
5461         * config/aplha/alpha.c: Add gimple-ssa.h to include list.
5463 2013-10-09  Easwaran Raman <eraman@google.com>
5465         * params.def (PARAM_MIN_SIZE_FOR_STACK_SHARING): New param...
5466         * cfgexpand.c (defer_stack_allocation): ...use here
5467         * doc/invoke.texi: Add documentation for min-size-for-stack-sharing.
5469 2013-10-09  Zhenqiang Chen  <zhenqiang.chen@arm.com>
5471         * tree-ssa-phiopts.c (rhs_is_fed_for_value_replacement): New function.
5472         (operand_equal_for_value_replacement): New function, extracted from
5473         value_replacement and enhanced to catch more cases.
5474         (value_replacement): Use operand_equal_for_value_replacement.
5476 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
5478         * loop-doloop.c (doloop_modify, doloop_optimize): Use
5479         get_max_loop_iterations.
5481 2013-10-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5483         * config/arm/aarch-common.c (arm_early_load_addr_dep):
5484         Place comment above function.
5486 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
5488         * tree-flow.h: Remove all remaining prototypes, enums and structs that
5489         are not related to tree-cfg.c.
5490         * tree-ssa-address.h: New file.  Relocate prototypes.
5491         * tree-ssa-address.c: (struct mem_address): Relocate from tree-flow.h.
5492         (addr_for_mem_ref): New.  Combine call to get_address_description and
5493         return addr_for_mem_ref.
5494         * expr.c (expand_expr_real_1): Use new addr_for_mem_ref routine.
5495         * tree-ssa-live.h: Adjust prototypes.
5496         * passes.c: Include tree-ssa-live.h.
5497         * gimple-pretty-print.h (gimple_dump_bb): Add prototype.
5498         * graphite.c (graphite_transform_loops): Make static.
5499         (graphite_transforms, gate_graphite_transforms, pass_data_graphite,
5500         make_pass_graphite, pass_data_graphite_transforms,
5501         make_pass_graphite_transforms): Relocate here from tree-ssa-loop.c.
5502         * ipa-pure-const.c (warn_function_noreturn): Make static.
5503         (execute_warn_function_noreturn, gate_warn_function_noreturn,
5504         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
5505         Relocate from tree-cfg.c
5506         * tree-cfg.c (tree_node_can_be_shared, gimple_empty_block_p): Make
5507         static.
5508         (execute_warn_function_noreturn, gate_warn_function_noreturn,
5509         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
5510         Move to ipa-pure-const.c.
5511         (execute_fixup_cfg, class pass_fixup_cfg, make_pass_fixup_cfg):
5512         Relocate from tree-optimize.c.
5513         * tree-optimize.c (execute_fixup_cfg, class pass_fixup_cfg,
5514         make_pass_fixup_cfg): Move to tree-cfg.c.
5515         * tree-chrec.h: (enum ev_direction): Relocate here from tree-flow.h.
5516         Relocate some prototypes.
5517         * tree-data-ref.h (tree_check_data_deps) Add prototype.
5518         * tree-dump.c (dump_function_to_file): Remove prototype.
5519         Add tree-flow.h to the include file.
5520         * tree-dump.h: Remove prototype.
5521         * tree-parloops.h: New File.  Add prototypes.
5522         * tree-parloops.c (gate_tree_parallelize_loops, tree_parallelize_loops,
5523         pass_data_parallelize_loops,  make_pass_parallelize_loops): Relocate
5524         from tree-ssa-loop.c.
5525         * tree-predcom.c (run_tree_predictive_commoning,
5526         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom):
5527         Relocate here from tree-ssa-loop.c.
5528         * tree-ssa-dom.c (tree_ssa_dominator_optimize) Don't call
5529         ssa_name_values.release ().
5530         * tree-ssa-threadedge.h: New File.  Relocate prototypes here.
5531         (ssa_name_values): Relocate from tree-flow.h.
5532         * tree-ssa.h: Include tree-ssa-threadedge.h and tree-ssa-address.h.
5533         * tree-ssa-loop.c (run_tree_predictive_commoning,
5534         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom,
5535         graphite_transforms, gate_graphite_transforms, pass_data_graphite,
5536         make_pass_graphite, pass_data_graphite_transforms,
5537         make_pass_graphite_transforms, gate_tree_parallelize_loops,
5538         tree_parallelize_loops, pass_data_parallelize_loops,
5539         make_pass_parallelize_loops): Move to other files.
5540         * tree-vectorizer.h (lpeel_tree_duplicate_loop_to_edge_cfg): Prototype
5541         moved here.
5542         * tree.h: Remove prototypes from tree-address.c.
5544 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
5546         * tree-flow.h (tm_restart_node, gimple_df): Move to gimple-ssa.h.
5547         (struct int_tree_map): Move to tree-hasher.h
5548         (SCALE, LABEL, PERCENT): Move to gimple.h
5549         * tree-flow-inline.h: Delete.  Move functions to other files.
5550         (unmodifiable_var_p, ref_contains_array_ref): Unused, so delete.
5551         * gimple-ssa.h (tm_restart_node, gimple_df): Relocate from tree-flow.h.
5552         (gimple_in_ssa_p, gimple_vop): Relocate from tree-flow-inline.h
5553         * gimple.h (imple_stmt_max_uid, set_gimple_stmt_max_uid,
5554         inc_gimple_stmt_max_uid, get_lineno): Relocate from tree-flow-inline.h.
5555         (SCALE, LABEL, PERCENT): Relocate from tree-flow.h
5556         * tree-hasher.h: Don't include tree-flow.h.
5557         (struct int_tree_map): Relocate from tree-flow.h.
5558         * tree-sra.c (contains_view_convert_expr_p): Relocate from
5559         tree-flow-inline.h and make static.
5560         * tree-ssa-alias.h (ranges_overlap_p): Relocate from
5561         tree-flow-inline.h.
5562         * tree-ssa-operands.c (gimple_ssa_operands): Relocate from
5563         tree-flow-inline.h and make static.
5564         * tree.h (is_global_var, may_be_aliased): Relocate from
5565         tree-flow-inline.h.
5566         * Makefile.in (GTFILES): Remove tree-flow.h and add gimple-ssa.h.
5567         * value-prof.c: No longer include tree-flow-inline.h.
5568         * tree-switch-conversion.c: No longer include tree-flow-inline.h.
5570 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
5572         * tree-flow.h: Move some protoypes.  Include new tree-ssa-loop.h.
5573         (struct affine_iv, struct tree_niter_desc): Move to tree-ssa-loop.h.
5574         (enum move_pos): Move to tree-ssa-loop-im.h
5575         * cfgloop.h: Move some prototypes.
5576         (gcov_type_to_double_int): relocate from tree-ssa-loop.niter.c.
5577         * tree-flow-inline.h (loop_containing_stmt): Move to tree-ssa-loop.h.
5578         * tree-ssa-loop.h: New File.  Include other tree-ssa-loop-*.h files.
5579         (struct affine_iv, struct tree_niter_desc): Relocate from tree-flow.h.
5580         (loop_containing_stmt): Relocate from tree-flow-inline.h.
5581         * tree-ssa-loop-ch.c: (do_while_loop_p): Make static.
5582         * tree-ssa-loop-im.c (for_each_index): Move to tree-ssa-loop.c.
5583         (enum move_pos): Relocate here.
5584         (lsm_tmp_name_add, gen_lsm_tmp_name, get_lsm_tmp_name): Move to
5585         tree-ssa-loop.c.
5586         (execute_sm_if_changed_flag_set): Change get_lsm_tmp_name call.
5587         (tree_ssa_loop_im, gate_tree_ssa_loop_im, pass_data_lim,
5588         make_pass_lim): Relocate here from tree-ssa-loop.c.
5589         * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Move to
5590         tree-ssa-loop.c.
5591         (loop_edge_to_cancel, unloop_loops): Make static.
5592         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
5593         make_pass_iv_canon): Relocate from tree-ssa-loop.c.
5594         (tree_complete_unroll, gate_tree_complete_unroll,
5595         pass_data_complete_unroll, make_pass_complete_unroll): Relocate here.
5596         (tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
5597         pass_data_complete_unrolli, make_pass_complete_unrolli): Relocate here.
5598         * tree-ssa-loop-ivopts.c: Remove local prototypes.
5599         (stmt_invariant_in_loop_p): Remove unused function.
5600         * tree-ssa-loop-ivopts.h: New file.  Add prototypes.
5601         * tree-ssa-loop-manip.h: New file.  Add prototypes.
5602         * tree-ssa-loop-niter.c (record_niter_bound): Move to cfgloop.c.
5603         (gcov_type_to_double_int): Move to cfgloop.h.
5604         (double_int_cmp, bound_index,
5605         estimate_numbers_of_iterations_loop): Make static.
5606         (estimated_loop_iterations): Factor out get_estimated_loop_iterations.
5607         (max_loop_iterations): Factor out get_max_loop_iterations.
5608         (estimated_loop_iterations_int, max_stmt_executions_int): Move to
5609         cfgloop.c.
5610         * tree-ssa-loop-niter.h: New file.  Add prototypes.
5611         * tree-ssa-loop-prefetch.c (tree_ssa_loop_prefetch,
5612         gate_tree_ssa_loop_prefetch, pass_data_loop_prefetch,
5613         make_pass_loop_prefetch): Relocate from tree-ssa-loop.c.
5614         * tree-ssa-loop-unswitch.c (tree_ssa_loop_unswitch,
5615         gate_tree_ssa_loop_unswitch, pass_data_tree_unswitch,
5616         make_pass_tree_unswitch): Relocate from tree-ssa-loop.c.
5617         * tree-ssa-loop.c (tree_ssa_loop_im, gate_tree_ssa_loop_im,
5618         pass_data_lim, make_pass_lim): Move to tree-ssa-loop-im.c.
5619         (tree_ssa_loop_unswitch, gate_tree_ssa_loop_unswitch,
5620         pass_data_tree_unswitch, make_pass_tree_unswitch): Move.
5621         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
5622         make_pass_iv_canon, tree_complete_unroll, gate_tree_complete_unroll,
5623         pass_data_complete_unroll, make_pass_complete_unroll,
5624         tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
5625         pass_data_complete_unrolli, make_pass_complete_unrolli): Move to
5626         tree-ssa-loop-ivcanon.c.
5627         (tree_ssa_loop_prefetch, gate_tree_ssa_loop_prefetch,
5628         pass_data_loop_prefetch, make_pass_loop_prefetch): Move to
5629         tree-ssa-loop-prefetch.c.
5630         (for_each_index, lsm_tmp_name_add, gen_lsm_tmp_name): Relocate from
5631         tree-ssa-loop-im.c.
5632         (get_lsm_tmp_name): Relocate and add suffix parameter.
5633         (tree_num_loop_insns): Relocate from tree-ssa-ivcanon.c.
5634         * tree-scalar-evolution.h (simple_iv): Don't use affive_iv typedef.
5635         * cfgloop.c (record_niter_bound, estimated_loop_iterations_int,
5636         max_stmt_executions_int): Move from tree-ssa-loop-niter.c.
5637         (get_estimated_loop_iterations): Factor out accessor from
5638         estimated_loop_iterations in tree-ssa-loop-niter.c.
5639         (get_max_loop_iterations): Factor out accessor from
5640         _max_loop_iterations in tree-ssa-niter.c.
5641         * loop-unroll.c (decide_unroll_constant_iterations,
5642         decide_unroll_runtime_iterations, decide_peel_simple,
5643         decide_unroll_stupid): Use new get_* accessors.
5645 2013-10-09  Marc Glisse  <marc.glisse@inria.fr>
5647         PR tree-optimization/20318
5648         * doc/extend.texi (returns_nonnull): New function attribute.
5649         * fold-const.c (tree_expr_nonzero_warnv_p): Look for returns_nonnull
5650         attribute.
5651         * tree-vrp.c (gimple_stmt_nonzero_warnv_p): Likewise.
5652         (stmt_interesting_for_vrp): Accept all GIMPLE_CALL.
5654 2013-10-09  Eric Botcazou  <ebotcazou@adacore.com>
5656         PR middle-end/58570
5657         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): Return
5658         false if both components are bitfields.
5660 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
5662         * config/aarch64/arm_neon.h (vclz_s8, vclz_s16, vclz_s32)
5663         (vclzq_s8, vclzq_s16, vclzq_s32, vclz_u8, vclz_u16, vclz_u32)
5664         (vclzq_u8, vclzq_u16, vclzq_u32): Replace ASM with C.
5665         * config/aarch64/aarch64.h
5666         (CLZ_DEFINED_VALUE_AT_ZERO): Macro fixed for clz.
5667         * config/aarch64/aarch64-simd-builtins.def
5668         (VAR1 (UNOP, clz, 0, v4si)): Replaced with iterator.
5670 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
5672         * config/aarch64/arm_neon.h (vadd_f64, vsub_f64): Implementation added.
5674 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
5676         * config/aarch64/arm_neon.h (vdiv_f64): Added.
5678 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
5680         * config/aarch64/arm_neon.h (vneg_f32): Asm replaced with C.
5681         (vneg_f64): New intrinsic.
5682         (vneg_s8): Asm replaced with C.
5683         (vneg_s16): Likewise.
5684         (vneg_s32): Likewise.
5685         (vneg_s64): New intrinsic.
5686         (vnegq_f32): Asm replaced with C.
5687         (vnegq_f64): Likewise.
5688         (vnegq_s8): Likewise.
5689         (vnegq_s16): Likewise.
5690         (vnegq_s32): Likewise.
5691         (vnegq_s64): Likewise.
5693 2013-10-09  Renlin Li  <Renlin.Li@arm.com>
5695         * config/arm/arm.c (arm_output_mi_thunk): Use plus_constant.
5697 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5699         * config/s390/s390.c (s390_register_info_stdarg_fpr): Remove
5700         packed stack special handling.
5701         (s390_frame_info, s390_emit_prologue, s390_emit_epilogue): Switch
5702         back to fixed stack slots for FPRs saved due to stdarg.
5704 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5706         * config/s390/s390.c (s390_frame_info): Restructure function.
5708 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5710         * config/s390/s390.c (struct s390_frame_layout): New field
5711         gpr_save_slots.
5712         (cfun_save_arg_fprs_p, cfun_gpr_save_slot): New macros.
5713         (s390_reg_clobbered_rtx, s390_regs_ever_clobbered): Change type of
5714         regs_ever_clobbered to char*.
5715         (s390_regs_ever_clobbered): Check crtl->saves_all_registers instead
5716         of cfun->has_nonlocal_label.  Ignore frame related restore INSNs.
5717         (s390_register_info): Enable FPR save slots.  Move/Copy some
5718         functionality into ...
5719         (s390_register_info_gprtofpr, s390_register_info_stdarg_fpr)
5720         (s390_register_info_stdarg_gpr, s390_optimize_register_info): New
5721         function.
5722         (s390_frame_info): Do gpr slot allocation here now.  stdarg does
5723         not imply a stack frame.
5724         (s390_init_frame_layout): Remove variable clobbered_regs.
5725         (s390_update_register_info): Remove function.
5726         (s390_hard_regno_rename_ok): Call-saved regs without a save slot
5727         cannot be used for register renaming.
5728         (s390_hard_regno_scratch_ok): New function.
5729         (TARGET_HARD_REGNO_SCRATCH_OK): Define target hook.
5730         (s390_initial_elimination_offset): Change offset calculation of
5731         the return address pointer.
5732         (save_gprs): Deal with only r6 being saved from the call-saved regs.
5733         (restore_gprs): Set frame related flag.
5734         (s390_save_gprs_to_fprs, s390_restore_gprs_from_fprs): New functions.
5735         (s390_emit_prologue): Call s390_register_info instead of
5736         s390_update_frame_layout.  Call s390_save_gprs_to_fprs.
5737         (s390_emit_epilogue): Call s390_restore_gprs_from_fprs.
5738         (s390_optimize_prologue): Call s390_optimize_register_info.
5739         Try to remove also FPR slot save/restore INSNs.  Remove frame
5740         related flags from restore INSNs.
5742 2013-10-08  DJ Delorie  <dj@redhat.com>
5744         * config/rl78/rl78-expand.md (movqi): use operands[] not operandN.
5745         (movhi): Likewise.
5747         * config/rl78/rl78.c (rl78_print_operand_1): Change %c to %C to
5748         avoid conflict with the MI use of %c.
5749         * config/rl78/rl78-real.md: change %c to %C throughout.
5750         * config/rl78/rl78-virt.md: Likewise.
5752 2013-10-08  Jan Hubicka  <jh@suse.cz>
5754         * config/i386/i386.c (ix86_option_override_internal): Switch
5755         to SSE math for -ffast-math when target ISA supports SSE2.
5757 2013-10-08  Andrew MacLeod  <amacleod@redhat.com>
5759         * tree-flow.h: Remove some prototypes.
5760         * tree.h: Remove some protypes, add a couple.
5761         * tree.c (using_eh_for_cleanups_flag, using_eh_for_cleanups,
5762         using_eh_for_cleanups_p): Add interface routines for front ends.
5763         * tree-eh.h: New file.  Add protoptyes.
5764         * tree-eh.c (using_eh_for_cleanups_p, using_eh_for_cleanups): Delete.
5765         (add_stmt_to_eh_lp_fn): Make static.
5766         (lower_try_finally): Use new using_eh_for_cleanups_p.
5767         * emit-rtl.c: Include tree-eh.h.
5768         * gimple.h: Include tree-eh.h.
5770 2013-10-08  Marc Glisse  <marc.glisse@inria.fr>
5772         PR tree-optimization/58480
5773         * tree-vrp.c (infer_nonnull_range): New function.
5774         (infer_value_range): Call infer_nonnull_range.
5776 2013-10-08  Dehao Chen  <dehao@google.com>
5778         PR tree-optimization/58619
5779         * tree-inline.c (copy_phis_for_bb): Combine location data
5780         only if non-null.
5782 2013-10-08  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
5784         PR target/58423
5785         * config/arm/arm.c (arm_emit_ldrd_pop): Attach
5786         RTX_FRAME_RELATED_P on INSN.
5788 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5790         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const_le): New.
5791         (altivec_expand_vec_perm_const): Call it.
5793 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5795         * config/rs6000/vector.md (mov<mode>): Emit permuted move
5796         sequences for LE VSX loads and stores at expand time.
5797         * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_move): New
5798         prototype.
5799         * config/rs6000/rs6000.c (rs6000_const_vec): New.
5800         (rs6000_gen_le_vsx_permute): New.
5801         (rs6000_gen_le_vsx_load): New.
5802         (rs6000_gen_le_vsx_store): New.
5803         (rs6000_gen_le_vsx_move): New.
5804         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): New.
5805         (*vsx_le_perm_load_v4si): New.
5806         (*vsx_le_perm_load_v8hi): New.
5807         (*vsx_le_perm_load_v16qi): New.
5808         (*vsx_le_perm_store_v2di): New.
5809         (*vsx_le_perm_store_v4si): New.
5810         (*vsx_le_perm_store_v8hi): New.
5811         (*vsx_le_perm_store_v16qi): New.
5812         (*vsx_xxpermdi2_le_<mode>): New.
5813         (*vsx_xxpermdi4_le_<mode>): New.
5814         (*vsx_xxpermdi8_le_V8HI): New.
5815         (*vsx_xxpermdi16_le_V16QI): New.
5816         (*vsx_lxvd2x2_le_<mode>): New.
5817         (*vsx_lxvd2x4_le_<mode>): New.
5818         (*vsx_lxvd2x8_le_V8HI): New.
5819         (*vsx_lxvd2x16_le_V16QI): New.
5820         (*vsx_stxvd2x2_le_<mode>): New.
5821         (*vsx_stxvd2x4_le_<mode>): New.
5822         (*vsx_stxvd2x8_le_V8HI): New.
5823         (*vsx_stxvd2x16_le_V16QI): New.
5825 2013-10-07  Renlin Li  <Renlin.Li@arm.com>
5827         * config/arm/arm-cores.def (cortex-a53): Use cortex tuning.
5829 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5831         * config/s390/s390.c (s390_register_info): Make the call-saved FPR
5832         loop to work also for 31bit ABI.
5833         Save the stack pointer for frame_size > 0.
5835 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5837         * config/s390/s390.md ("tbegin", "tbegin_nofloat", "tbegin_retry")
5838         ("tbegin_retry_nofloat", "tend", "tabort", "tx_assist"): Remove
5839         constraint letters from expanders.
5840         ("tbegin_retry", "tbegin_retry_nofloat"): Change predicate of the
5841         retry count to general_operand.
5842         ("tabort"): Give operand 0 a mode.
5843         ("tabort_1"): Add mode and constraint letter for operand 0.
5844         * doc/extend.texi: Fix protoype of __builtin_non_tx_store.
5846 2013-10-04  Jeff Law  <law@redhat.com>
5848         * tree-ssa-threadedge.c: Fix some trailing whitespace problems.
5850         * tree-ssa-threadedge.c (thread_through_normal_block): Broken
5851         out of ...
5852         (thread_across_edge): Here.  Call it.
5854 2013-10-04  Cary Coutant  <ccoutant@google.com>
5856         * dwarf2out.c (dw_sra_loc_expr): Release addr_table entries when
5857         discarding a location list expression (or a piece of one).
5859 2013-10-03  Jan Hubicka  <jh@suse.cz>
5861         * config/i386/i386.c (ix86_issue_rate): Pentium4, Nocona has issue
5862         rate of 2.  Core2, Corei7 and Haswell has issue rate of 4.
5863         (ix86_adjust_cost): Remove Atom case; fix core2/corei7/Haswell case.
5865 2013-10-03  Jan Hubicka  <jh@suse.cz>
5867         * config/i386/i386.c (ix86_option_override_internal): Do not enable
5868         accumulate-outgoing-args when producing unwind info.
5870 2013-10-03  Wei Mi  <wmi@google.com>
5872         * lra-constraints.c (insert_move_for_subreg): New function
5873         extracted from simplify_operand_subreg.
5874         (simplify_operand_subreg): Add reload for paradoxical subreg.
5876 2013-10-03  Rong Xu  <xur@google.com>
5878         * ipa-inline-analysis.c (find_foldable_builtin_expect): Find
5879         the candidate of builtin_expect such that we should fix the
5880         size/time estimation.
5881         (estimate_function_body_sizes): Do the acutally size/time fix-up
5882         for builtin_expect.
5884 2013-10-03  Rong Xu  <xur@google.com>
5886         * predict.c (tree_predict_by_opcode): Get the probability
5887         for builtin_expect from param builtin_expect_probability.
5888         * params.def (BUILTIN_EXPECT_PROBABILITY): New parameter.
5889         * predict.def (PRED_BUILTIN_EXPECT_RELAXED): Fix comments.
5890         * doc/invoke.texi: Add documentation for builtin-expect-probability.
5892 2013-10-03  Marc Glisse  <marc.glisse@inria.fr>
5894         PR c++/19476
5895         * common.opt (fcheck-new): Moved from c.opt. Make it 'Common'.
5896         * calls.c (alloca_call_p): Use get_callee_fndecl.
5897         * fold-const.c (tree_expr_nonzero_warnv_p): Handle operator new.
5898         * tree-vrp.c (gimple_stmt_nonzero_warnv_p, stmt_interesting_for_vrp):
5899         Likewise.
5900         (vrp_visit_stmt): Remove duplicated code.
5902 2013-10-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
5904         * config/rs6000/rs6000-builtin.def (XSRDPIM): Use floatdf2,
5905         ceildf2, btruncdf2, instead of vsx_* name.
5907         * config/rs6000/vsx.md (vsx_add<mode>3): Change arithmetic
5908         iterators to only do V2DF and V4SF here.  Move the DF code to
5909         rs6000.md where it is combined with SF mode.  Replace <VSv> with
5910         just 'v' since only vector operations are handled with these insns
5911         after moving the DF support to rs6000.md.
5912         (vsx_sub<mode>3): Likewise.
5913         (vsx_mul<mode>3): Likewise.
5914         (vsx_div<mode>3): Likewise.
5915         (vsx_fre<mode>2): Likewise.
5916         (vsx_neg<mode>2): Likewise.
5917         (vsx_abs<mode>2): Likewise.
5918         (vsx_nabs<mode>2): Likewise.
5919         (vsx_smax<mode>3): Likewise.
5920         (vsx_smin<mode>3): Likewise.
5921         (vsx_sqrt<mode>2): Likewise.
5922         (vsx_rsqrte<mode>2): Likewise.
5923         (vsx_fms<mode>4): Likewise.
5924         (vsx_nfma<mode>4): Likewise.
5925         (vsx_copysign<mode>3): Likewise.
5926         (vsx_btrunc<mode>2): Likewise.
5927         (vsx_floor<mode>2): Likewise.
5928         (vsx_ceil<mode>2): Likewise.
5929         (vsx_smaxsf3): Delete scalar ops that were moved to rs6000.md.
5930         (vsx_sminsf3): Likewise.
5931         (vsx_fmadf4): Likewise.
5932         (vsx_fmsdf4): Likewise.
5933         (vsx_nfmadf4): Likewise.
5934         (vsx_nfmsdf4): Likewise.
5935         (vsx_cmpdf_internal1): Likewise.
5937         * config/rs6000/rs6000.h (TARGET_SF_SPE): Define macros to make it
5938         simpler to select whether a target has SPE or traditional floating
5939         point support in iterators.
5940         (TARGET_DF_SPE): Likewise.
5941         (TARGET_SF_FPR): Likewise.
5942         (TARGET_DF_FPR): Likewise.
5943         (TARGET_SF_INSN): Macros to say whether floating point support
5944         exists for a given operation for expanders.
5945         (TARGET_DF_INSN): Likewise.
5947         * config/rs6000/rs6000.c (Ftrad): New mode attributes to allow
5948         combining of SF/DF mode operations, using both traditional and VSX
5949         registers.
5950         (Fvsx): Likewise.
5951         (Ff): Likewise.
5952         (Fv): Likewise.
5953         (Fs): Likewise.
5954         (Ffre): Likewise.
5955         (FFRE): Likewise.
5956         (abs<mode>2): Combine SF/DF modes using traditional floating point
5957         instructions.  Add support for using the upper DF registers with
5958         VSX support, and SF registers with power8-vector support.  Update
5959         expanders for operations supported by both the SPE and traditional
5960         floating point units.
5961         (abs<mode>2_fpr): Likewise.
5962         (nabs<mode>2): Likewise.
5963         (nabs<mode>2_fpr): Likewise.
5964         (neg<mode>2): Likewise.
5965         (neg<mode>2_fpr): Likewise.
5966         (add<mode>3): Likewise.
5967         (add<mode>3_fpr): Likewise.
5968         (sub<mode>3): Likewise.
5969         (sub<mode>3_fpr): Likewise.
5970         (mul<mode>3): Likewise.
5971         (mul<mode>3_fpr): Likewise.
5972         (div<mode>3): Likewise.
5973         (div<mode>3_fpr): Likewise.
5974         (sqrt<mode>3): Likewise.
5975         (sqrt<mode>3_fpr): Likewise.
5976         (fre<Fs>): Likewise.
5977         (rsqrt<mode>2): Likewise.
5978         (cmp<mode>_fpr): Likewise.
5979         (smax<mode>3): Likewise.
5980         (smin<mode>3): Likewise.
5981         (smax<mode>3_vsx): Likewise.
5982         (smin<mode>3_vsx): Likewise.
5983         (negsf2): Delete SF operations that are merged with DF.
5984         (abssf2): Likewise.
5985         (addsf3): Likewise.
5986         (subsf3): Likewise.
5987         (mulsf3): Likewise.
5988         (divsf3): Likewise.
5989         (fres): Likewise.
5990         (fmasf4_fpr): Likewise.
5991         (fmssf4_fpr): Likewise.
5992         (nfmasf4_fpr): Likewise.
5993         (nfmssf4_fpr): Likewise.
5994         (sqrtsf2): Likewise.
5995         (rsqrtsf_internal1): Likewise.
5996         (smaxsf3): Likewise.
5997         (sminsf3): Likewise.
5998         (cmpsf_internal1): Likewise.
5999         (copysign<mode>3_fcpsgn): Add VSX/power8-vector support.
6000         (negdf2): Delete DF operations that are merged with SF.
6001         (absdf2): Likewise.
6002         (nabsdf2): Likewise.
6003         (adddf3): Likewise.
6004         (subdf3): Likewise.
6005         (muldf3): Likewise.
6006         (divdf3): Likewise.
6007         (fred): Likewise.
6008         (rsqrtdf_internal1): Likewise.
6009         (fmadf4_fpr): Likewise.
6010         (fmsdf4_fpr): Likewise.
6011         (nfmadf4_fpr): Likewise.
6012         (nfmsdf4_fpr): Likewise.
6013         (sqrtdf2): Likewise.
6014         (smaxdf3): Likewise.
6015         (smindf3): Likewise.
6016         (cmpdf_internal1): Likewise.
6017         (lrint<mode>di2): Use TARGET_<MODE>_FPR macro.
6018         (btrunc<mode>2): Delete separate expander, and combine with the
6019         insn and add VSX instruction support.  Use TARGET_<MODE>_FPR.
6020         (btrunc<mode>2_fpr): Likewise.
6021         (ceil<mode>2): Likewise.
6022         (ceil<mode>2_fpr): Likewise.
6023         (floor<mode>2): Likewise.
6024         (floor<mode>2_fpr): Likewise.
6025         (fma<mode>4_fpr): Combine SF and DF fused multiply/add support.
6026         Add support for using the upper registers with VSX and
6027         power8-vector.  Move insns to be closer to the define_expands. On
6028         VSX systems, prefer the traditional form of FMA over the VSX
6029         version, since the traditional form allows the target not to
6030         overlap with the inputs.
6031         (fms<mode>4_fpr): Likewise.
6032         (nfma<mode>4_fpr): Likewise.
6033         (nfms<mode>4_fpr): Likewise.
6035 2013-10-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6036             Richard Earnshaw  <richard.earnshaw@arm.com>
6038         * config/arm/aarch-common-protos.h (struct alu_cost_table): New.
6039         (struct mult_cost_table): Likewise.
6040         (struct mem_cost_table): Likewise.
6041         (struct fp_cost_table): Likewise.
6042         (struct vector_cost_table): Likewise.
6043         (cpu_cost_table): Likewise.
6044         * config/arm/arm.opt (mold-rts-costs): New option.
6045         (mnew-generic-costs): Likewise.
6046         * config/arm/arm.c (generic_extra_costs): New table.
6047         (cortexa15_extra_costs): Likewise.
6048         (arm_slowmul_tune): Use NULL as new costs.
6049         (arm_fastmul_tune): Likewise.
6050         (arm_strongarm_tune): Likewise.
6051         (arm_xscale_tune): Likewise.
6052         (arm_9e_tune): Likewise.
6053         (arm_v6t2_tune): Likewise.
6054         (arm_cortex_a5_tune): Likewise.
6055         (arm_cortex_a9_tune): Likewise.
6056         (arm_v6m_tune): Likewise.
6057         (arm_fa726te_tune): Likewise.
6058         (arm_cortex_a15_tune): Use cortex15_extra_costs.
6059         (arm_cortex_tune): Use generict_extra_costs.
6060         (shifter_op_p): New function.
6061         (arm_unspec_cost): Likewise.
6062         (LIBCALL_COST): Define.
6063         (arm_new_rtx_costs): New function.
6064         (arm_rtx_costs): Use arm_new_rtx_costs when core-specific
6065         table is available. Use old costs otherwise unless mnew-generic-costs
6066         is specified.
6067         * config/arm/arm-protos.h (tune_params): Add insn_extra_cost field.
6068         (cpu_cost_table): Declare.
6070 2013-10-03  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6072         PR target/58460
6073         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>)
6074         (*subs_mul_imm_<mode>)
6075         (*add_<shift>_<mode>, *add_<shift>_si_uxtw,*add_mul_imm_<mode>)
6076         (*sub_<shift>_<mode>)
6077         (*sub_<shift>_si_uxtw,*sub_mul_imm_<mode>, *sub_mul_imm_si_uxtw):
6078         Remove k constraint.
6080 2013-10-03  Ian Bolton  <ian.bolton@arm.com>
6082         * config/aarch64/aarch64.c (aarch64_secondary_reload): Remove legacy
6083         code.
6084         * config/aarch64/aarch64.md (reload_sp_immediate): Likewise.
6086 2013-10-02  Teresa Johnson  <tejohnson@google.com>
6088         * predict.c (probably_never_executed): New function.
6089         (probably_never_executed_bb_p): Invoke probably_never_executed.
6090         (probably_never_executed_edge_p): Ditto.
6091         * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
6092         Treat profile insanities conservatively.
6094 2013-10-02  John David Anglin  <danglin@gcc.gnu.org>
6096         * config.gcc (hppa*64*-*-linux*): Don't add pa/t-linux to tmake_file.
6098 2013-10-02  Vladimir Makarov  <vmakarov@redhat.com>
6100         * lra-constraints.c (process_alt_operand): Calculate scratch_p and
6101         use it.  Use smaller increase for scratch.  Don't increase reject
6102         for early clobber scratch.
6103         * lra-eliminations.c (eliminate_regs_in_insn): Remove all insns
6104         setting eliminated regs except setting fp from hfp.
6105         (lra_eliminate): Check lra_insn_recog_data on NULL.
6107 2013-10-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
6109         PR target/58587
6110         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Turn off
6111         setting -mvsx-timode by default until the underlying problem is fixed.
6112         (RS6000_CPU, power7 defaults): Likewise.
6114 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
6116         * config/x-linux (host-linux.o): Remove header dependencies.
6117         Use $(COMPILE) and $(POSTCOMPILE).
6118         * config/t-linux-android (linux-android.o): Ditto.
6120 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
6122         * Makefile.in (expmed.o-warn): Remove.
6124 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
6126         * graphite-scop-detection.c: Include tree-ssa-propagate,h.
6127         * graphite-sese-to-poly.c: Include tree-ssa-propagate.h.
6129 2013-10-02  Teresa Johnson  <tejohnson@google.com>
6131         * dojump.c (do_jump_1): Divide probability between
6132         both conditions of a TRUTH_ANDIF_EXPR/TRUTH_ORIF_EXPR.
6134 2013-10-02  Tom Tromey  <tromey@redhat.com>
6136         * Makefile.in (DRIVER_DEFINES): Use $(if), not $(and).
6138 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
6140         * tree-flow.h: Remove some prototypes.
6141         * tree-ssa-dce.c (mark_virtual_operand_for_renaming,
6142         mark_virtual_phi_result_for_renaming): Move to tree-into-ssa.c.
6143         * tree-into-ssa.c (mark_virtual_operand_for_renaming,
6144         mark_virtual_phi_result_for_renaming): Relocate here.
6145         * tree-into-ssa.h: Add prototypes.
6146         * tree-ssa-phiopt.c: (tree_ssa_phiopt_worker) Use
6147         single_pred_before_succ_order.
6148         (blocks_in_phiopt_order): Rename and move to cfganal.c.
6149         (nonfreeing_call_p) Move to gimple.c.
6150         * cfganal.c (single_pred_before_succ_order): Move and renamed from
6151         tree-ssa-phiopt.c.
6152         * basic-block.h (single_pred_before_succ_order): Add prototype.
6153         * gimple.c (nonfreeing_call_p): Relocate here.
6154         * gimple.h: Add prototype.
6155         * tree-ssa-ifcombine.c: Include tree-ssa-phiopt.h.
6156         * tree-ssa-dom.h: New file.  Relocate prototypes here.
6157         * tree-ssa.h: Include tree-ssa-dom.h.
6159 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
6161         * config/i386/x-i386 (driver-i386.o): Remove header dependencies.
6162         Use $(COMPILE) and $(POSTCOMPILE).
6164         * config/alpha/x-alpha (driver-alpha.o): Ditto.
6166 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
6168         * tree-flow.h: Remove some prototypes.
6169         * gimple-fold.h: Add prototypes from gimple.h and tree-flow.h.
6170         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
6171         * tree-ssa-copy.c (may_propagate*, propagate_value, replace_exp,
6172         propagate_tree_value*): Move from here to...
6173         * tree-ssa-propagate.c (may_propagate*, propagate_value, replace_exp,
6174         propagate_tree_value*): Relocate here.
6175         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
6176         * gimple.h: Include gimple-fold.h, move prototypes into gimple-fold.h.
6177         * gimple-fold.c: Remove gimple-fold.h from include list.
6178         * tree-vrp.c: Remove gimple-fold.h from include list.
6179         * tree-ssa-sccvn.c: Remove gimple-fold.h from include list.
6180         * tree-ssa-ccp.c: Remove gimple-fold.h from include list.
6181         * tree-scalar-evolution.c: Add tree-ssa-propagate.h to include list.
6182         * tree-ssa-pre.c: Add tree-ssa-propagate.h to include list.
6183         * sese.c: Add tree-ssa-propagate.h to include list.
6185 2013-10-02  Richard Biener  <rguenther@suse.de>
6187         * tree-loop-distribution.c: Include tree-vectorizer.h for
6188         find_loop_location.
6189         (enum partition_kind): Remove PKIND_REDUCTION.
6190         (struct partition_s): Remove has_writes member, add reduction_p member.
6191         (partition_alloc): Adjust.
6192         (partition_builtin_p): Likewise.
6193         (partition_has_writes): Remove.
6194         (partition_reduction_p): New function.
6195         (partition_merge_into): Likewise.
6196         (generate_code_for_partition): Commonize builtin partition
6197         handling tail.
6198         (rdg_cannot_recompute_vertex_p): Remove.
6199         (already_processed_vertex_p): Likewise.
6200         (rdg_flag_vertex): Do not set has_writes.
6201         (classify_partition): Adjust.
6202         (rdg_build_partitions): Do not set has_writes, treat all
6203         partitions as useful.
6204         (distribute_loop): Record number of library calls generated.  Adjust.
6205         (tree_loop_distribution): Report number of loops and library
6206         calls generated as opt-info.
6208 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
6210         * tree-flow.h: Include new .h files.  Move prototypes.
6211         * tree-cfgcleanup.h: New file.  Add prototypes from tree-flow.h.
6212         * tree-dfa.h: New File.  Add prototypes from tree-flow.h.
6213         (get_addr_base_and_unit_offset_1) Move from tree-flow-inline.h.
6214         * tree-pretty-print.h: Add prototypes from tree-flow.h.
6215         * tree-into-ssa.h: New File.  Add prototypes from tree-flow.h.
6216         ({debug|dump}*): Move debugging prototypes out of tree-into-ssa.c.
6217         * tree-into-ssa.c ({debug|dump}*): Move prototypes to header file.
6218         * tree.h (get_ref_base_and_extent): Move prototype out.
6219         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Move to
6220         tree-dfa.h.
6221         * gimple-low.h: New File.  Add prototypes from tree-flow.h.
6222         * gimple-low.c (try_catch_may_fallthru, block_may_fallthru): Move to...
6223         * tree.c (try_catch_may_fallthru, block_may_fallthru): Here.
6224         * tree-scalar-evolution.c: Include tree.h.
6225         * sese.c: Include tree.h.
6226         * dumpfile.c: Move gimple-pretty-print.h include after tree.h.
6227         * dwarf2out.c: Include tree-dfa.h.
6228         * tree-chrec.c: Include tree.h.
6229         * tree-data-ref.c: Include tree.h.
6231 2013-10-02  Yufeng Zhang  <yufeng.zhang@arm.com>
6233         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref):
6234         Fix whitespace.
6236 2013-10-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
6238         * config/t-sol2 (sol2-c.o): Remove header dependencies.
6239         Use $(COMPILE) and $(POSTCOMPILE).
6240         (sol2-cxx.o): Likewise.
6241         (sol2-stubs.o): Likewise.
6242         (sol2.o): Likewise.
6243         * config/x-solaris (host-solaris.o): Likewise.
6245         * config/sparc/t-sparc (sparc.o): Remove.
6246         (sparc-c.o): Remove header dependencies.
6247         Use $(COMPILE) and $(POSTCOMPILE).
6248         * config/sparc/x-sparc: Likewise.
6250 2013-10-02  Joern Rennecke  <joern.rennecke@embecosm.com>
6252         * config/arc/arc-opts.h: Add 2013 to Copyright years.
6253         * config/arc/arc700.md: Likewise.
6254         * config/arc/arc-modes.def: Likewise.
6255         * config/arc/arc-simd.h: Likewise.
6256         * config/arc/t-arc-uClibc: Likewise.
6257         * config/arc/t-arc-newlib: Likewise.
6259 2013-10-02  Renlin Li  <renlin.li@arm.com>
6261         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6262         plus_constant.
6263         (aarch64_expand_epilogue): Likewise.
6265 2013-10-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6266             Yufeng Zhang  <yufeng.zhang@arm.com>
6268         * gimple-ssa-strength-reduction.c (legal_cast_p_1): Forward
6269         declaration.
6270         (backtrace_base_for_ref): Call get_unwidened with 'base_in' if
6271         'base_in' represent a conversion and legal_cast_p_1 holds; set
6272         'base_in' with the returned value from get_unwidened.
6274 2013-10-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6276         * config/arm/arm.c (arm_legitimize_reload_address): Explain why
6277         plus_constant is not used.
6279 2013-10-01  Wei Mi  <wmi@google.com>
6281         * config/i386/x86-tune.def (DEF_TUNE): Remove m_CORE_ALL.
6282         * config/i386/i386.md: Add define_peephole2 to
6283         break partial reg stall for cvtss2sd/cvtsd2ss.
6285 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
6287         * config/arc/arc.c (pass_arc_ifcvt::clone):
6288         Update for ctxt_ -> m_ctxt change.
6290 2013-10-01  Jeff Law  <law@redhat.com>
6292         * tree-ssa-threadupdate.c (struct redirection_data): Delete
6293         outgoing_edge and intermediate_edge fields.  Instead store the path.
6294         (redirection_data::hash): Hash on the last edge's destination index.
6295         (redirection_data::equal): Check the entire thread path.
6296         (lookup_redirectio_data): Corresponding changes.
6297         (create_edge_and_update_destination_phis): Likewise.
6298         (thread_single_edge): Likewise.
6300 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
6301             Diego Novillo <dnovillo@google.com>
6303         * config/arc/simdext.md (UNSPEC_ARC_SIMD_VLD32WH): Delete.
6304         (UNSPEC_ARC_SIMD_VLD32WL): Likewise.
6305         (vld32wh_insn, vld32wl_insn): Delete commented-out old
6306         versions of these patterns.
6308         * doc/extend.texi (long_call/medium_call/short_call): Typo fix.
6309         (__builtin_arc_aligned): Likewise.
6311         * config/arc/arc.md: Expand adc_0 comment stating the intended
6312         purpose and why it isn't ready.
6313         Replace commented out call_value_via_label_mixed with a
6314         plain comment about bl_s.
6316         * config/arc/arc.c (stdio.h): Don't include directly.
6317         (arc_expand_epilogue): Remove [0]: Remove fp_restored_p.
6318         Remove if (1) condition.
6319         (arc_encode_section_info): Fix comment.
6321 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
6323         * config/arc/arc.c (arc_conditional_register_usage):
6324         Use ARC_FIRST_SIMD_VR_REG / ARC_LAST_SIMD_VR_REG.
6325         Also set reg_alloc_order for DMA config regs.
6327 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
6328             Jeremy Bennett  <jeremy.bennett@embecosm.com>
6330         * doc/install.texi (--with-cpu): Mention ARC.
6331         (arc-*-elf32): New paragraph.
6332         (arc-linux-uclibc): Likewise.
6333         * doc/md.texi (Machine Constraints): Add ARC part.
6334         * doc/invoke.texi: (menu): Add ARC Options.
6335         (Machine Dependent Options) <ARC Options>: Add synopsis.
6336         (node ARC Options): Add.
6337         * doc/extend.texi (long_call / short_call attribute): Add ARC.
6338         (ARC Built-in Functions): New section defining
6339         generic ARC built-in functions.
6340         (ARC SIMD Built-in Functions): New section defining SIMD specific
6341         built-in functions.
6342         (Declaring Attributes of Functions): Extended
6343         description of short_call and long_call attributes for ARC and
6344         added index entries.
6346 2013-10-01  Saurabh Verma  <saurabh.verma@codito.com>
6347             Ramana Radhakrishnan  <ramana.radhakrishnan@codito.com>
6348             Joern Rennecke  <joern.rennecke@embecosm.com>
6349             Muhammad Khurram Riaz  <khurram.riaz@arc.com>
6350             Brendan Kehoe  <brendan@zen.org>
6351             Michael Eager  <eager@eagercon.com>
6352             Simon Cook  <simon.cook@embecosm.com>
6353             Jeremy Bennett  <jeremy.bennett@embecosm.com>
6355         * config/arc, common/config/arc: New directories.
6357 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
6358             Brendan Kehoe  <brendan@zen.org>
6359             Simon Cook  <simon.cook@embecosm.com>
6361         * config.gcc (arc*-*-elf*, arc*-*-linux-uclibc*): New configurations.
6363 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
6365         * tree-ssa-live.h (coalesce_ssa_name): Move Prototype to...
6366         * tree-ssa-coalesce.h: New. Move prototype to here.
6367         * tree-outof-ssa.h: Include tree-ssa-coalesce.h.
6368         * tree-ssa-coalesce.c: Include tree-outof-ssa.h.
6369         (gimple_can_coalesce_p): Move to...
6370         * gimple.c (gimple_can_coalesce_p): Here.
6372 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
6374         * tree-into-ssa.c (enum need_phi_state): Relocate from tree-flow.h.
6375         (dump_decl_set): Move to gimple.c.
6376         * gimple.h: Don't include tree-ssa-operands.h.
6377         (dump_decl_set): Add prototype.
6378         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
6379         Move to gimple-ssa.h.
6380         (phi_ssa_name_p, phi_nodes, phi_nodes_ptr, gimple_phi_arg_def,
6381         gimple_phi_arg_def_ptr, gimple_phi_arg_edge, gimple_phi_arg_location,
6382         gimple_phi_arg_location_from_edge, gimple_phi_arg_set_location,
6383         gimple_phi_arg_has_location): Relocate from tree-flow-inline.h
6384         * gimple.c (walk_stmt_load_store_ops): Use gimple_phi_arg_def rather
6385         than PHI_ARG_DEF.
6386         (dump_decl_set): Relocate here.
6387         * gimple-ssa.h: New file.
6388         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
6389         Relocate from gimple.h.
6390         * tree-cfg.c (has_zero_uses_1, single_imm_use_1): Move to...
6391         * tree-ssa-operands.c (swap_ssa_operands): Rename from
6392         swap_tree_operands and remove non-ssa path.
6393         (has_zero_uses_1, single_imm_use_1): Relocate from tree-cfg.c.
6394         * tree-ssa-reassoc.c (linearize_expr_tree, repropagate_negates): Use
6395         swap_ssa_operands.
6396         * tree-vect-loop.c (destroy_loop_vec_info, vect_is_slp_reduction,
6397         vect_is_simple_reduction_1): Use swap_ssa_operands.
6398         * tree-flow.h: Move various prototypes to tree-phinodes.h.
6399         (enum need_phi_state): Move to tree-into-ssa.c.
6400         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
6401         BREAK_FROM_IMM_USE_STMT): Move to ssa-iterators.h.
6402         (swap_tree_operands): Rename and move prototype to tree-ssa-operands.h.
6403         * tree-flow-inline.h (delink_imm_use, link_imm_use_to_list,
6404         link_imm_use, set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
6405         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
6406         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
6407         num_imm_uses): Move to ssa-iterators.h.
6408         (get_use_from_ptr, get_def_from_ptr): Move to tree-ssa-operands.h
6409         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Move to
6410         tree-phinodes.h.
6411         (op_iter_done, op_iter_next_def, op_iter_next_tree,
6412         clear_and_done_ssa_iter, op_iter_init, op_iter_init_use,
6413         op_iter_init_def, op_iter_init_tree, single_ssa_tree_operand,
6414         single_ssa_use_operand, single_ssa_def_operand, zero_ssa_operands,
6415         num_ssa_operands, delink_stmt_imm_use, single_phi_def,
6416         op_iter_init_phiuse, op_iter_init_phidef, end_imm_use_stmt_p,
6417         end_imm_use_stmt_traverse, move_use_after_head, link_use_stmts_after,
6418         first_imm_use_stmt, next_imm_use_stmt, first_imm_use_on_stmt,
6419         end_imm_use_on_stmt_p, next_imm_use_on_stmt): Move to ssa-iterators.h.
6420         (gimple_phi_arg_def, gimple_phi_arg_def_ptr, gimple_phi_arg_edge,
6421         gimple_phi_arg_location, gimple_phi_arg_location_from_edge,
6422         gimple_phi_arg_set_location, gimple_phi_arg_has_location, phi_nodes,
6423         phi_nodes_ptr, phi_ssa_name_p): Move to gimple.h.
6424         (set_phi_nodes): Move to tree-phinodes.h.
6425         * tree-ssa-operands.h (enum ssa_op_iter_type,
6426         struct ssa_operand_iterator_d, SSA_OP*, FOR_EACH_SSA*, SINGLE_SSA*,
6427         ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS): Move to ssa-iterators.h.
6428         (dump_decl_set): Remove prototype.
6429         (get_use_from_ptr, get_def_from_ptr): Relocate from tree-flow.h.
6430         * tree-phinodes.h: New file.  Move some prototypes from tree-flow.h.
6431         (set_phi_nodes): Relocate from tree-flow-inline.h.
6432         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Relocate from
6433         tree-flow-inline.h
6434         * tree-ssa.h: Add tree-phinodes.h, gimple-ssa.h, ssa-iterators.h to
6435         include list.  Temporarily add gimple.h to include list.
6436         * ssa-iterators.h: New file.
6437         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
6438         BREAK_FROM_IMM_USE_STMT): Relocate from tree-flow.h.
6439         (enum ssa_op_iter_type, struct ssa_operand_iterator_d, SSA_OP*,
6440         FOR_EACH_SSA*, SINGLE_SSA*, ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS):
6441         Relocate from tree-ssa-operands.h.
6442         (delink_imm_use, link_imm_use_to_list, link_imm_use,
6443         set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
6444         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
6445         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
6446         num_imm_uses, get_use_from_ptr, get_def_from_ptr,
6447         phi_arg_index_from_use, op_iter_done, op_iter_next_def,
6448         op_iter_next_tree, clear_and_done_ssa_iter, op_iter_init,
6449         op_iter_init_use, op_iter_init_def, op_iter_init_tree,
6450         single_ssa_tree_operand, single_ssa_use_operand, single_ssa_def_operand,
6451         zero_ssa_operands, num_ssa_operands, delink_stmt_imm_use,
6452         single_phi_def, op_iter_init_phiuse, op_iter_init_phidef,
6453         end_imm_use_stmt_p, end_imm_use_stmt_traverse, move_use_after_head,
6454         link_use_stmts_after, first_imm_use_stmt, next_imm_use_stmt,
6455         first_imm_use_on_stmt, end_imm_use_on_stmt_p, next_imm_use_on_stmt):
6456         Relocate from tree-flow-inline.h.
6457         * tree-outof-ssa.h: Change _SSAEXPAND_H macro to GCC_TREE_OUTOF_SSA_H.
6459 2013-10-01  Vidya Praveen  <vidyapraveen@arm.com>
6461         * aarch64-simd.md
6462         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l2<mode>_internal): Rename to ...
6463         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): ... this;
6464         Insert '\t' to output template.
6465         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): New.
6466         (aarch64_saddl2<mode>, aarch64_uaddl2<mode>): Modify to call
6467         gen_aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal() instead.
6468         (aarch64_ssubl2<mode>, aarch64_usubl2<mode>): Ditto.
6470 2013-10-01  Uros Bizjak  <ubizjak@gmail.com>
6472         * doc/install.texi (Host/target specific installation notes for GCC):
6473         Put @anchor before @heading.
6474         * doc/gcc.texi (titlepage): Use @uref and http:// prefix for website.
6475         Use @email for email addresses.
6477 2013-10-01  Jeff Law  <law@redhat.com>
6479         * tree-ssa-threadedge.c (thread_across_edge): Make path a pointer to
6480         a vec.  Only delete the path if we create one without successfully
6481         registering a jump thread.
6482         * tree-ssa-threadupdate.h (register_jump_thread): Pass in path vector
6483         as a pointer.
6484         * tree-ssa-threadupdate.c (threaded_edges): Remove.  No longer used
6485         (paths): New vector of jump threading paths.
6486         (THREAD_TARGET, THREAD_TARGET2): Remove accessor macros.
6487         (THREAD_PATH): New accessor macro for the entire thread path.
6488         (lookup_redirection_data): Get intermediate and final outgoing edge
6489         from the thread path.
6490         (create_edge_and_update_destination_phis): Copy the threading path.
6491         (ssa_fix_duplicate_block_edges): Get edges and block types from the
6492         jump threading path.
6493         (ssa_redirect_edges): Get edges and block types from the jump threading
6494         path.  Free the path vector.
6495         (thread_block): Get edges from the jump threading path.  Look at the
6496         entire path to see if we thread to a loop exit.  If we cancel a jump
6497         thread request, then free the path vector.
6498         (thread_single_edge): Get edges and block types from the jump threading
6499         path.  Free the path vector.
6500         (thread_through_loop_header): Get edges and block types from the jump
6501         threading path.  Free the path vector.
6502         (mark_threaded_blocks): Iterate over the vector of paths and store
6503         the path on the appropriate edge.  Get edges and block types from the
6504         jump threading path.
6505         (mark_threaded_blocks): Get edges and block types from the jump
6506         threading path.  Free the path vector.
6507         (thread_through_all_blocks): Use the vector of paths rather than
6508         a vector of 3-edge sets.
6509         (register_jump_thread): Accept pointer to a path vector rather
6510         than the path vector itself.  Store the path vector for later use.
6511         Simplify.
6513 2013-10-01  Jakub Jelinek  <jakub@redhat.com>
6514             Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6516         PR target/58574
6517         * config/s390/s390.c (s390_split_branches): Modify check for table
6518         jump insns.
6519         (s390_chunkify_start): Rearrange table jump insn check in order to
6520         deal with compare and branch insns correctly.
6522 2013-10-01  Kugan Vivekanandarajah  <kuganv@linaro.org>
6524         PR target/58578
6525         Revert
6526         2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
6527         * config/arm/arm.md (arm_ashldi3_1bit):  define_insn into
6528         define_insn_and_split.
6529         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
6530         (shiftsi3_compare): New pattern.
6531         (rrx): New pattern.
6532         * config/arm/unspecs.md (UNSPEC_RRX): New.
6534 2013-10-01  Alan Modra  <amodra@gmail.com>
6536         * stmt.c (expand_asm_operands): Revert part of 2013-09-24 special
6537         casing inout operands.
6539 2013-10-01  Richard Biener  <rguenther@suse.de>
6541         PR tree-optimization/58553
6542         * tree-loop-distribution.c (struct partition_s): Add niter member.
6543         (classify_partition): Populate niter member for the partition
6544         and properly identify whether the relevant store happens before
6545         or after the loop exit.
6546         (generate_memset_builtin): Use niter member from the partition.
6547         (generate_memcpy_builtin): Likewise.
6549 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
6551         * vec.h (vec_prefix, vec): Prefix member names with "m_".
6552         * vec.c (vec_prefix::calculate_allocation): Update accordingly.
6554 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
6556         * basic-block.h (edge_list): Prefix member names with "m_".
6557         * context.h (context): Likewise.
6558         * domwalk.h (dom_walker): Likewise.
6559         * gengtype-state.c (s_expr_writer, state_writer): Likewise.
6560         * graphite-sese-to-poly.c (sese_dom_walker): Likewise.
6561         * hash-table.h (hash_table): Likewise.
6562         * machmode.h (bit_field_mode_iterator): Likewise.
6563         * pass_manager.h (pass_list): Likewise.
6564         * tree-into-ssa.c (mark_def_dom_walker): Likewise.
6565         * tree-pass.h (pass_data): Likewise.
6566         * tree-ssa-dom.c (dom_opt_dom_walker): Likewise.
6567         * tree-ssa-phiopt.c (nontrapping_dom_walker): Likewise,
6568         * tree-ssa-uncprop.c (uncprop_dom_walker): Likewise.
6569         * asan.c (pass_data_asan): Update accordingly.
6570         * cfganal.c (control_dependences::find_control_dependence): Likewise.
6571         (control_dependences::control_dependences): Likewise.
6572         (control_dependences::~control_dependences): Likewise.
6573         (control_dependences::~control_dependences): Likewise.
6574         (control_dependences::get_edges_dependent_on): Likewise.
6575         * cgraphbuild.c (pass_data_rebuild_cgraph_edges::clone): Likewise.
6576         (pass_data_remove_cgraph_callee_edges::clone): Likewise.
6577         * context.c (gcc::context::context): Likewise.
6578         * cprop.c (pass_rtl_cprop::clone): Likewise.
6579         * domwalk.c (dom_walker::walk): Likewise.
6580         * ipa-inline-analysis.c (pass_inline_parameters::clone): Likewise.
6581         * ipa-pure-const.c (pass_local_pure_const::clone): Likewise.
6582         * mode-switching.c (pass_mode_switching::clone): Likewise.
6583         * passes.c (opt_pass::opt_pass): Likewise.
6584         (pass_manager::pass_manager): Likewise.
6585         * predict.c (pass_strip_predict_hints::clone): Likewise.
6586         * recog.c (pass_data pass_data_peephole2::clone): Likewise.
6587         (pass_split_all_insns::clone): Likewise.
6588         * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
6589         Likewise.
6590         (bit_field_mode_iterator::next_mode): Likewise.
6591         (bit_field_mode_iterator::prefer_smaller_modes): Likewise.
6592         * tree-cfg.c (pass_split_crit_edges::clone): Likewise.
6593         * tree-cfgcleanup.c (pass_merge_phi::clone): Likewise.
6594         * tree-complex.c (pass_lower_complex::clone): Likewise.
6595         * tree-eh.c (pass_cleanup_eh::clone): Likewise.
6596         * tree-object-size.c (pass_object_sizes::clone): Likewise.
6597         * tree-optimize.c (pass_fixup_cfg::clone): Likewise.
6598         * tree-ssa-ccp.c (pass_data_ccp::clone): Likewise.
6599         (pass_fold_builtins::clone): Likewise.
6600         * tree-ssa-copy.c (pass_data_copy_prop::clone): Likewise.
6601         * tree-ssa-copyrename.c (pass_rename_ssa_copies::clone): Likewise.
6602         * tree-ssa-dce.c (pass_dce::clone, pass_dce_loop::clone): Likewise.
6603         (pass_cd_dce::clone): Likewise.
6604         * tree-ssa-dom.c (pass_dominator::clone): Likewise.
6605         (pass_phi_only_cprop::clone): Likewise.
6606         * tree-ssa-dse.c (pass_dse::clone): Likewise.
6607         * tree-ssa-forwprop.c (pass_forwprop::clone): Likewise.
6608         * tree-ssa-loop.c (pass_lim::clone): Likewise.
6609         * tree-ssa-phiopt.c (pass_phiopt::clone): Likewise.
6610         * tree-ssa-pre.c (pass_fre::clone): Likewise.
6611         * tree-ssa-reassoc.c (pass_reassoc::clone): Likewise.
6612         * tree-ssa-uninit.c (pass_late_warn_uninitialized::clone): Likewise.
6613         * tree-tailcall.c (pass_tail_recursion::clone): Likewise.
6614         * tree-vect-generic.c (pass_lower_vector_ssa::clone): Likewise.
6615         * tree-vrp.c (pass_vrp::clone): Likewise.
6616         * tsan.c (pass_tsan::clone): Likewise.
6618 2013-09-30  Jakub Jelinek  <jakub@redhat.com>
6620         PR middle-end/58564
6621         * fold-const.c (tree_unary_nonnegative_warnv_p): Use
6622         INTEGRAL_TYPE_P (t) instead of TREE_CODE (t) == INTEGER_TYPE.
6624         PR middle-end/58564
6625         * fold-const.c (fold_ternary_loc): For A < 0 : <sign bit of A> : 0
6626         optimization, punt if sign_bit_p looked through any zero extension.
6628 2013-09-30  Teresa Johnson  <tejohnson@google.com>
6630         * tree-ssa-threadupdate.c (ssa_fix_duplicate_block_edges):
6631         Update redirected out edge count in joiner case.
6632         (ssa_redirect_edges): Common the joiner and non-joiner cases
6633         so that joiner case gets profile updates.
6635 2013-09-30  Richard Biener  <rguenther@suse.de>
6637         PR tree-optimization/58554
6638         * tree-loop-distribution.c (classify_partition): Require
6639         unconditionally executed stores for memcpy and memset recognition.
6640         (tree_loop_distribution): Calculate dominance info.
6642 2013-09-30  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
6644         * config/aarch64/aarch64.h (MCOUNT_NAME): Define.
6645         (NO_PROFILE_COUNTERS): Likewise.
6646         (PROFILE_HOOK): Likewise.
6647         (FUNCTION_PROFILER): Likewise.
6648         * config/aarch64/aarch64.c (aarch64_function_profiler): Remove.
6650 2013-09-30  Iain Sandoe  <iain@codesourcery.com>
6652         * config/rs6000/darwin.md (load_macho_picbase_si): Wrap machopic
6653         calls and defines in TARGET_MACHO conditional.
6654         (load_macho_picbase_di): Likewise.
6655         (reload_macho_picbase): Likewise.
6656         (reload_macho_picbase_si): Likewise.
6657         (reload_macho_picbase_di): Likewise.
6658         (nonlocal_goto_receiver): Likewise.
6660 2013-09-30  Nick Clifton  <nickc@redhat.com>
6662         * config/msp430/msp430.c (msp430x_names): New array.  Lists MCUs
6663         that use the MSP430X ISA.
6664         (msp430_option_override): Scan -mmcu command line option for any
6665         MCU name that supports the MSP430X ISA.
6666         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add matches for known
6667         -mmcu options which enable the MSP430X ISA.
6669 2013-09-30  Richard Biener  <rguenther@suse.de>
6671         PR middle-end/58532
6672         * tree-cfg.c (make_abnormal_goto_edges): Skip debug statements
6673         before looking for setjmp-like calls.
6675 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
6677         PR target/10901
6678         * config/darwin-protos.h (machopic_get_function_picbase): New.
6679         * config/darwin.c (machopic_get_function_picbase): New.
6680         * config/rs6000/darwin.md (load_macho_picbase_si): Update picbase
6681         label for a new func.  (load_macho_picbase_di): Likewise.
6682         (reload_macho_picbase): New expand.
6683         (reload_macho_picbase_si): New insn.
6684         (reload_macho_picbase_di): New insn.
6685         (nonlocal_goto_receiver): New define and split.
6686         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_RELD_MPIC.
6687         (unspecv enum): Add UNSPECV_NLGR.
6689 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
6691         * config/rs6000/rs6000.c (rs6000_init_dwarf_reg_sizes_extra): Ensure
6692         that altivec registers are correctly sized on Darwin.
6694 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
6696         * config/t-darwin (darwin.o, darwin-c.o, darwin-f.o,
6697         darwin-driver.o): Use COMPILE and POSTCOMPILE.
6698         * config/x-darwin (host-darwin.o): Likewise.
6699         * config/i386/x-darwin (host-i386-darwin.o): Likewise.
6700         * config/rs6000/x-darwin (host-ppc-darwin.o): Likewise.
6701         * config/rs6000/x-darwin64 (host-ppc64-darwin.o): Likewise.
6703 2013-09-29  Uros Bizjak  <ubizjak@gmail.com>
6705         * doc/invoke.texi: Fix usage of @tie{} command.
6707 2013-09-29  Eric Botcazou  <ebotcazou@adacore.com>
6709         * config/sparc/sync.md: Add peephole for consecutive memory barriers.
6711 2013-09-28  Jan Hubicka  <jh@suse.cz>
6713         * config/i386/x86-tune.def: Add documentation for each of the options;
6714         add whitespace.
6716 2013-09-28  Jan Hubicka  <jh@suse.cz>
6718         * x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL): Enable for
6719         generic.
6720         (X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Likewise.
6721         (X86_TUNE_FOUR_JUMP_LIMIT): Drop for generic and buldozer.
6722         (X86_TUNE_PAD_RETURNS): Drop for buldozer chips.
6723         (X86_TUNE_AVOID_VECTOR_DECODE): Drop for generic.
6724         (X86_TUNE_REASSOC_FP_TO_PARALLEL): Enable for generic.
6726 2013-09-28  Richard Sandiford  <rdsandiford@googlemail.com>
6728         * alloc-pool.c, asan.c, auto-inc-dec.c, basic-block.h, bb-reorder.c,
6729         bitmap.c, bitmap.h, bt-load.c, builtins.c, calls.c, cfgcleanup.c,
6730         cfgexpand.c, cfghooks.c, cfgloop.c, cfgloopmanip.c, cfgrtl.c, cgraph.c,
6731         cgraph.h, cgraphbuild.c, cgraphclones.c, cgraphunit.c, collect2.c,
6732         combine-stack-adj.c, combine.c, compare-elim.c, context.c, context.h,
6733         cprop.c, cse.c, cselib.c, dbxout.c, dce.c, defaults.h, df-core.c,
6734         df-problems.c, df-scan.c, df.h, diagnostic.c, double-int.c, dse.c,
6735         dumpfile.c, dwarf2asm.c, dwarf2cfi.c, dwarf2out.c, emit-rtl.c,
6736         errors.c, except.c, expmed.c, expr.c, file-find.c, final.c,
6737         fixed-value.c, fold-const.c, function.c, fwprop.c, gcc-ar.c, gcc.c,
6738         gcov-io.c, gcov-io.h, gcov.c, gcse.c, genattr-common.c, genattr.c,
6739         genattrtab.c, genautomata.c, genconfig.c, genemit.c, genextract.c,
6740         genflags.c, gengenrtl.c, gengtype-state.c, gengtype.c, genmodes.c,
6741         genopinit.c, genoutput.c, genpeep.c, genpreds.c, genrecog.c,
6742         gensupport.c, ggc-common.c, ggc-page.c, gimple-fold.c, gimple-low.c,
6743         gimple-pretty-print.c, gimple-ssa-strength-reduction.c, gimple.c,
6744         gimple.h, godump.c, graphite-clast-to-gimple.c,
6745         graphite-optimize-isl.c, graphite-poly.h, graphite-sese-to-poly.c,
6746         graphite.c, haifa-sched.c, hash-table.c, hash-table.h, hwint.c,
6747         hwint.h, ifcvt.c, incpath.c, init-regs.c, input.h, intl.c, intl.h,
6748         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
6749         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c,
6750         ipa-utils.c, ipa.c, ira-build.c, ira.c, jump.c, loop-doloop.c,
6751         loop-init.c, loop-invariant.c, loop-iv.c, lower-subreg.c, lto-cgraph.c,
6752         lto-streamer-in.c, lto-streamer-out.c, lto-wrapper.c, mcf.c,
6753         mode-switching.c, modulo-sched.c, omp-low.c, optabs.c, opts.c,
6754         pass_manager.h, passes.c, plugin.c, postreload-gcse.c, postreload.c,
6755         predict.c, prefix.c, pretty-print.c, print-rtl.c, print-tree.c,
6756         profile.c, read-md.c, real.c, real.h, recog.c, ree.c, reg-stack.c,
6757         regcprop.c, reginfo.c, regmove.c, regrename.c, regs.h, regstat.c,
6758         reload1.c, reorg.c, rtl.c, rtl.h, rtlanal.c, sbitmap.c, sched-rgn.c,
6759         sdbout.c, sel-sched-ir.c, sel-sched.c, sparseset.c, stack-ptr-mod.c,
6760         statistics.c, stmt.c, stor-layout.c, store-motion.c, streamer-hooks.h,
6761         system.h, target-hooks-macros.h, targhooks.c, targhooks.h, toplev.c,
6762         tracer.c, trans-mem.c, tree-browser.c, tree-call-cdce.c, tree-cfg.c,
6763         tree-cfgcleanup.c, tree-complex.c, tree-data-ref.c, tree-data-ref.h,
6764         tree-eh.c, tree-emutls.c, tree-flow.h, tree-if-conv.c, tree-into-ssa.c,
6765         tree-iterator.c, tree-loop-distribution.c, tree-mudflap.c,
6766         tree-nested.c, tree-nomudflap.c, tree-nrv.c, tree-object-size.c,
6767         tree-optimize.c, tree-pass.h, tree-pretty-print.c, tree-profile.c,
6768         tree-scalar-evolution.c, tree-sra.c, tree-ssa-ccp.c,
6769         tree-ssa-coalesce.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
6770         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
6771         tree-ssa-ifcombine.c, tree-ssa-live.c, tree-ssa-loop-ch.c,
6772         tree-ssa-loop-im.c, tree-ssa-loop-ivopts.c, tree-ssa-loop-prefetch.c,
6773         tree-ssa-loop.c, tree-ssa-math-opts.c, tree-ssa-operands.c,
6774         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
6775         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
6776         tree-ssa-structalias.c, tree-ssa-threadedge.c, tree-ssa-threadupdate.c,
6777         tree-ssa-uncprop.c, tree-ssa-uninit.c, tree-ssa.c, tree-ssanames.c,
6778         tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
6779         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
6780         tree-vect-stmts.c, tree-vectorizer.c, tree-vectorizer.h, tree-vrp.c,
6781         tree.c, tree.h, tsan.c, tsystem.h, value-prof.c, var-tracking.c,
6782         varasm.c, vec.h, vmsdbgout.c, vtable-verify.c, web.c: Add missing
6783         whitespace before "(".
6785 2013-09-28  Sandra Loosemore  <sandra@codesourcery.com>
6787         * expr.h (extract_bit_field): Remove packedp parameter.
6788         * expmed.c (extract_fixed_bit_field): Remove packedp parameter
6789         from forward declaration.
6790         (store_split_bit_field): Remove packedp arg from calls to
6791         extract_fixed_bit_field.
6792         (extract_bit_field_1): Remove packedp parameter and packedp
6793         argument from recursive calls and calls to extract_fixed_bit_field.
6794         (extract_bit_field): Remove packedp parameter and corresponding
6795         arg to extract_bit_field_1.
6796         (extract_fixed_bit_field): Remove packedp parameter.  Remove code
6797         to issue warnings.
6798         (extract_split_bit_field): Remove packedp arg from call to
6799         extract_fixed_bit_field.
6800         * expr.c (emit_group_load_1): Adjust calls to extract_bit_field.
6801         (copy_blkmode_from_reg): Likewise.
6802         (copy_blkmode_to_reg): Likewise.
6803         (read_complex_part): Likewise.
6804         (store_field): Likewise.
6805         (expand_expr_real_1): Likewise.
6806         * calls.c (store_unaligned_arguments_into_pseudos): Adjust call
6807         to extract_bit_field.
6808         * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Adjust
6809         call to extract_bit_field.
6810         * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Adjust
6811         call to extract_bit_field.
6812         * doc/invoke.texi (Code Gen Options): Remove mention of warnings
6813         and special packedp behavior from -fstrict-volatile-bitfields
6814         documentation.
6816 2013-09-27  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
6818         * lra-eliminations.c (init_elim_table): Guard value_p.
6820 2013-09-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
6822         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Allow
6823         DFmode, DImode, and SFmode in the upper VSX registers based on the
6824         -mupper-regs-{df,sf} flags.  Fix wu constraint to be ALTIVEC_REGS
6825         if -mpower8-vector.  Combine -mvsx-timode handling with the rest
6826         of the VSX register handling.
6828         * config/rs6000/rs6000.md (f32_lv): Use %x0 for VSX regsters.
6829         (f32_sv): Likewise.
6830         (zero_extendsidi2_lfiwzx): Add support for loading into the
6831         Altivec registers with -mpower8-vector.  Use wu/wv constraints to
6832         only do VSX memory options on Altivec registers.
6833         (extendsidi2_lfiwax): Likewise.
6834         (extendsfdf2_fpr): Likewise.
6835         (mov<mode>_hardfloat, SF/SD modes): Likewise.
6836         (mov<mode>_hardfloat32, DF/DD modes): Likewise.
6837         (mov<mode>_hardfloat64, DF/DD modes): Likewise.
6838         (movdi_internal64): Likewise.
6840 2013-09-27  Xinliang David Li  <davidxl@google.com>
6842         * opts.c (finish_options): Adjust parameters
6843         according to vect cost model.
6844         (common_handle_option): Set dynamic vect cost
6845         model for FDO.
6846         targhooks.c (default_add_stmt_cost): Compute stmt cost
6847         unconditionally.
6848         * tree-vect-loop.c (vect_estimate_min_profitable_iters):
6849         Use helper function.
6850         * tree-vectorizer.h (unlimited_cost_model): New function.
6851         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use helper function.
6852         * tree-vect-data-refs.c (vect_peeling_hash_insert): Use helper
6853         function.
6854         (vect_enhance_data_refs_alignment): Ditto.
6855         * flag-types.h: New enum.
6856         * common/config/i386/i386-common.c (ix86_option_init_struct):
6857         No need to initialize vect_cost_model flag.
6858         * config/i386/i386.c (ix86_add_stmt_cost): Compute stmt cost
6859         unconditionally.
6861 2013-09-27  Diego Novillo  <dnovillo@google.com>
6863         * gimple.h (enum ssa_mode): Remove.
6865 2013-09-27  Paulo Matos  <pmatos@broadcom.com>
6867         * cfgloop.h (number_of_loops): Fix typo in check for null.
6869 2013-09-27  Jakub Jelinek  <jakub@redhat.com>
6871         PR middle-end/58551
6872         * tree-cfg.c (move_sese_region_to_fn): Also move loops that
6873         are children of outermost saved_cfun's loop, and set it up to
6874         be moved to dest_cfun's outermost loop.  Fix up num_nodes adjustments
6875         if loop != loop0 and SESE region contains bbs that belong to loop0.
6877 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
6879         * rtlanal.c (must_be_base_p, must_be_index_p): Delete.
6880         (binary_scale_code_p, get_base_term, get_index_term): New functions.
6881         (set_address_segment, set_address_base, set_address_index)
6882         (set_address_disp): Accept the argument unconditionally.
6883         (baseness): Remove must_be_base_p and must_be_index_p checks.
6884         (decompose_normal_address): Classify as much as possible in the
6885         main loop.
6887 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
6889         * cse.c (count_reg_usage): Handle INT_LIST.
6890         * lra-eliminations.c (lra_eliminate_regs_1): Likewise.
6891         * reginfo.c (reg_scan_mark_refs): Likewise.
6892         * reload1.c (eliminate_regs_1): Likewise.
6894 2013-09-27  Iain Sandoe  <iain@codesourcery.com>
6896         PR middle-end/58547
6897         * rtlanal.c (lsb_bitfield_op_p): Make both parts of the comparison
6898         signed.
6900 2013-09-27  Richard Biener  <rguenther@suse.de>
6902         PR tree-optimization/58459
6903         * tree-ssa-forwprop.c (forward_propagate_addr_expr): Remove
6904         restriction not propagating into loops.
6906 2013-09-26  Florian Weimer  <fw@deneb.enyo.de>
6908         * tree-ssa.h (walk_use_def_chains_fn, walk_use_def_chains): Delete.
6909         * tree-ssa.c (walk_use_def_chains_1, walk_use_def_chains): Delete.
6910         * doc/tree-ssa.texi (Walking use-def chains): Delete.
6912 2013-09-26  Richard Biener  <rguenther@suse.de>
6914         * tree-into-ssa.c (rewrite_into_ssa): Make more SSA names to anonymous.
6916 2013-09-26  Richard Biener  <rguenther@suse.de>
6918         * alias.h (component_uses_parent_alias_set): Rename to ...
6919         (component_uses_parent_alias_set_from): ... this.
6920         * alias.c (component_uses_parent_alias_set): Rename to ...
6921         (component_uses_parent_alias_set_from): ... this and return
6922         the desired parent.
6923         (reference_alias_ptr_type_1): Use the result from
6924         component_uses_parent_alias_set_from instead of stripping
6925         components one at a time.
6926         * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
6928 2013-09-26  Andrew MacLeod  <amacleod@redhat.com>
6930         * tree-ssa-live.h (find_replaceable_exprs, dump_replaceable_exprs):
6931         Move prototypes to...
6932         * tree-ssa-ter.h: New File.  Move prototypes here.
6933         * tree-flow.h (stmt_is_replaceable_p): Remove prototype.
6934         * tree-outof-ssa.h: New. Rename ssaexpand.h, include tree-ssa-ter.h.
6935         * tree-outof-ssa.c (ssa_is_replaceable_p): New.  Refactor common bits
6936         from is_replaceable_p.
6937         * tree-ssa-ter.c (is_replaceable_p, stmt_is_replaceable_p): Delete.
6938         (ter_is_replaceable_p): New.  Use new refactored ssa_is_replaceable_p.
6939         (process_replaceable): Use ter_is_replaceable_p.
6940         (find_replaceable_in_bb): Use ter_is_replaceable_p.
6941         * expr.c (stmt_is_replaceable_p): Relocate from tree-ssa-ter.c.  Use
6942         newly refactored ssa_is_replaceable_p.
6943         * cfgexpand.c: Include tree-outof-ssa.h.
6944         * ssaexpand.h: Delete.
6946 2013-09-26  Andrew MacLeod <amacleod@redhat.com>
6948         * gimple.c (gimple_replace_lhs): Move to tree-ssa.c and rename.
6949         (struct count_ptr_d, count_ptr_derefs, count_uses_and_derefs): Move to
6950         tree-ssa.c
6951         (create_gimple_tmp): Delete.
6952         (get_expr_type, build_assign, build_type_cast): Move to...
6953         * gimple-builder.c: New File.
6954         (get_expr_type): Relocate from gimple.c.
6955         (build_assign, build_type_cast): Change to only create ssanames.
6956         * gimple.h: Move prototypes to...
6957         * gimple-builder.h: New File. Here.
6958         * tree-ssa.h: And here.
6959         * tree-ssa.c (struct count_ptr_d, count_ptr_derefs,
6960         count_uses_and_derefs): Relocate from gimple.c.
6961         (gimple_replace_ssa_lhs): Renamed gimple_replace_ssa from gimple.c
6962         * tree-ssa-reassoc.c (repropagate_negates): Use gimple_replace_ssa_lhs.
6963         * tree-ssa-math-opts (execute_cse_reciprocals): Use
6964         gimple_replace_ssa_lhs.
6965         * asan.c: Include gimple-builder.h.
6966         * Makefile.in: Add gimple-builder.o.
6968 2013-09-26  Richard Biener  <rguenther@suse.de>
6970         * tree-ssa-live.c (var_map_base_init): Handle SSA names with
6971         DECL_IGNORED_P base VAR_DECLs like anonymous SSA names.
6972         (loe_visit_block): Use gcc_checking_assert.
6973         * tree-ssa-coalesce.c (create_outofssa_var_map): Use
6974         gimple_assign_ssa_name_copy_p.
6975         (gimple_can_coalesce_p): Adjust according to the var_map_base_init
6976         change.
6978 2013-09-26  David Edelsohn  <dje.gcc@gmail.com>
6980         * config/rs6000/t-rs6000 (rs6000.o): Remove.
6981         (rs6000-c.o): Use COMPILE and POSTCOMPILE.
6983 2013-09-26  Richard Biener  <rguenther@suse.de>
6985         PR tree-optimization/58539
6986         * tree-vect-loop.c (vect_create_epilog_for_reduction): Honor
6987         the fact that debug statements are not taking part in loop-closed
6988         SSA construction.
6990 2013-09-26  Nick Clifton  <nickc@redhat.com>
6992         * config/msp430/msp430.c (msp430_expand_epilogue): Fix compile
6993         time warning message.
6994         (msp430_print_operand_raw): Delete unused letter parameter.
6995         (TARGET_PRINT_OPERAND_ADDRESS): Define.
6996         (msp430_print_operand_address): New function.
6997         (msp430_print_operand): Move address printing code from here to
6998         new function.
6999         * config/msp430/msp430.md (movsipsi2): Add comment in generated
7000         assembler.
7001         (zero_extendpsisi2): Likewise.
7002         (extendpsisi2): New pattern.
7003         (andneghi3): New pattern.
7005 2013-09-26  Yvan Roux  <yvan.roux@linaro.org>
7007         * config/aarch64/aarch64.opt (mlra): New option.
7008         * config/aarch64/aarch64.c (aarch64_lra_p): New function.
7009         (TARGET_LRA_P): Define.
7011 2013-09-26  Eric Botcazou  <ebotcazou@adacore.com>
7013         * expr.c (expand_assignment): Remove obsolete comment.
7015 2013-09-25  Jeff Law  <law@redhat.com>
7017         * tree-flow.h (thread_through_all_blocks): Prototype moved into
7018         tree-ssa-threadupdate.h.
7019         (register_jump_thread): Similarly.
7020         * tree-ssa-threadupdate.h: New header file.
7021         * tree-ssa-dom.c: Include tree-ssa-threadupdate.h.
7022         * tree-vrp.c: Likewise.
7023         * tree-ssa-threadedge.c: Include tree-ssa-threadupdate.h.
7024         (thread_around_empty_blocks): Change type of path vector argument to
7025         an edge,type pair from just an edge.  Initialize both elements when
7026         appending to a jump threading path.  Tweak references to elements
7027         appropriately.
7028         (thread_across_edge): Similarly.  Release memory for the elements
7029         as needed.
7030         * tree-ssa-threadupdate.c: Include tree-ssa-threadupdate.h.
7031         (dump_jump_thread_path): New function broken out from
7032         register_jump_thread.
7033         (register_jump_thread): Use dump_jump_thread_path.  Change type of
7034         path vector entries.  Search the path for NULL edges and dump
7035         the path if one is found.  Tweak the conversion of path to 3-edge
7036         form to use the block copy type information embedded in the path.
7038 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
7040         * lra.c (update_inc_notes): Remove all REG_DEAD and REG_UNUSED notes.
7042 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
7043             Vladimir Makarov  <vmakarov@redhat.com>
7045         * rtlanal.c (lsb_bitfield_op_p): New predicate for bitfield operations
7046         from the least significant bit.
7047         (strip_address_mutations): Add bitfield operations handling.
7048         (must_be_index_p): Add shifting and rotate operations handling.
7049         (set_address_base): Use must_be_base_p predicate.
7050         (set_address_index): Use must_be_index_p predicate.
7052 2013-09-25  Alexander Ivchenko  <alexander.ivchenko@intel.com>
7053             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
7054             Sergey Lega  <sergey.s.lega@intel.com>
7055             Anna Tikhonova  <anna.tikhonova@intel.com>
7056             Ilya Tocar  <ilya.tocar@intel.com>
7057             Andrey Turetskiy  <andrey.turetskiy@intel.com>
7058             Ilya Verbin  <ilya.verbin@intel.com>
7059             Kirill Yukhin  <kirill.yukhin@intel.com>
7060             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
7062         * config/i386/i386.c (ix86_avx256_split_vector_move_misalign):
7063         Use new names.
7064         (ix86_expand_vector_move_misalign): Support new unaligned load and
7065         stores and use new names.
7066         (CODE_FOR_sse2_storedqu): Rename to ...
7067         (CODE_FOR_sse2_storedquv16qi): ... this.
7068         (CODE_FOR_sse2_loaddqu): Rename to ...
7069         (CODE_FOR_sse2_loaddquv16qi): ... this.
7070         (CODE_FOR_avx_loaddqu256): Rename to ...
7071         (CODE_FOR_avx_loaddquv32qi): ... this.
7072         (CODE_FOR_avx_storedqu256): Rename to ...
7073         (CODE_FOR_avx_storedquv32qi): ... this.
7074         * config/i386/i386.md (fpint_logic): New.
7075         * config/i386/sse.md (VMOVE): Extend for AVX512.
7076         (VF): Ditto.
7077         (VF_128_256): New.
7078         (VF_512): Ditto.
7079         (VI_UNALIGNED_LOADSTORE): Ditto.
7080         (sse2_avx_avx512f): Ditto.
7081         (sse2_avx2): Extend for AVX512.
7082         (sse4_1_avx2): Ditto.
7083         (avx2_avx512f): New.
7084         (sse): Extend for AVX512.
7085         (sse2): Ditto.
7086         (sse4_1): Ditto.
7087         (avxsizesuffix): Ditto.
7088         (sseintvecmode): Ditto.
7089         (ssePSmode): Ditto.
7090         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Ditto.
7091         (<sse>_storeu<ssemodesuffix><avxsizesuffix>): Ditto.
7092         (<sse2>_loaddqu<avxsizesuffix>): Extend for AVX512 and rename to ...
7093         (<sse2_avx_avx512f>_loaddqu<mode>): ... this.
7094         (<sse2>_storedqu<avxsizesuffix>): Extend for AVX512 and rename to ...
7095         (<sse2_avx_avx512f>_storedqu<mode): ... this.
7096         (<sse>_movnt<mode>): Replace constraint "x" with "v".
7097         (STORENT_MODE): Extend for AVX512.
7098         (*absneg<mode>2): Replace constraint "x" with "v".
7099         (*mul<mode>3): Ditto.
7100         (*ieee_smin<mode>3): Ditto.
7101         (*ieee_smax<mode>3): Ditto.
7102         (avx_cmp<mode>3): Replace VF with VF_128_256.
7103         (*<sse>_maskcmp<mode>3_comm): Ditto.
7104         (<sse>_maskcmp<mode>3): Ditto.
7105         (<sse>_andnot<mode>3): Extend for AVX512.
7106         (<code><mode>3, anylogic): Replace VF with VF_128_256.
7107         (<code><mode>3, fpint_logic): New.
7108         (*<code><mode>3): Extend for AVX512.
7109         (avx512flogicsuff): New.
7110         (avx512f_<logic><mode>): Ditto.
7111         (<sse>_movmsk<ssemodesuffix><avxsizesuffix>): Replace VF with
7112         VF_128_256.
7113         (<sse4_1>_blend<ssemodesuffix><avxsizesuffix>): Ditto.
7114         (<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>): Ditto.
7115         (<sse4_1>_dp<ssemodesuffix><avxsizesuffix>): Ditto.
7116         (avx_vtest<ssemodesuffix><avxsizesuffix>): Ditto.
7117         (<sse4_1>_round<ssemodesuffix><avxsizesuffix>): Ditto.
7118         (xop_vpermil2<mode>3): Ditto.
7119         (*avx_vpermilp<mode>): Extend for AVX512 and rename to ...
7120         (*<sse2_avx_avx512f>_vpermilp<mode>): ... this.
7121         (avx_vpermilvar<mode>3): Extend for AVX512 and rename to ...
7122         (<sse2_avx_avx512f>_vpermilvar<mode>3): ... this.
7124 2013-09-25  Tom Tromey  <tromey@redhat.com>
7126         * Makefile.in (PARTITION_H, LTO_SYMTAB_H, COMMON_TARGET_DEF_H)
7127         (RTL_ERROR_H, TRANS_MEM_H, COVERAGE_H, DEMANGLE_H, ALIAS_H)
7128         (SCHED_INT_H, SEL_SCHED_IR_H, SEL_SCHED_DUMP_H, VALTRACK_H, DDG_H)
7129         (GGC_INTERNAL_H, DECNUM_H, BACKTRACE_H, MKDEPS_H, TREE_HASHER_H)
7130         (TREE_SSA_LIVE_H, SSAEXPAND_H, DWARF2OUT_H, SCEV_H, OMEGA_H)
7131         (TREE_DATA_REF_H, IRA_INT_H, LRA_INT_H, DBGCNT_H, DATA_STREAMER_H)
7132         (GIMPLE_STREAMER_H, TREE_STREAMER_H, STREAMER_HOOKS_H)
7133         (TREE_VECTORIZER_H, IPA_INLINE_H, GSTAB_H, LIBFUNCS_H)
7134         (GRAPHITE_HTAB_H): Remove.
7136 2013-09-25  Tom Tromey  <tromey@redhat.com>
7138         * config/mcore/t-mcore (CROSS_FLOAT_H): Remove.
7140 2013-09-25  Tom Tromey  <tromey@redhat.com>
7142         * config/t-glibc (glibc-c.o): Use COMPILE and POSTCOMPILE.
7144 2013-09-25  Tom Tromey  <tromey@redhat.com>
7146         * config/i386/t-i386 (i386.o): Remove.
7147         (i386-c.o): Use COMPILE and POSTCOMPILE.
7149 2013-09-25  Tom Tromey  <tromey@redhat.com>
7151         * Makefile.in ($(out_object_file)): Use COMPILE and POSTCOMPILE.
7153 2013-09-25  Tom Tromey  <tromey@redhat.com>
7155         * Makefile.in (graph.o, sbitmap.o, sparseset.o, gcc-ar.o)
7156         (gcc-ranlib.o, gcc-nm.o, collect2.o, collect2-aix.o, tlink.o)
7157         (lto-wrapper.o, default-c.o, attribs.o, incpath.o, prefix.o)
7158         (gcc.o, options.o, options-save.o, version.o, gtype-desc.o)
7159         (trans-mem.o, ggc-common.o, ggc-page.o, ggc-none.o, stringpool.o)
7160         (convert.o, double-int.o, lto-compress.o, data-streamer-in.o)
7161         (data-streamer-out.o, data-streamer.o, gimple-streamer-in.o)
7162         (gimple-streamer-out.o, tree-streamer.o, tree-streamer-in.o)
7163         (tree-streamer-out.o, streamer-hooks.o, lto-cgraph.o)
7164         (lto-streamer-in.o, lto-streamer-out.o, lto-section-in.o)
7165         (lto-section-out.o, lto-opts.o, lto-streamer.o, langhooks.o)
7166         (test-dump.o, tree.o, tree-dump.o, tree-inline.o, print-tree.o)
7167         (stor-layout.o, asan.o, tsan.o, ubsan.o, tree-ssa-tail-merge.o)
7168         (tree-ssa-structalias.o, tree-ssa-uninit.o, tree-ssa.o)
7169         (tree-into-ssa.o, tree-ssa-ter.o, tree-ssa-coalesce.o)
7170         (tree-outof-ssa.o, tree-ssa-dse.o, tree-ssa-forwprop.o)
7171         (tree-ssa-phiprop.o, tree-ssa-ifcombine.o, tree-ssa-phiopt.o)
7172         (tree-nrv.o, tree-ssa-copy.o, tree-ssa-propagate.o)
7173         (tree-ssa-dom.o, tree-ssa-uncprop.o, tree-ssa-threadedge.o)
7174         (tree-ssa-threadupdate.o, tree-ssanames.o, tree-phinodes.o)
7175         (domwalk.o, tree-ssa-live.o, tree-ssa-copyrename.o)
7176         (tree-ssa-pre.o, tree-ssa-sccvn.o)
7177         (gimple-ssa-strength-reduction.o, tree-vrp.o, tree-cfg.o)
7178         (tree-cfgcleanup.o, tree-tailcall.o, tree-ssa-sink.o)
7179         (tree-nested.o, tree-if-conv.o, tree-iterator.o, tree-dfa.o)
7180         (tree-ssa-operands.o, tree-eh.o, tree-ssa-loop.o)
7181         (tree-ssa-loop-unswitch.o, tree-ssa-address.o)
7182         (tree-ssa-loop-niter.o, tree-ssa-loop-ivcanon.o)
7183         (tree-ssa-loop-ch.o, tree-ssa-loop-prefetch.o, tree-predcom.o)
7184         (tree-ssa-loop-ivopts.o, tree-affine.o, tree-ssa-loop-manip.o)
7185         (tree-ssa-loop-im.o, tree-ssa-math-opts.o, tree-ssa-alias.o)
7186         (tree-ssa-reassoc.o, tree-optimize.o, gimplify.o)
7187         (gimple-iterator.o, gimple-fold.o, gimple-low.o, omp-low.o)
7188         (tree-browser.o, omega.o, tree-chrec.o, tree-scalar-evolution.o)
7189         (tree-data-ref.o, sese.o, graphite.o, graphite-blocking.o)
7190         (graphite-clast-to-gimple.o, graphite-dependences.o)
7191         (graphite-interchange.o, graphite-poly.o)
7192         (graphite-scop-detection.o, graphite-sese-to-poly.o)
7193         (graphite-optimize-isl.o, tree-vect-loop.o)
7194         (tree-vect-loop-manip.o, tree-vect-patterns.o, tree-vect-slp.o)
7195         (tree-vect-stmts.o, tree-vect-data-refs.o, tree-vectorizer.o)
7196         (vtable-verify.o, tree-loop-distribution.o, tree-parloops.o)
7197         (tree-stdarg.o, tree-object-size.o, internal-fn.o, gimple.o)
7198         (gimple-pretty-print.o, tree-mudflap.o, tree-nomudflap.o)
7199         (tree-pretty-print.o, tree-diagnostic.o, fold-const.o)
7200         (diagnostic.o, diagnostic-color.o, opts.o, opts-global.o)
7201         (opts-common.o, targhooks.o, common/common-targhooks.o, input.o)
7202         (toplev.o, hwint.o, passes.o, plugin.o, main.o, host-default.o)
7203         (rtl-error.o, rtl.o, print-rtl.o, rtlanal.o, varasm.o, function.o)
7204         (statistics.o, stmt.o, except.o, expr.o, dojump.o, builtins.o)
7205         (calls.o, expmed.o, explow.o, optabs.o, dbxout.o, debug.o)
7206         (sdbout.o, dwarf2out.o, dwarf2cfi.o, dwarf2asm.o, vmsdbgout.o)
7207         (xcoffout.o, godump.o, emit-rtl.o, real.o, realmpfr.o, dfp.o)
7208         (fixed-value.o, jump.o, simplify-rtx.o, symtab.o, cgraph.o)
7209         (cgraphunit.o, cgraphclones.o, cgraphbuild.o, varpool.o, ipa.o)
7210         (ipa-profile.o, ipa-devirt.o, ipa-prop.o, ipa-ref.o, ipa-cp.o)
7211         (ipa-split.o, ipa-inline.o, ipa-inline-analysis.o)
7212         (ipa-inline-transform.o, ipa-utils.o, ipa-reference.o)
7213         (ipa-pure-const.o, coverage.o, cselib.o, cse.o, dce.o, dumpfile.o)
7214         (dse.o, fwprop.o, web.o, ree.o, cprop.o, gcse.o, store-motion.o)
7215         (resource.o, lcm.o, mode-switching.o, tree-ssa-dce.o)
7216         (tree-call-cdce.o, tree-ssa-ccp.o, tree-ssa-strlen.o, tree-sra.o)
7217         (tree-switch-conversion.o, tree-complex.o, tree-emutls.o)
7218         (tree-vect-generic.o, df-core.o, df-problems.o, df-scan.o)
7219         (regstat.o, valtrack.o, var-tracking.o, profile.o, mcf.o)
7220         (tree-profile.o, value-prof.o, loop-doloop.o, alloc-pool.o)
7221         (auto-inc-dec.o, cfg.o, cfghooks.o, cfgexpand.o, cfgrtl.o)
7222         (cfganal.o, cfgbuild.o, cfgcleanup.o, cfgloop.o, cfgloopanal.o)
7223         (graphds.o, loop-iv.o, loop-invariant.o, cfgloopmanip.o)
7224         (loop-init.o, loop-unswitch.o, loop-unroll.o, dominance.o)
7225         (et-forest.o, combine.o, reginfo.o, bitmap.o, vec.o, hash-table.o)
7226         (reload.o, reload1.o, rtlhooks.o, postreload.o, postreload-gcse.o)
7227         (caller-save.o, bt-load.o, reorg.o, alias.o, stack-ptr-mod.o)
7228         (init-regs.o, ira-build.o, ira-costs.o, ira-conflicts.o)
7229         (ira-color.o, ira-emit.o, ira-lives.o, ira.o, lra.o)
7230         (lra-assigns.o, lra-coalesce.o, lra-constraints.o)
7231         (lra-eliminations.o, lra-lives.o, lra-spills.o, regmove.o)
7232         (combine-stack-adj.o, compare-elim.o, ddg.o, modulo-sched.o)
7233         (haifa-sched.o, sched-deps.o, sched-rgn.o, sched-ebb.o)
7234         (sched-vis.o, sel-sched.o, sel-sched-dump.o, sel-sched-ir.o)
7235         (final.o, recog.o, reg-stack.o, sreal.o, predict.o, lists.o)
7236         (bb-reorder.o, tracer.o, timevar.o, regcprop.o, regrename.o)
7237         (ifcvt.o, params.o, pointer-set.o, hooks.o, pretty-print.o)
7238         (errors.o, dbgcnt.o, lower-subreg.o, target-globals.o)
7239         (hw-doloop.o, file-find.o, context.o, $(common_out_object_file))
7240         (insn-attrtab.o, insn-automata.o, insn-dfatab.o, insn-emit.o)
7241         (insn-enums.o, insn-extract.o, insn-latencytab.o, insn-modes.o)
7242         (insn-opinit.o, insn-output.o, insn-peep.o, insn-preds.o)
7243         (insn-recog.o, intl.o, cppbuiltin.o, cppdefault.o, gcov.o)
7244         (gcov-dump.o): Remove.
7245         (default-c.o): Use COMPILE and POSTCOMPILE.
7246         (CFLAGS-gcc.o): New variable.
7247         ($(common_out_object_file)): Use COMPILE and POSTCOMPILE.
7249 2013-09-25  Tom Tromey  <tromey@redhat.com>
7251         * Makefile.in (c-family/cppspec.o, c-family/c-common.o)
7252         (c-family/c-cppbuiltin.o, c-family/c-dump.o, c-family/c-format.o)
7253         (c-family/c-gimplify.o, c-family/c-lex.o, c-family/c-omp.o)
7254         (c-family/c-opts.o, c-family/c-pch.o, c-family/c-ppoutput.o)
7255         (c-family/c-pragma.o, c-family/c-pretty-print.o)
7256         (c-family/c-semantics.o, c-family/c-ada-spec.o)
7257         (c-family/array-notation-common.o, c-family/stub-objc.o)
7258         (c-family/c-ubsan.o): Remove.
7260 2013-09-25  Tom Tromey  <tromey@redhat.com>
7262         * Makefile.in (C_TREE_H): Reference c/c-tree.h.
7264 2013-09-25  Tom Tromey  <tromey@redhat.com>
7266         * Makefile.in (DRIVER_DEFINES): Use $(and), not shell code,
7267         to add -DENABLE_SHARED_LIBGCC.
7268         (gcc.o): Don't use subshell.
7270 2013-09-25  Tom Tromey  <tromey@redhat.com>
7272         * Makefile.in (OUTPUT_OPTION): Define as "-o $@".
7273         * configure.ac: Don't invoke AM_PROG_CC_C_O.
7274         (NO_MINUS_C_MINUS_O, OUTPUT_OPTION): Don't subst.
7275         * configure, config.in: Rebuild.
7277 2013-09-25  Tom Tromey  <tromey@redhat.com>
7279         * Makefile.in (CCDEPMODE, DEPDIR, depcomp, COMPILE.base)
7280         (COMPILE, POSTCOMPILE): New variables.
7281         (.cc.o .c.o): Use COMPILE, POSTCOMPILE.
7282         (DEPFILES): New variable.
7283         Include ".Po" files.
7284         * configure.ac: Add checks for dependency checking.
7285         * configure, aclocal.m4: Regenerate.
7287 2013-09-25  Tom Tromey  <tromey@redhat.com>
7289         * Makefile.in (ALL_HOST_BACKEND_OBJS): Add lto-wrapper.o.
7290         ($(ALL_HOST_OBJS)): Move order-only dependency to end of file.
7292 2013-09-25  Tom Tromey  <tromey@redhat.com>
7294         * Makefile.in (generated_files): Add options.h,
7295         target-hooks-def.h, insn-opinit.h,
7296         common/common-target-hooks-def.h, pass-instances.def,
7297         c-family/c-target-hooks-def.h.
7299 2013-09-25  Jeff Law  <law@redhat.com>
7301         * tree-ssa-threadedge.c (thread_across_edge): Use foo.last () rather
7302         than foo[foo.length () - 1] to access last member in a vec.
7303         * tree-ssa-threadupdate.c (register_jump_thread): Similarly.
7305 2013-09-25  Richard Biener  <rguenther@suse.de>
7307         PR middle-end/58521
7308         * tree.c (iterative_hash_expr): Remove MEM_REF special handling.
7310 2013-09-25  Jan Hubicka  <jh@suse.cz>
7312         * cgraph.c (cgraph_resolve_speculation): Use semantical equivalency
7313         test.
7315 2013-09-25  Marek Polacek  <polacek@redhat.com>
7317         PR sanitizer/58420
7318         * ubsan.c (ubsan_type_descriptor): Handle IDENTIFIER_NODEs
7319         when determining the type name.
7321 2013-09-24  Oleg Endo  <olegendo@gcc.gnu.org>
7323         * config/sh/sh.md: Fix formatting.
7325 2013-09-24  Xinliang David Li  <davidxl@google.com>
7327         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Check
7328         max peel iterations parameter.
7329         * param.def: New parameter.
7330         * doc/invoke.texi: Document New parameter.
7332 2013-09-24  Christophe Lyon  <christophe.lyon@linaro.org>
7334         * gimple-pretty-print.c: Various whitespace tweaks.
7335         * tree-core.h: Likewise.
7336         * tree-pretty-print.c: Likewise.
7337         * tree-ssa-alias.c: Likewise.
7338         * tree-ssa-copy.c: Likewise.
7339         * tree-ssanames.c: Likewise.
7340         * tree-ssanames.h: Likewise.
7341         * tree-vrp.c: Likewise.
7343 2013-09-24  Alan Modra  <amodra@gmail.com>
7345         PR middle-end/57134
7346         PR middle-end/57586
7347         * stmt.c (expand_asm_operands): Call expand_expr with EXPAND_MEMORY
7348         for output operands that disallow regs.  Don't use EXPAND_WRITE on
7349         inout operands.
7351 2013-09-24  Richard Biener  <rguenther@suse.de>
7353         PR middle-end/58513
7354         * tree.c (reference_alias_ptr_type): Move ...
7355         * alias.c (reference_alias_ptr_type): ... here and implement
7356         in terms of the new reference_alias_ptr_type_1.
7357         (ref_all_alias_ptr_type_p): New helper.
7358         (get_deref_alias_set_1): Drop flag_strict_aliasing here,
7359         use ref_all_alias_ptr_type_p.
7360         (get_deref_alias_set): Add flag_strict_aliasing check here.
7361         (reference_alias_ptr_type_1): New function, split out from ...
7362         (get_alias_set): ... here.
7363         (alias_ptr_types_compatible_p): New function.
7364         * alias.h (reference_alias_ptr_type): Declare.
7365         (alias_ptr_types_compatible_p): Likewise.
7366         * tree.h (reference_alias_ptr_type): Remove.
7367         * fold-const.c (operand_equal_p): Use alias_ptr_types_compatible_p
7368         to compare MEM_REF alias types.
7370 2013-09-24  Richard Biener  <rguenther@suse.de>
7372         * tree-vrp.c (vrp_finalize): Check for SSA name presence.
7374 2013-09-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
7376         * config/rs6000/rs6000.c (rs6000_vector_reload): Delete, combine
7377         reload helper function arrays into a single array reg_addr.
7378         (reload_fpr_gpr): Likewise.
7379         (reload_gpr_vsx): Likewise.
7380         (reload_vsx_gpr): Likewise.
7381         (struct rs6000_reg_addr): Likewise.
7382         (reg_addr): Likewise.
7383         (rs6000_debug_reg_global): Change rs6000_vector_reload,
7384         reload_fpr_gpr, reload_gpr_vsx, reload_vsx_gpr uses to reg_addr.
7385         (rs6000_init_hard_regno_mode_ok): Likewise.
7386         (rs6000_secondary_reload_direct_move): Likewise.
7387         (rs6000_secondary_reload): Likewise.
7389         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add new
7390         constraints: wu, ww, and wy.  Repurpose wv constraint added during
7391         power8 changes.  Put wg constraint in alphabetical order.
7393         * config/rs6000/rs6000.opt (-mvsx-scalar-float): New debug switch
7394         for future work to add ISA 2.07 VSX single precision support.
7395         (-mvsx-scalar-double): Change default from -1 to 1, update
7396         documentation comment.
7397         (-mvsx-scalar-memory): Rename debug switch to -mupper-regs-df.
7398         (-mupper-regs-df): New debug switch to control whether DF values
7399         can go in the traditional Altivec registers.
7400         (-mupper-regs-sf): New debug switch to control whether SF values
7401         can go in the traditional Altivec registers.
7403         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print wu, ww,
7404         and wy constraints.
7405         (rs6000_init_hard_regno_mode_ok): Use ssize_t instead of int for
7406         loop variables.  Rename -mvsx-scalar-memory to -mupper-regs-df.
7407         Add new constraints, wu/ww/wy.  Repurpose wv constraint.
7408         (rs6000_debug_legitimate_address_p): Print if we are running
7409         before, during, or after reload.
7410         (rs6000_secondary_reload): Add a comment.
7411         (rs6000_opt_masks): Add -mupper-regs-df, -mupper-regs-sf.
7413         * config/rs6000/constraints.md (wa constraint): Sort w<x>
7414         constraints.  Update documentation string.
7415         (wd constraint): Likewise.
7416         (wf constraint): Likewise.
7417         (wg constraint): Likewise.
7418         (wn constraint): Likewise.
7419         (ws constraint): Likewise.
7420         (wt constraint): Likewise.
7421         (wx constraint): Likewise.
7422         (wz constraint): Likewise.
7423         (wu constraint): New constraint for ISA 2.07 SFmode scalar
7424         instructions.
7425         (ww constraint): Likewise.
7426         (wy constraint): Likewise.
7427         (wv constraint): Repurpose ISA 2.07 constraint that we did not use
7428         in the previous submissions.
7429         * doc/md.texi (PowerPC and IBM RS6000): Likewise.
7431 2013-09-23  Richard Sandiford  <rdsandiford@googlemail.com>
7433         * doc/rtl.texi (REG_NOTES): Say that int_list can also be used.
7434         (REG_BR_PROB): Say that the probability is stored in an int_list.
7435         * reg-notes.def: Update commentary to mention INT_LIST.
7436         * rtl.def (EXPR_LIST, INSN_LIST): Capitalize comments.
7437         (INT_LIST): New rtx.
7438         * rtl.h (add_int_reg_note, add_shallow_copy_of_reg_note): Declare.
7439         * rtlanal.c (int_reg_note_p): New function.
7440         (alloc_reg_note): Assert that the note does not have an int argument.
7441         (add_int_reg_note, add_shallow_copy_of_reg_note): New functions.
7442         * combine.c (distribute_notes): Use add_shallow_copy_of_rtx.
7443         * cse.c (cse_process_notes_1): Expect REG_EQUAL to be an EXPR_LIST
7444         rather than an INSN_LIST.  Handle INT_LIST.
7445         * ifcvt.c (cond_exec_process_insns): Take the probability as an int
7446         rather than an rtx.  Use gen_rtx_INT_LIST to create a REG_BR_PROB note.
7447         (cond_exec_process_if_block): Use XINT to extract REG_BR_PROB values.
7448         Manipulate them as ints rather than rtxes.
7449         * reg-stack.c (subst_asm_stack_regs): Only handle EXPR_LIST notes.
7450         * regmove.c (copy_src_to_dest): Likewise.
7451         * sched-vis.c (print_insn_with_notes): Handle INT_LIST.
7453         * config/i386/winnt.c (i386_pe_seh_unwind_emit): Sink pat assignment
7454         into the cases that need it.
7455         * config/arm/arm.c (arm_unwind_emit): Likewise.
7457         * asan.c (asan_clear_shadow): Use add_int_reg_note for REG_BR_PROB.
7458         * emit-rtl.c (try_split, emit_copy_of_insn_after): Likewise.
7459         * loop-doloop.c (add_test, doloop_modify): Likewise.
7460         * loop-unswitch.c (compare_and_jump_seq): Likewise.
7461         * optabs.c (emit_cmp_and_jump_insn_1): Likewise.
7462         * predict.c (combine_predictions_for_insn): Likewise.
7463         * print-rtl.c (print_rtx): Handle INT_LIST.
7464         * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Likewise.
7465         * config/alpha/alpha.c (emit_unlikely_jump): Likewise.
7466         * config/arm/arm.c (emit_unlikely_jump): Likewise.
7467         * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise.
7468         (ix86_split_fp_branch, predict_jump): Likewise.
7469         * config/rs6000/rs6000.c (emit_unlikely_jump): Likewise.
7470         * config/sh/sh.c (expand_cbranchsi4): Likewise.
7471         * config/spu/spu.c (ea_load_store_inline): Likewise.
7473         * cfgbuild.c (compute_outgoing_frequencies): Use XINT to access the
7474         value of a REG_BR_PROB note.
7475         * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
7476         (update_br_prob_note, rtl_verify_edges, purge_dead_edges): Likewise.
7477         * emit-rtl.c (try_split): Likewise.
7478         * predict.c (br_prob_note_reliable_p): Likewise.
7479         (invert_br_probabilities, combine_predictions_for_insn): Likewise.
7480         * reorg.c (mostly_true_jump): Likewise.
7481         * config/bfin/bfin.c (cbranch_predicted_taken_p): Likewise.
7482         * config/frv/frv.c (frv_print_operand_jump_hint): Likewise.
7483         * config/i386/i386.c (ix86_print_operand): Likewise.
7484         * config/ia64/ia64.c (ia64_print_operand): Likewise.
7485         * config/mmix/mmix.c (mmix_print_operand): Likewise.
7486         * config/rs6000/rs6000.c (output_cbranch): Likewise.
7487         * config/s390/s390.c (s390_expand_tbegin): Likewise.
7488         * config/sh/sh.c (sh_print_operand, sh_adjust_cost): Likewise.
7489         * config/sparc/sparc.c (output_cbranch): Likewise.
7490         * config/spu/spu.c (get_branch_target): Likewise.
7491         * config/tilegx/tilegx.c (cbranch_predicted_p): Likewise.
7492         * config/tilepro/tilepro.c (cbranch_predicted_p): Likewise.
7494 2013-09-23  Jan Hubicka  <jh@suse.cz>
7496         * ipa-cp.c (ipa_get_indirect_edge_target_1): Add sanity check
7497         for ipa-devirt.
7498         * ipa-utils.h (possible_polymorphic_call_target_p): New function.
7499         * ipa-devirt.c (possible_polymorphic_call_target_p): Be tolerant
7500         of external calls
7501         * gimple-fold.c: Include ipa-utils.h and gimple-pretty-print.h
7502         (gimple_fold_call): Dump inconsistent devirtualizations; add
7503         sanity check for type based devirtualizations.
7504         * ipa-prop.c: Include ipa-utils.h
7505         (ipa_intraprocedural_devirtualization): Add sanity check.
7506         (try_make_edge_direct_virtual_call): Likewise.
7508 2013-09-23  Eric Botcazou  <ebotcazou@adacore.com>
7510         * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Recurse on copy
7511         assignment statements.
7513 2013-09-23  Kugan Vivekanandarajah  <kuganv@linaro.org>
7515         * gimple-pretty-print.c (dump_ssaname_info): New function.
7516         (dump_gimple_phi): Call it.
7517         (pp_gimple_stmt_1): Likewise.
7518         * tree-core.h (tree_ssa_name): New union ssa_name_info_type field.
7519         (range_info_def): Declare.
7520         * tree-pretty-print.c (pp_double_int): New function.
7521         (dump_generic_node): Call it.
7522         * tree-pretty-print.h (pp_double_int): Declare.
7523         * tree-ssa-alias.c (dump_alias_info): Check pointer type.
7524         * tree-ssanames.h (range_info_def): New structure.
7525         (value_range_type): Move definition here.
7526         (set_range_info, value_range_type, duplicate_ssa_name_range_info):
7527         Declare.
7528         * tree-ssanames.c (make_ssa_name_fn): Check pointer type at
7529         initialization.
7530         (set_range_info): New function.
7531         (get_range_info): Likewise.
7532         (duplicate_ssa_name_range_info): Likewise.
7533         (duplicate_ssa_name_fn): Check pointer type and call
7534         duplicate_ssa_name_range_info.
7535         * tree-ssa-copy.c (fini_copy_prop): Likewise.
7536         * tree-vrp.c (value_range_type): Remove definition, now in
7537         tree-ssanames.h.
7538         (vrp_finalize): Call set_range_info to update value range of SSA_NAMEs.
7539         * tree.h (SSA_NAME_PTR_INFO): Macro changed to access via union.
7540         (SSA_NAME_RANGE_INFO): New macro.
7542 2013-09-23  Richard Biener  <rguenther@suse.de>
7544         PR tree-optimization/58464
7545         * tree-ssa-pre.c (phi_trans_lookup): Remove.
7546         (phi_trans_add): Change to add conditionally on being not present.
7547         (phi_translate_1): Remove recursion detection here.
7548         (phi_translate): Pre-seed the cache with NULL to catch
7549         recursion here in a more generic way.
7550         (bitmap_find_leader): Adjust comment.
7551         (get_representative_for): Dump value-numbers.
7552         (create_expression_by_pieces): Likewise.
7553         (insert_into_preds_of_block): Likewise.
7555 2013-09-23  Christian Bruel  <christian.bruel@st.com>
7557         PR target/58475
7558         * config/sh/sh.md (movsf_ie): Allow fpul_operand.
7559         * config/sh/predicate.md (arith_reg_operand): Disallow FPUL_REG.
7561 2013-09-23  James Greenhalgh  <james.greenhalgh@arm.com>
7563         Revert r202780:
7564         2013-09-20  Renlin Li  <renlin.li@arm.com>
7566         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
7567         plus_constant.
7568         (aarch64_expand_epilogue): Likewise.
7569         (aarch64_legitimize_reload_address): Likewise.
7571 2013-09-22  Eric Botcazou  <ebotcazou@adacore.com>
7573         * gimplify.c (gimplify_asm_expr): Reset the TREE_CHAIN of clobbers to
7574         NULL_TREE before pushing them onto the vector.  Likewise for labels.
7576 2013-09-21  Eric Botcazou  <ebotcazou@adacore.com>
7578         * config/ia64/predicates.md (ia64_cbranch_operator): Accept unordered
7579         comparison operators when -fno-trapping-math is in effect.
7580         * config/ia64/ia64.c (ia64_expand_compare): Add support for unordered
7581         comparison operators in TFmode and assert that unsupported operators
7582         cannot reach here.
7583         (ia64_print_operand): Likewise.
7585 2013-09-21  Jan Hubicka  <jh@suse.cz>
7587         * x86-tune.def (partial_reg_stall): Disable for CoreI7 and newer.
7588         (sse_typeless_stores): Enable for core
7589         (sse_load0_by_pxor): Likewise.
7590         (four_jump_limit): Disable for core.
7591         (pad_returns): Likewise.
7592         (avoid_vector_decode): Likewise.
7593         (fuse_cmp_and_branch): Enable for cores.
7594         * i386.c (x86_accumulate_outgoing_args): Disable for cores.
7596 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
7598         PR middle-end/56791
7599         * config/pa/pa.c (pa_option_override): Disable auto increment and
7600         decrement instructions until reload is completed.
7602         * config/pa/pa-linux.h (TARGET_OS_CPP_BUILTINS): Define
7603         __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1, __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2,
7604         and __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
7606 2013-09-20  DJ Delorie  <dj@redhat.com>
7607             Nick Clifton  <nickc@redhat.com>
7609         * config/rl78/rl78.c: Various whitespace and comment tweaks.
7610         (need_to_save): Save bank 0 on interrupts.
7611         (characterize_address): Strip far address wrappers.
7612         (rl78_as_legitimate_address): Likewise.
7613         (transcode_memory_rtx): Likewise.
7614         (rl78_peep_movhi_p): Disable this peephole after devirt.
7615         (rl78_propogate_register_origins): Forget all origins when a
7616         CLOBBER is seen.
7617         * config/rl78/rl78-virt.md: Various whitespace tweaks.
7618         * config/rl78/rl78-real.md: Various whitespace tweaks.  Additional
7619         peephole2's.
7620         * config/rl78/rl78.md (sel_rb): Disable for G10 just in case.
7621         * config/rl78/rl78-expand.md (movqi): Check for subregs of consts.
7622         * config/rl78/rl78.h (LINK_SPEC): Pass -gc-sections unless
7623         relocating.
7624         * config/rl78/constraints.md: Various whitespace and paren tweaks.
7626 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
7628         * config/pa/pa.md: In "scc" insn patterns, change output template to
7629         handle const0_rtx in reg_or_0_operand operands.
7631 2013-09-20  Martin Husemann  <martin@NetBSD.org>
7633         PR target/56875
7634         * config/vax/vax.c (vax_output_int_move): Use D format specifier.
7635         * config/vax/vax.md (ashldi3, <unnamed>): Ditto.
7637 2013-09-20  Richard Biener  <rguenther@suse.de>
7639         PR middle-end/58484
7640         * tree-scalar-evolution.c (struct scev_info_str): Shrink by
7641         remembering SSA name version and block index.
7642         (new_scev_info_str): Adjust.
7643         (hash_scev_info): Likewise.  Also hash the block index.
7644         (eq_scev_info): Adjust.
7645         (find_var_scev_info): Likewise.
7646         (struct instantiate_cache_entry): Remove.
7647         (struct instantiate_cache_type): Use a htab to map name, block
7648         to chrec.
7649         (instantiate_cache_type::~instantiate_cache_type): Adjust.
7650         (get_instantiated_value_entry): Likewise.
7651         (hash_idx_scev_info, eq_idx_scev_info): New functions.
7652         (instantiate_scev_name): Adjust.
7654 2013-09-20  Jeff Law  <law@redhat.com>
7656         * tree-ssa-dom.c (record_temporary_equivalences): Add comment.
7658 2013-09-20  Yufeng Zhang  <yufeng.zhang@arm.com>
7660         * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7661         Call aarch64_simd_expand_args to update op[argc].
7663 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
7665         * plugin.c (parse_plugin_arg_opt): Accept equal sign inside
7666         plugin argument.
7668 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
7670         * gengtype.c (file_rules): Added rule for *.cc files.
7671         (get_output_file_with_visibility): Give fatal message when no
7672         rules found.
7674 2013-09-20  Renlin Li  <renlin.li@arm.com>
7676         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
7677         (aarch64_expand_epilogue): Likewise.
7678         (aarch64_legitimize_reload_address): Likewise.
7680 2013-09-20  Bernd Edlinger  <bernd.edlinger@hotmail.de>
7682         PR middle-end/57748
7683         * expr.c (expand_assignment): Remove misalignp code path.
7685 2013-09-20  Marek Polacek  <polacek@redhat.com>
7687         PR sanitizer/58413
7688         * ubsan.c (get_ubsan_type_info_for_type): Use TYPE_SIZE instead of
7689         TYPE_PRECISION.  Add asserts.
7691 2013-09-20  Richard Biener  <rguenther@suse.de>
7693         PR tree-optimization/58453
7694         * tree-loop-distribution.c (distribute_loop): Apply the cost
7695         model for -ftree-loop-distribute-patterns, too.
7697 2013-09-20  Richard Biener  <rguenther@suse.de>
7699         PR middle-end/58473
7700         * tree-chrec.h (build_polynomial_chrec): Use gcc_checking_assert,
7701         make type comparison less strict.
7703 2013-09-20  Alan Modra  <amodra@gmail.com>
7705         * configure: Regenerate.
7706         * aclocal.m4: Regenerate.
7708 2013-09-20  Marek Polacek  <polacek@redhat.com>
7710         PR other/58467
7711         * doc/extend.texi: Document that attribute used is meant to be used
7712         on variables with static storage duration.
7714 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
7716         PR tree-optimization/58472
7717         * tree-vect-stmts.c (vectorizable_store, vectorizable_load): For
7718         simd_lane_access set inv_p = false.
7719         * omp-low.c (lower_rec_input_clauses): Set TREE_NO_WARNING on
7720         the simduid magic VAR_DECL.
7722 2013-09-19  Jan Hubicka  <jh@suse.cz>
7724         * i386.c (generic_memcpy, generic_memset): Fix 32bit template.
7726 2013-09-17  Jeff Law  <law@redhat.com>
7728         * tree-ssa-dom.c (record_temporary_equivalences): New function
7729         split out of dom_opt_dom_walker::after_dom_children.
7730         (dom_opt_dom_walker::thread_across_edge): Move common code
7731         in here from dom_opt_dom_walker::after_dom_children.
7732         (dom_opt_dom_walker::after_dom_children): Corresponding simplifictions.
7734 2013-09-19  Jan Hubicka  <jh@suse.cz>
7736         * i386.h (TARGET_GENERIC32, TARGET_GENERIC64): Remove.
7737         (TARGET_GENERIC): Use PROCESOR_GENERIC
7738         (enum processor_type): Unify generic32 and 64.
7739         * i386.md (cpu): Likewise.
7740         * x86-tune.def (use_leave): Enable for generic32.
7741         (avoid_vector_decode, slow_imul_imm32_mem, slow_imul_imm8): Likewise.
7742         * athlon.md: Change generic64 to generic in all occurences.
7743         * i386-c.c (ix86_target_macros_internal): Unify generic64 and 32.
7744         (ix86_target_macros_internal): Likewise.
7745         * driver-i386.c (host_detect_local_cpu): Likewise.
7746         * i386.c (generic64_memcpy, generic64_memset, generic64_cost): Rename
7747         to ..
7748         (generic_memcpy, generic_memset, generic_cost): This one.
7749         (generic32_memcpy, generic32_memset, generic32_cost): Remove.
7750         (m_GENERIC32, m_GENERIC64): Remove.
7751         (m_GENERIC): Turn into one flag.
7752         (processor_target): Unify generic tunnings.
7753         (ix86_option_override_internal): Replace generic32/64 by generic.
7754         (ix86_issue_rate): Likewise.
7755         (ix86_adjust_cost): Likewise.
7757 2013-09-19  Jan Hubicka  <jh@suse.cz>
7759         * cgraph.c (cgraph_create_edge_1): Avoid uninitialized read
7760         of speculative flag.
7762 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
7764         * omp-low.c (expand_omp_sections): Always pass len - 1 to
7765         GOMP_sections_start, even if !exit_reachable.
7767 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
7769         * lra-constraints.c (need_for_all_save_p): Use macro
7770         HARD_REGNO_CALL_PART_CLOBBERED.
7771         * lra-lives.c (check_pseudos_live_through_calls): Use the macro to
7772         set up pseudo conflict hard regs.
7774 2013-09-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
7776         PR target/58452
7777         * config/rs6000/paired.md (movmisalignv2sf): Fix to allow memory
7778         operands.
7780 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
7782         PR rtl-optimization/58438
7783         * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
7784         * lra-constraints.c (undo_optional_reloads): Keep optional reloads
7785         from previous subpasses.
7787 2013-09-18  Richard Earnshaw  <rearnsha@arm.com>
7789         * arm.c (arm_get_frame_offsets): Validate architecture supports
7790         LDRD/STRD before accepting the tuning preference.
7791         (arm_expand_prologue): Likewise.
7792         (arm_expand_epilogue): Likewise.
7794 2013-09-18  Richard Biener  <rguenther@suse.de>
7796         PR tree-optimization/58417
7797         * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
7798         have chrecs with symbols defined in the loop as operands.
7799         (chrec_fold_multiply): Likewise.
7800         * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
7801         parameters before folding binary operations.
7802         (struct instantiate_cache_entry_hasher): Remove.
7803         (struct instantiate_cache_type): Use a pointer-map.
7804         (instantiate_cache_type::instantiate_cache_type): New function.
7805         (instantiate_cache_type::get): Likewise.
7806         (instantiate_cache_type::set): Likewise.
7807         (instantiate_cache_type::~instantiate_cache_type): Adjust.
7808         (get_instantiated_value_entry): Likewise.
7809         (global_cache): New global.
7810         (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
7811         instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
7812         instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
7813         (instantiate_scev_name): Adjust.
7814         (instantiate_scev): Construct global instead of local cache.
7815         (resolve_mixers): Likewise.
7817 2013-09-18  Daniel Morris  <danielm@ecoscentric.com>
7818             Paolo Carlini  <paolo.carlini@oracle.com>
7820         PR c++/58458
7821         * doc/implement-cxx.texi: Fix references to the C++ standards.
7823 2013-09-18  Jakub Jelinek  <jakub@redhat.com>
7825         * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
7826         * tree-vect-data-refs.c (vect_analyze_data_refs): For
7827         simd_lane_access drs, update also DR_ALIGNED_TO.
7829 2013-09-18  Marek Polacek  <polacek@redhat.com>
7831         PR sanitizer/58411
7832         * doc/extend.texi: Document no_sanitize_undefined attribute.
7833         * builtins.c (fold_builtin_0): Don't sanitize function if it has the
7834         no_sanitize_undefined attribute.
7836 2013-09-18  Nick Clifton  <nickc@redhat.com>
7838         * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
7839         (ASM_DECLARE_FUNCTION_NAME): Define.
7841 2013-09-17  Trevor Saunders  <tsaunders@mozilla.com>
7843         * compare-elim.c (find_comparison_dom_walker): New class
7844         (find_comparisons_in_bb): Rename to
7845         find_comparison_dom_walker::before_dom_children
7846         (find_comparisons): Adjust
7847         * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
7848         adjust.
7849         (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
7850         * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
7851         (init_walk_dominator_tree): Remove declaration.
7852         (fini_walk_dominator_tree): Remove declaration.
7853         * fwprop.c (single_def_use_dom_walker): New class
7854         (single_def_use_enter_block): Convert to
7855         single_def_use_dom_walker::before_dom_children.
7856         (single_def_use_leave_block): Convert to
7857         single_def_use_dom_walker::after_dom_children.
7858         (build_single_def_use_links): Adjust.
7859         * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
7860         class.
7861         (find_candidates_in_block): Convert to
7862         find_candidates_dom_walker::before_dom_children.
7863         (execute_strength_reduction): Adjust.
7864         * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
7865         (sese_dom_walker): New class.
7866         (sese_dom_walker::sese_dom_walker): New constructor.
7867         (sese_dom_walker::~sese_dom_walker): New destructor.
7868         (build_sese_conditions_before): Convert to
7869         sese_dom_walker::before_dom_children.
7870         (build_sese_conditions_after): Convert to
7871         sese_dom_walker::after_dom_children.
7872         (build_poly_scop): Adjust
7873         * tree-into-ssa.c (rewrite_dom_walker): New class
7874         (rewrite_enter_block): Convert to
7875         rewrite_dom_walker::before_dom_children.
7876         (rewrite_leave_block): Convert to
7877         rewrite_dom_walker::after_dom_children.
7878         (rewrite_update_dom_walker): New class.
7879         (rewrite_update_enter_block): Convert to
7880         rewrite_update_dom_walker::before_dom_children.
7881         (rewrite_update_leave_block): Convert to
7882         rewrite_update_dom_walker::after_dom_children.
7883         (rewrite_blocks, rewrite_into_ssa): Adjust.
7884         (mark_def_dom_walker): New class.
7885         (mark_def_dom_walker::mark_def_dom_walker): New constructor.
7886         (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
7887         (mark_def_sites_blocks): Convert to
7888         mark_def_dom_walker::before_dom_children.
7889         (mark_def_site_blocks): Remove.
7890         * tree-ssa-dom.c (dom_opt_dom_walker): New class.
7891         (tree_ssa_dominator_optimize): Adjust.
7892         (dom_thread_across_edge): Convert to method
7893         dom_opt_dom_walker::thread_across_edge.
7894         (dom_opt_enter_block): Convert to member function
7895         dom_opt_dom_walker::before_dom_children.
7896         (dom_opt_leave_block): Convert to member function
7897         dom_opt_dom_walker::after_dom_children.
7898         * tree-ssa-dse.c (dse_dom_walker): New class.
7899         (dse_enter_block): Convert to member function
7900         dse_dom_walker::before_dom_children.
7901         (tree_ssa_dse): Adjust.
7902         * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
7903         (determine_invariantness_stmt): Convert to method
7904         invariantness_dom_walker::before_dom_children.
7905         (determine_invariantness): Remove
7906         (move_computations_dom_walker): New class.
7907         (move_computations_stmt): Convert to method
7908         move_computations_dom_walker::before_dom_children.
7909         (move_computations, tree_ssa_lim): Adjust.
7910         * tree-ssa-phiopt.c (nontrapping_dom_walker): New class.
7911         (nt_init_block): Convert to method
7912         notrappping_dom_walker::before_dom_children.
7913         (nt_fini_block): Convert to method
7914         method nontrapping_dom_walker::after_dom_children.
7915         (get_non_trapping): Adjust.
7916         * tree-ssa-pre.c (eliminate_dom_walker): New class.
7917         (eliminate_bb): Convert to method
7918         eliminate_dom_walker::before_dom_children.
7919         (eliminate_leave_block): Convert to method
7920         eliminate_dom_walker::after_dom_children.
7921         (eliminate): Adjust.
7922         * tree-ssa-strlen.c (strlen_dom_walker): New class.
7923         (strlen_enter_block): Convert to method
7924         strlen_dom_walker::before_dom_children.
7925         (strlen_leave_block): Convert to method
7926         method strlen_dom_walker::after_dom_children.
7927         (tree_ssa_strlen): Adjust.
7928         * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
7929         (tree_ssa_uncprop): Adjust.
7930         (uncprop_leave_block): Convert to method
7931         uncprop_dom_walker::after_dom_children.
7932         (uncprop_leave_block): Convert to method
7933         uncprop_dom_walker::before_dom_children.
7935 2013-09-18  Bin Cheng  <bin.cheng@arm.com>
7937         * config/arm/arm.c (thumb1_reorg): Search for flag setting insn before
7938         branch in same basic block.  Check both src and dest of the move insn.
7940 2013-09-17  Nick Clifton  <nickc@redhat.com>
7942         * config/rl78/rl78-real.md (bf): New pattern.
7943         (bt): New pattern.
7944         * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
7945         (rl78_print_operand): Do not put a # before a %B.
7946         * config/rl78/rl78.opt: Tweak doc strings.
7948 2013-09-17  DJ Delorie  <dj@redhat.com>
7950         * config/rl78/constraints.md (Wcv): Allow up to $r31.
7951         * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
7952         (rl78_option_override): Likewise, if -mallregs.
7953         (is_virtual_register): Likewise.
7954         * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
7955         (REGNO_OK_FOR_BASE_P): Likewise.
7956         * config/rl78/rl78.opt (-mallregs): New.
7958 2013-09-17  Nick Clifton  <nickc@redhat.com>
7960         * config/rl78/rl78.c (need_to_save): Change return type to bool.
7961         For interrupt functions: save all call clobbered registers if the
7962         interrupt handler is not a leaf function.
7963         (rl78_expand_prologue): Always recompute the frame information.
7964         For interrupt functions: only select bank 0 if one of the bank 0
7965         registers is going to be psuhed.
7967 2013-09-17  DJ Delorie  <dj@redhat.com>
7969         * config/rl78/constraints.md: For each W* constraint, rename to C*
7970         and create a W* constraint that checks for an optional ES: prefix
7971         pattern also.
7972         * config/rl78/rl78.md (UNS_ES_ADDR): New.
7973         (es_addr): New.  Used to wrap far addresses.
7974         * config/rl78/rl78-protos.h (rl78_es_addr): New.
7975         (rl78_es_base): New.
7976         * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
7977         wrapped far addresses.
7978         (rl78_print_operand_1): Unwrap far addresses before processing.
7979         (rl78_lo16): Wrap far addresses in unspecs.
7980         (rl78_es_addr): New.
7981         (rl78_es_base): New.
7982         (insn_ok_now): Check for not-yet-wrapped far addresses.
7983         (transcode_memory_rtx): Properly re-wrap far addresses.
7985 2013-09-17  Sebastian Huber  <sebastian.huber@embedded-brains.de>
7987         * config/sparc/t-rtems: Add leon3 multilibs.
7989 2013-09-17  Cong Hou  <congh@google.com>
7991         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
7992         when checking the dot production pattern. The type of rhs operand
7993         of multiply is now checked correctly.
7995 2013-09-17  Jeff Law  <law@redhat.com>
7997         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
7998         edge implied equivalences into successor phis.
7999         * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
8000         here from tree-ssa-threadedge.c.
8001         (mark_threaded_blocks): When threading through a joiner, if both
8002         successors of the joiner's clone reach the same block, verify the
8003         PHI arguments are equal.  If not, cancel the jump threading request.
8004         * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
8005         tree-ssa-threadupdate.c
8006         (thread_across_edge): Don't check PHI argument equality when
8007         threading through joiner block here.
8009 2013-09-17  Andrew MacLeod <amacleod@redhat.com>
8011         * tree-flow.h (ssa_undefined_value_p): Remove prototype.
8012         * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
8013         (warn_uninit, warn_uninitialized_vars,
8014         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
8015         Move to tree-ssa-uninit.c.
8016         * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c.
8017         (has_undefined_value_p): New.  Pass dependant parts of
8018         ssa_undefined_value_p.
8019         (uninit_undefined_value_p): Use has_undefined_value_p.
8020         (warn_uninit, warn_uninitialized_vars,
8021         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
8022         Move from tree-ssa.c.
8023         * tree-ssa.h: Adjust prototypes.
8025 2013-09-17  Jan Hubicka  <jh@suse.cz>
8027         PR middle-end/58332
8028         * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
8029         * ipa-inline.c (can_inline_edge_p): Do not downgrade
8030         FUNCTION_NOT_OPTIMIZED.
8031         * ipa-inline-analysis.c (compute_inline_parameters): Function
8032         not optimized is not inlinable unless it is alwaysinline.
8033         (inline_analyze_function): Force calls in not optimized
8034         function not inlinable.
8036 2013-09-17  Jan Hubicka  <jh@suse.cz>
8038         PR middle-end/58329
8039         * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
8040         to return NULL.
8041         * ipa.c (function_and_variable_visibility): Likewise.
8042         * ipa-profile.c (ipa_profile): Likewise.
8044 2013-09-17  Bernd Edlinger  <bernd.edlinger@hotmail.de>
8046         PR ipa/58398
8047         * cgraph.c (cgraph_function_body_availability): Check for ifunc
8048         attribute, and don't inline the resolver in this case.
8050 2013-09-17  Teresa Johnson  <tejohnson@google.com>
8052         * coverage.c (get_coverage_counts): Add missing newline.
8054 2013-09-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8056         PR tree-optimization/58088
8057         * fold-const.c (mask_with_trailing_zeros): New function.
8058         (fold_binary_loc): Make sure we don't recurse infinitely
8059         when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
8060         Use mask_with_trailing_zeros where appropriate.
8062 2013-09-17  Yuri Rumyantsev  <ysrumyan@gmail.com>
8064         * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
8065         of 'prev' var to get better distance estimation.
8067 2013-09-17  Eric Botcazou  <ebotcazou@adacore.com>
8069         * tree-inline.h (struct copy_body_data): Add transform_parameter.
8070         * tree-inline.c (is_parameter_of): New predicate.
8071         (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
8072         a parameter has been remapped.
8073         (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
8074         (optimize_inline_calls): Initialize transform_parameter.
8075         (copy_gimple_seq_and_replace_locals): Likewise.
8076         (tree_function_versioning): Likewise.
8077         (maybe_inline_call_in_expr): Likewise.
8079 2013-09-17  Nick Clifton  <nickc@redhat.com>
8081         * config/msp430/msp430-protos.h: Add prototypes for new functions.
8082         * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
8083         interrupt handlers.
8084         (is_attr_func): New function.
8085         (msp430_is_interrupt_func): New function.
8086         (is_naked_func): New function.
8087         (is_reentrant_func): New function.
8088         (is_critical_func): New function.
8089         (msp430_start_function): Add annotations for function attributes.
8090         (msp430_attr): New function.
8091         (msp430_attribute_table): New.
8092         (msp430_function_section): New function.
8093         (TARGET_ASM_FUNCTION_SECTION): Define.
8094         (msp430_builtin): New enum.
8095         (msp430_init_builtins): New function.
8096         (msp430_builtin_devl): New function.
8097         (msp430_expand_builtin): New function.
8098         (TARGET_INIT_BUILTINS): Define.
8099         (TARGET_EXPAND_BUILTINS): Define.
8100         (TARGET_BUILTIN_DECL): Define.
8101         (msp430_expand_prologue): Add support for naked, interrupt,
8102         critical and reentrant functions.
8103         (msp430_expand_epilogue): Likewise.
8104         (msp430_print_operand): Handle 'O' character.
8105         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
8106         NO_TRAMPOLINES.
8107         * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
8108         UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
8109         (pushm): Use a 'n' rather than an 'i' constraint.
8110         (msp_return): Add generation of the interrupt return instruction.
8111         (disable_interrupts): New pattern.
8112         (enable_interrupts): New pattern.
8113         (push_intr_state): New pattern.
8114         (pop_intr_state): New pattern.
8115         (bic_SR): New pattern.
8116         (bis_SR): New pattern.
8117         * doc/extend.texi: Document MSP430 function attributes and builtin
8118         functions.
8120 2013-09-17  Richard Biener  <rguenther@suse.de>
8122         PR tree-optimization/58432
8123         * tree-loop-distribution.c (tree_loop_distribution): Also
8124         scan PHIs for outside loop uses and seed a partition from them.
8126 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
8128         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
8129         (restructure_reference): Call backtrace_base_for_ref.
8131 2013-09-17  Alan Modra  <amodra@gmail.com>
8133         PR target/57589
8134         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11
8135         patch.
8137 2013-09-16  DJ Delorie  <dj@redhat.com>
8139         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
8140         vregs location for RL78/G10.
8141         (rl78_expand_prologue): Avoid SEL on G10.
8142         (rl78_expand_epilogue): Likewise.
8143         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
8144         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
8145         __RL78_G10__ when appropriate.
8146         (ASM_SPEC): Pass -mg10 along to the assembler.
8147         * config/rl78/rl78.md (sel_rb): Disable for G10.
8148         * config/rl78/rl78.opt: Add -mg10 option.
8149         * config/rl78/t-rl78: Add -mg10 multilib.
8151 2013-09-16  Xinliang David Li  <davidxl@google.com>
8153         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
8154         * omp-low.c (omp_max_vf): Ditto.
8155         (expand_omp_simd): Ditto.
8156         * tree-vectorizer.c (vectorize_loops): Ditto.
8157         (gate_vect_slp): Ditto.
8158         (gate_increase_alignment): Ditto.
8159         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
8160         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
8161         (gate_tree_vectorize): Name change.
8162         (tree_vectorize): Ditto.
8163         (pass_vectorize::gate): Call new function.
8164         (pass_vectorize::execute): Ditto.
8165         * opts.c: O3 default setting change.
8166         (finish_options): Check new flag.
8167         * doc/invoke.texi: Document new flags.
8168         * common.opt: New flags.
8170 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
8172         * doc/tm.texi.in (Cond Exec Macros): Remove node.
8173         (Condition Code): Don't reference it.
8174         * doc/tm.texi: Regenerate.
8176 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
8178         PR middle-end/58418
8179         * lra-constraints.c (undo_optional_reloads): Consider all optional
8180         reload even if it did not get a hard reg.
8182 2013-09-16  Teresa Johnson  <tejohnson@google.com>
8184         * dumpfile.c (dump_loc): Remove newline emission.
8185         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
8186         emission to dump_printf_loc calls where missing.
8187         (vect_mark_for_runtime_alias_test): Ditto.
8188         (vect_analyze_data_ref_dependence): Ditto.
8189         (vect_analyze_data_ref_dependences): Ditto.
8190         (vect_slp_analyze_data_ref_dependence): Ditto.
8191         (vect_slp_analyze_data_ref_dependences): Ditto.
8192         (vect_compute_data_ref_alignment): Ditto.
8193         (vect_update_misalignment_for_peel): Ditto.
8194         (vect_verify_datarefs_alignment): Ditto.
8195         (vector_alignment_reachable_p): Ditto.
8196         (vect_get_data_access_cost): Ditto.
8197         (vect_enhance_data_refs_alignment): Ditto.
8198         (vect_find_same_alignment_drs): Ditto.
8199         (vect_analyze_data_refs_alignment): Ditto.
8200         (vect_analyze_group_access): Ditto.
8201         (vect_analyze_data_ref_access): Ditto.
8202         (vect_analyze_data_ref_accesses): Ditto.
8203         (vect_prune_runtime_alias_test_list): Ditto.
8204         (vect_analyze_data_refs): Ditto.
8205         (vect_create_addr_base_for_vector_ref): Ditto.
8206         (vect_create_data_ref_ptr): Ditto.
8207         (vect_grouped_store_supported): Ditto.
8208         (vect_grouped_load_supported): Ditto.
8209         * value-prof.c (check_counter): Ditto.
8210         (check_ic_target): Ditto.
8211         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
8212         (vect_recog_widen_mult_pattern): Ditto.
8213         (vect_recog_widen_sum_pattern): Ditto.
8214         (vect_recog_over_widening_pattern): Ditto.
8215         (vect_recog_widen_shift_pattern): Ditto.
8216         (vect_recog_rotate_pattern): Ditto.
8217         (vect_recog_vector_vector_shift_pattern): Ditto.
8218         (vect_recog_divmod_pattern): Ditto.
8219         (vect_recog_mixed_size_cond_pattern): Ditto.
8220         (vect_recog_bool_pattern): Ditto.
8221         (vect_pattern_recog_1): Ditto.
8222         (vect_pattern_recog): Ditto.
8223         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
8224         (vect_is_simple_iv_evolution): Ditto.
8225         (vect_analyze_scalar_cycles_1): Ditto.
8226         (vect_get_loop_niters): Ditto.
8227         (vect_analyze_loop_1): Ditto.
8228         (vect_analyze_loop_form): Ditto.
8229         (vect_analyze_loop_operations): Ditto.
8230         (vect_analyze_loop_2): Ditto.
8231         (vect_analyze_loop): Ditto.
8232         (report_vect_op): Ditto.
8233         (vect_is_slp_reduction): Ditto.
8234         (vect_is_simple_reduction_1): Ditto.
8235         (vect_get_known_peeling_cost): Ditto.
8236         (vect_estimate_min_profitable_iters): Ditto.
8237         (vect_model_reduction_cost): Ditto.
8238         (vect_model_induction_cost): Ditto.
8239         (get_initial_def_for_induction): Ditto.
8240         (vect_create_epilog_for_reduction): Ditto.
8241         (vectorizable_reduction): Ditto.
8242         (vectorizable_induction): Ditto.
8243         (vectorizable_live_operation): Ditto.
8244         (vect_loop_kill_debug_uses): Ditto.
8245         (vect_transform_loop): Ditto.
8246         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
8247         (vect_stmt_relevant_p): Ditto.
8248         (process_use): Ditto.
8249         (vect_mark_stmts_to_be_vectorized): Ditto.
8250         (vect_model_simple_cost): Ditto.
8251         (vect_model_promotion_demotion_cost): Ditto.
8252         (vect_model_store_cost): Ditto.
8253         (vect_get_store_cost): Ditto.
8254         (vect_model_load_cost): Ditto.
8255         (vect_get_load_cost): Ditto.
8256         (vect_init_vector_1): Ditto.
8257         (vect_get_vec_def_for_operand): Ditto.
8258         (vect_finish_stmt_generation): Ditto.
8259         (vectorizable_call): Ditto.
8260         (vectorizable_conversion): Ditto.
8261         (vectorizable_assignment): Ditto.
8262         (vectorizable_shift): Ditto.
8263         (vectorizable_operation): Ditto.
8264         (vectorizable_store): Ditto.
8265         (vectorizable_load): Ditto.
8266         (vectorizable_condition): Ditto.
8267         (vect_analyze_stmt): Ditto.
8268         (vect_transform_stmt): Ditto.
8269         (vect_is_simple_use): Ditto.
8270         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
8271         (vect_can_advance_ivs_p): Ditto.
8272         (vect_update_ivs_after_vectorizer): Ditto.
8273         (vect_do_peeling_for_loop_bound): Ditto.
8274         (vect_gen_niters_for_prolog_loop): Ditto.
8275         (vect_update_inits_of_drs): Ditto.
8276         (vect_create_cond_for_alias_checks): Ditto.
8277         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
8278         (vect_build_slp_tree_1): Ditto.
8279         (vect_supported_load_permutation_p): Ditto.
8280         (vect_analyze_slp_instance): Ditto.
8281         (vect_analyze_slp): Ditto.
8282         (vect_make_slp_decision): Ditto.
8283         (vect_detect_hybrid_slp): Ditto.
8284         (vect_bb_vectorization_profitable_p): Ditto.
8285         (vect_slp_analyze_bb_1): Ditto.
8286         (vect_update_slp_costs_according_to_vf): Ditto.
8287         (vect_get_mask_element): Ditto.
8288         (vect_transform_slp_perm_load): Ditto.
8289         (vect_schedule_slp_instance): Ditto.
8290         (vect_schedule_slp): Ditto.
8291         (vect_slp_transform_bb): Ditto.
8292         * profile.c (read_profile_edge_counts): Ditto.
8293         (compute_branch_probabilities): Ditto.
8294         * coverage.c (get_coverage_counts): Ditto.
8296 2013-09-16  Diego Novillo  <dnovillo@google.com>
8298         * tree-core.h: Add missing comment lines from refactoring of tree.h.
8300 2013-09-16  Jan Hubicka  <jh@suse.cz>
8302         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
8303         abstract functions; for static functions check the presence of body.
8305 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
8307         * config/aarch64/aarch64-simd-builtins.def (fma): New.
8308         * config/aarch64/aarch64-simd.md
8309         (aarch64_mla_elt<mode>): New.
8310         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
8311         (aarch64_mls_elt<mode>): Likewise.
8312         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
8313         (aarch64_fma4_elt<mode>): Likewise.
8314         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
8315         (aarch64_fma4_elt_to_128v2df): Likewise.
8316         (aarch64_fma4_elt_to_64df): Likewise.
8317         (fnma<mode>4): Likewise.
8318         (aarch64_fnma4_elt<mode>): Likewise.
8319         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8320         (aarch64_fnma4_elt_to_128v2df): Likewise.
8321         (aarch64_fnma4_elt_to_64df): Likewise.
8322         * config/aarch64/iterators.md (VDQSF): New.
8323         * config/aarch64/arm_neon.h
8324         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
8325         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
8327 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
8329         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
8330         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8331         (aarch64_mul3_elt_to_128df): Likewise.
8332         (aarch64_mul3_elt_to_64v2df): Likewise.
8333         * config/aarch64/iterators.md (VEL): Also handle DFmode.
8334         (VMUL): New.
8335         (VMUL_CHANGE_NLANES) Likewise.
8336         (h_con): Likewise.
8337         (f): Likewise.
8338         * config/aarch64/arm_neon.h
8339         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
8341 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
8343         * config/aarch64/arm_neon.h
8344         (vcvtx_high_f32_f64): Fix parameters.
8346 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
8347             Uros Bizjak  <ubizjak@gmail.com>
8349         * config/alpha.c: Include tree-ssanames.h.
8351 2013-09-16  Richard Biener  <rguenther@suse.de>
8353         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
8354         (dot_rdg_1): Handle control_dd.
8355         (create_edge_for_control_dependence): New function.
8356         (create_rdg_edges): Add control dependences if asked for.
8357         (build_rdg): Likewise.
8358         (generate_loops_for_partition): If there are not necessary
8359         control stmts remove all their dependencies.
8360         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
8361         (distribute_loop): Pass on control dependences.
8362         (tree_loop_distribution): Compute control dependences and remove
8363         restriction on number of loop nodes.
8365 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
8367         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
8368         for internal calls.
8370 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
8372         * cse.c (try_const_anchors): Punt on CC modes.
8374 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
8376         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
8378 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
8380         PR target/58382
8381         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
8382         calls to word_mode.
8384 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
8386         PR target/48094
8387         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
8388         seen.
8389         (darwin_objc1_section): Likewise.
8390         (darwin_file_end): Emit Image Info section when required.
8392 2013-09-14  Jan Hubicka  <jh@suse.cz>
8394         * tree-into-ssa.c (gate_into_ssa): New.
8395         (pass_data_build_ssa): Use it.
8396         * cgraph.h (expand_thunk): Update prototype.
8397         * cgraphunit.c (analyze_function): Expand thunks early.
8398         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
8399         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
8400         set lowered flag; do not add new function.
8401         (assemble_thunks_and_aliases): Update.
8402         * tree-ssa.c (gate_init_datastructures): New gate.
8403         (pass_data_init_datastructures): Use it.
8405 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
8407         PR target/58269
8408         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
8409         xmm register set described in the psABI.
8411 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
8413         * dwarf2out.c (should_emit_struct_debug): Add check
8414         for type_decl variable is not NULL.
8416 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
8418         * config.gcc: Use new winnt-c.c target hooks
8419         * config/t-winnt: New file
8420         * config/winnt-c.c: New file
8421         * doc/tm.texi.in: Document new hook
8422         * doc/tm.texi: Regenerated
8424 2013-09-13  Jan Hubicka  <jh@suse.cz>
8426         PR middle-end/58094
8427         * ipa-inline.c (check_callers): New function.
8428         (check_caller_edge): Remove.
8429         (want_inline_function_to_all_callers_p): Also permit alises that are
8430         called dirrectly.
8431         (inline_to_all_callers): Terminate the walk when devirtualization
8432         introduce new calls.
8434 2013-09-13  Jan Hubicka  <jh@suse.cz>
8436         * ipa-inline-analysis.c (struct growth_data): Add node.
8437         (do_estimate_growth_1): Fix detection of recursion.
8439 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
8441         PR tree-optimization/58392
8442         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
8443         to avoid shadowing of outer loop variable.  If
8444         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
8445         replace_by_duplicate_decl simduid of loops that have it set and
8446         set dest_cfun->has_simduid_loops and/or
8447         dest_cfun->has_force_vect_loops.
8448         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
8449         instead of maybe_lookup_decl.
8450         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
8451         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
8452         call.  Copy over force_vect and copy and remap simduid.  Set
8453         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
8454         (copy_cfg_body): Remove blocks_to_copy argument.  Use
8455         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
8456         caller.  Don't set cfun->has_simduid_loops and/or
8457         cfun->has_force_vect_loops here.
8458         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
8459         caller.
8460         (expand_call_inline, tree_function_versioning): Adjust copy_body
8461         callers.
8463 2013-09-13  Martin Jambor  <mjambor@suse.cz>
8465         PR bootstrap/58388
8466         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
8467         the assert if the edge was a speculative one.
8469 2013-09-13  Richard Biener  <rguenther@suse.de>
8471         * tree-data-ref.h (known_dependences_p): Move here ...
8472         * tree-loop-distribution.c (known_dependences_p): ... from here.
8473         (dump_rdg_component, debug_rdg_component): Remove.
8474         (dump_rdg): Adjust.
8475         (generate_loops_for_partition): Use gimple_uid instead of
8476         relying on matching stmt visit order.
8477         (rdg_build_partitions): Take starting stmt vector.
8478         (ldist_gen): Merge into ...
8479         (distribute_loop): ... this function.  Do not compute starting
8480         vertices vector.
8481         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
8483 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8485         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
8486         Set type attribute correctly. Set predicable_short_it attribute.
8487         (cmpsi_shiftsi): Remove %? from output template.
8489 2013-09-13  Richard Biener  <rguenther@suse.de>
8491         * tree-loop-distribution.c (struct rdg_component,
8492         rdg_defs_used_in_other_loops_p, free_rdg_components,
8493         rdg_build_components): Remove.
8494         (stmts_from_loop): Do not record virtual PHIs.
8495         (generate_loops_for_partition): Skip virtual PHIs.
8496         (build_rdg_partition_for_component): Rename to ...
8497         (build_rdg_partition_for_vertex): ... this and adjust.
8498         (rdg_build_partitions): Take a vector of starting vertices
8499         instead of components.  Remove unnecessary leftover handling.
8500         (ldist_gen): Do not build components or record other stores.
8501         (distribute_loop): Do not distribute loops containing stmts
8502         with side-effects.
8504 2013-09-13  Christian Bruel  <christian.bruel@st.com>
8506         PR target/58314
8507         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
8509 2013-09-13  Kai Tietz  <ktietz@redhat.com>
8511         * config.gcc: Separate cases for mingw and cygwin targets,
8512         and add 64-bit cygwin target case.
8514         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
8515         dll-export inline-functions.
8516         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
8518 2013-09-13  Jeff Law  <law@redhat.com>
8520         PR middle-end/58387
8521         Revert:
8522         2013-09-06  Jeff Law  <law@redhat.com>
8524         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
8525         edge implied equivalences into successor phis.
8527 2013-09-12  DJ Delorie  <dj@redhat.com>
8529         * config/rl78/rl78-virt.md: Change from | to \; for asm line
8530         separators.
8532 2013-09-12  Brooks Moses  <bmoses@google.com>
8534         PR driver/42955
8535         * Makefile.in: Do not install driver binaries in $(target)/bin.
8537 2013-09-12  DJ Delorie  <dj@redhat.com>
8539         * config/rl78/rl78.opt (mrelax): New.
8540         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
8541         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
8543         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
8544         between SP and FP.
8545         (rl78_expand_epilogue): Likewise.
8547 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
8549         PR middle-end/58335
8550         * lra-eliminations.c (remove_reg_equal_offset_note): New.
8551         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
8552         pointer elimination with using remove_reg_equal_offset_note.
8554 2013-09-12  DJ Delorie  <dj@redhat.com>
8556         * config/msp430/: New port.
8557         * config.gcc (msp430): Added.
8558         * doc/invoke.texi: Document MSP430 options.
8559         * doc/install.texi: Document msp430-elf
8560         * doc/md.texi: Document msp430-elf
8561         * doc/contrib.texi: Document msp430-elf
8563         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
8564         PSImode.
8566 2013-09-12  Martin Jambor  <mjambor@suse.cz>
8568         PR ipa/58389
8569         * ipa-prop.c (remove_described_reference): Give up if the edge in the
8570         reference descriptor is NULL.
8571         (ipa_edge_removal_hook): If owning a reference descriptor, set its
8572         edge to NULL.
8574 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
8576         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
8577         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
8578         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
8579         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
8580         make_temp_ssa_name): move to tree-ssanames.h
8581         * tree-ssa-alias.h: Move prototype.
8582         * tree-ssa.h: Include tree-ssanames.h.
8583         * tree-ssanames.c (FREE_SSANAMES): Move to here.
8584         * tree-ssanames.h: New.  Move items from tree-flow*.h
8585         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
8587 2013-09-12  Richard Biener  <rguenther@suse.de>
8589         PR tree-optimization/58404
8590         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
8591         propagate non-invariant addresses into dereferences wrapped
8592         in component references.
8594 2013-09-12  Richard Biener  <rguenther@suse.de>
8596         PR tree-optimization/58402
8597         * passes.def: Move pass_late_warn_uninitialized later.
8599 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
8601         * tree-ssa.h: New.  Move content from tree-flow.h and
8602         tree-flow-inline.h.
8603         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
8604         Move prototypes belonging to tree-ssa.c.
8605         * tree-flow-inline.h (redirect_edge_var_map_def,
8606         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
8607         tree-ssa.h.
8608         * gimple.h: Adjust prototypes.
8609         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
8610         to...
8611         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
8612         * tree.h: Move prototype to tree-ssa.h.
8613         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
8614         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
8615         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
8616         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
8617         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
8618         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
8619         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
8620         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
8621         graphite-blocking.c, graphite-clast-to-gimple.c,
8622         graphite-dependences.c, graphite-interchange.c,
8623         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
8624         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
8625         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
8626         ipa-reference.c, ipa-split.c, ipa-utils.c,
8627         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
8628         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
8629         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
8630         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
8631         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
8632         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
8633         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
8634         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
8635         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
8636         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
8637         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
8638         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
8639         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
8640         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
8641         value-prof.c, var-tracking.c,
8642         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
8644 2013-09-12  Richard Biener  <rguenther@suse.de>
8646         PR tree-optimization/58396
8647         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
8648         (build_rdg): Take a loop-nest parameter, fix memleaks.
8649         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
8651 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
8653         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
8654         for SLM.
8656 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
8658         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
8659         The FMA instruction names should have a 'v' prefix.
8661 2013-09-12  Richard Biener  <rguenther@suse.de>
8663         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
8664         (dot_rdg): Use popen instead of system in optional code.
8665         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
8666         (already_processed_vertex_p): Adjust.
8667         (has_anti_or_output_dependence, predecessor_has_mem_write,
8668         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
8669         rdg_flag_uses): Remove.
8670         (rdg_flag_vertex): Simplify.
8671         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
8672         remove recursion.
8673         (build_rdg_partition_for_component): Process the first vertex
8674         of a component only.
8675         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
8677 2013-09-12  Alan Modra  <amodra@gmail.com>
8679         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
8681 2013-09-11  DJ Delorie  <dj@redhat.com>
8682             Nick Clifton  <nickc@redhat.com>
8684         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
8685         (rl78_stack_based_mem): New.
8686         * config/rl78/constraints.md (Iv08): New.
8687         (Iv16): New.
8688         (Iv24): New.
8689         (Is09): New.
8690         (Is17): New.
8691         (Is25): New.
8692         (ISsi): New.
8693         (IShi): New.
8694         (ISqi): New.
8695         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
8696         (movhi): Likewise.
8697         (movsi): Change from expand to insn-and-split.
8698         (ashrsi3): Clobber AX.
8699         (lshrsi3): New.
8700         (ashlsi3): New.
8701         (cbranchsi4): New.
8702         * config/rl78/rl78.md (CC_REG): Fix.
8703         (addsi3): Allow memory and immediate operands.
8704         (addsi3_internal): Split into...
8705         (addsi3_internal_virt): ...new, and ...
8706         (addsi3_internal_real): ...new.
8707         (subsi): New.
8708         (subsi3_internal_virt): New.
8709         (subsi3_internal_real): New.
8710         (mulsi3): Add memory operand.
8711         (mulsi3_rl78): Likewise.
8712         (mulsi3_g13): Likewise.
8713         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
8714         (cbranchqi4_real): Add more constraint options.
8715         (cbranchhi4_real): Expand pattern.
8716         (cbranchhi4_real_signed): New.
8717         (cbranchhi4_real_inverted): New.
8718         (cbranchsi4_real_lt): New.
8719         (cbranchsi4_real_ge): New.
8720         (cbranchsi4_real_signed): New.
8721         (cbranchsi4_real): New.
8722         (peephole2): New.
8723         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
8724         constant shifts.
8725         (lshrsi3_virt): Likewise.
8726         (ashlsi3_virt): Likewise.
8727         (cbranchqi4_virt_signed): New.
8728         (cbranchhi4_virt_signed): New.
8729         (cbranchsi4_virt): New.
8730         * config/rl78/rl78.c: Whitespace fixes throughout.
8731         (move_elim_pass): New.
8732         (pass_data_rl78_move_elim): New.
8733         (pass_rl78_move_elim): New.
8734         (make_pass_rl78_move_elim): New.
8735         (rl78_devirt_info): Run devirt earlier.
8736         (rl78_move_elim_info): New.
8737         (rl78_asm_file_start): Register it.
8738         (rl78_split_movsi): New.
8739         (rl78_as_legitimate_address): Allow virtual base registers when
8740         appropriate.
8741         (rl78_addr_space_convert): Remove spurious debug stuff.
8742         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
8743         (rl78_print_operand): More cases for not printing '#'.
8744         (rl78_expand_compare): Remove most of the logic.
8745         (content_memory): New.
8746         (clear_content_memory): New.
8747         (get_content_index): New.
8748         (get_content_name): New.
8749         (display_content_memory): New.
8750         (update_content): New.
8751         (record_content): New.
8752         (already_contains): New.
8753         (insn_ok_now): Re-recog insns with virtual registers.
8754         (add_postponed_content_update): New.
8755         (process_postponed_content_update): New.
8756         (gen_and_emit_move): New.
8757         (transcode_memory_rtx): Record new location content.
8758         Use gen_and_emit_move.
8759         (force_into_acc): New.
8760         (move_to_acc): Use gen_and_emit_move.
8761         (move_from_acc): Likewise.
8762         (move_acc_to_reg): Likewise.
8763         (move_to_x): Likewise.
8764         (move_to_hl): Likewise.
8765         (move_to_de): Likewise.
8766         (rl78_alloc_physical_registers_op1): Record location content.
8767         (has_constraint): New.
8768         (rl78_alloc_physical_registers_op2): Record location content.
8769         Optimize use of HL.
8770         (rl78_alloc_physical_registers_ro1): Likewise.
8771         (rl78_alloc_physical_registers_cmp): Likewise.
8772         (rl78_alloc_physical_registers_umul): Likewise.
8773         (rl78_alloc_address_registers_macax): New.
8774         (rl78_alloc_physical_registers): Initialize and set location
8775         content memory as needed.
8776         (rl78_reorg): Make sure split2 is called.
8777         (rl78_rtx_costs): New.
8779 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
8781         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
8782         for (not (neg ...)) and (neg (not ...)) cases.
8784 2013-09-11  Richard Biener  <rguenther@suse.de>
8786         PR middle-end/58377
8787         * passes.def: Split critical edges before late uninit warning passes.
8788         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
8790 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
8792         PR tree-optimization/58385
8793         * fold-const.c (build_range_check): If both low and high are NULL,
8794         use omit_one_operand_loc to preserve exp side-effects.
8796 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8798         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
8800 2013-09-11  Richard Biener  <rguenther@suse.de>
8802         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
8803         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
8804         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
8805         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
8806         stmts_from_loop, known_dependences_p, build_empty_rdg,
8807         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
8808         * tree-loop-distribution.c: ... here.
8809         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
8810         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
8811         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
8812         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
8813         * tree-loop-distribution.c: ... here.
8814         * tree-loop-distribution.c: Include gimple-pretty-print.h.
8815         (struct partition_s): Add loops member.
8816         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
8817         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
8818         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
8820 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8821             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
8822             Sergey Lega  <sergey.s.lega@intel.com>
8823             Anna Tikhonova  <anna.tikhonova@intel.com>
8824             Ilya Tocar  <ilya.tocar@intel.com>
8825             Andrey Turetskiy  <andrey.turetskiy@intel.com>
8826             Ilya Verbin  <ilya.verbin@intel.com>
8827             Kirill Yukhin  <kirill.yukhin@intel.com>
8828             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
8830         * config/i386/constraints.md (k): New.
8831         (Yk): Ditto.
8832         * config/i386/i386.c (const regclass_map): Add new mask registers.
8833         (dbx_register_map): Ditto.
8834         (dbx64_register_map): Ditto.
8835         (svr4_dbx_register_map): Ditto.
8836         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
8837         disabled.
8838         (ix86_preferred_reload_class): Disable constants for mask registers.
8839         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
8840         (ix86_hard_regno_mode_ok): Support new mask registers.
8841         (x86_order_regs_for_local_alloc): Ditto.
8842         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
8843         (FIXED_REGISTERS): Add new mask registers.
8844         (CALL_USED_REGISTERS): Ditto.
8845         (REG_ALLOC_ORDER): Ditto.
8846         (VALID_MASK_REG_MODE): New.
8847         (FIRST_MASK_REG): Ditto.
8848         (LAST_MASK_REG): Ditto.
8849         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
8850         (MAYBE_MASK_CLASS_P): New.
8851         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
8852         (REG_CLASS_CONTENTS): Ditto.
8853         (MASK_REGNO_P): New.
8854         (ANY_MASK_REG_P): Ditto.
8855         (HI_REGISTER_NAMES): Add new mask registers.
8856         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
8857         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
8858         mask registers.
8859         (attribute "type"): Add mskmov, msklog.
8860         (attribute "length_immediate"): Support them.
8861         (attribute "memory"): Ditto.
8862         (attribute "prefix_0f"): Ditto.
8863         (*movhi_internal): Support new mask registers.
8864         (*movqi_internal): Ditto.
8865         (define_split): Split out clobber pattern is a logic
8866         insn on mask registers.
8867         (*k<logic><mode>): New.
8868         (*andhi_1): Extend to support mask regs.
8869         (*andqi_1): Extend to support mask regs.
8870         (kandn<mode>): New.
8871         (define_split): Split and-not to and and not if operands
8872         are not mask regs.
8873         (*<code><mode>_1): Separate HI mode to new pattern...
8874         (*<code>hi_1): This.
8875         (*<code>qi_1): Extend to support mask regs.
8876         (kxnor<mode>): New.
8877         (kortestzhi): Ditto.
8878         (kortestchi): Ditto.
8879         (kunpckhi): Ditto.
8880         (*one_cmpl<mode>2_1): Remove HImode and handle it...
8881         (*one_cmplhi2_1): ...Here, now with mask registers support.
8882         (*one_cmplqi2_1): Support new mask registers.
8883         (HI/QImode arithmetics splitter): Don't split if mask registers
8884         are used.
8885         (HI/QImode not splitter): Ditto.
8886         * config/i386/predicated.md (mask_reg_operand): New.
8887         (general_reg_operand): Ditto.
8889 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8891         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
8892         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
8894 2013-09-10  Jeff Law  <law@redhat.com>
8896         PR tree-optimization/58380
8897         * tree-ssa-threadupdate.c (thread_block): Recognize another case
8898         of threading through a buried loop header.
8900         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
8901         return value for single successor case.
8903 2013-09-10  Jan Hubicka  <jh@suse.cz>
8905         * ipa-devirt.c (ipa_devirt): Enable with LTO.
8907 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
8909         PR target/58361
8910         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
8911         support conditional execution.
8912         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
8914 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
8916         * lra.c (lra): Clear lra_optional_reload_pseudos before every
8917         constraint pass.
8918         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
8919         Check destination too to check move insn.
8920         (undo_optional_reloads): Add check that the original peudo did not
8921         changed its allocation and the optional reload was inherited on last
8922         inheritance pass.  Break loop after deciding to keep optional reload.
8923         (lra_undo_inheritance): Add check that inherited pseudo still in
8924         memory.
8926 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
8928         * config/aarch64/aarch64.md (generic_sched): New.
8929         * config/aarch64/aarch64-generic.md (load): Make conditional
8930         on generic_sched attribute.
8931         (nonload): Likewise.
8933 2013-09-10  Jan Hubicka  <jh@suse.cz>
8935         * lto-cgraph.c: Include ipa-utils.h.
8936         (compute_ltrans_boundary): Also add possible targets into the boundary.
8938 2013-09-10  Jan Hubicka  <jh@suse.cz>
8940         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
8941         VAR_DECL of vtable rather than full expression.
8943 2013-09-10  Jan Hubicka  <jh@suse.cz>
8944             Paolo Carlini  <paolo.carlini@oracle.com>
8946         * cgraphunit.c (analyze_functions): Save input_location, set it
8947         to UNKNOWN_LOCATION and restore it at the end.
8949 2013-09-10  Martin Jambor  <mjambor@suse.cz>
8951         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
8952         represented by a thunk.
8954 2013-09-10  Jeff Law  <law@redhat.com>
8956         PR tree-optimization/58343
8957         * tree-ssa-threadupdate.c (thread_block): Identify and disable
8958         jump threading requests through loop headers buried in the middle
8959         of a jump threading path.
8961         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
8962         in return value/type.
8964 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
8966         PR rtl-optimization/58365
8967         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
8968         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
8969         it differs.
8971 2013-09-10  Richard Biener  <rguenther@suse.de>
8973         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
8974         * tree-data-ref.c (create_rdg_vertices): Collect all data
8975         references, signal failure to the caller, use data-ref API.
8976         (build_rdg): Compute data references only once.  Maintain lifetime
8977         of data references and data dependences from within RDG.
8978         (free_rdg): Free dependence relations.
8979         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
8980         inventing extra dependences.
8981         (distribute_loop): Update for RDG API changes.
8983 2013-09-10  Kai Tietz  <ktietz@redhat.com>
8985         * doc/invoke.texi (fms-extensions): Document changed
8986         behavior for ms-abi targets.
8987         * config/i386/i386.c (ix86_option_override_internal):
8988         Set default value of option -fms-extension for ms-abi targets.
8990 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
8992         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
8994 2013-09-10  Alan Modra  <amodra@gmail.com>
8996         PR target/58330
8997         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
8999 2013-09-10  Alan Modra  <amodra@gmail.com>
9001         * config/rs6000/predicates.md (add_cint_operand): New.
9002         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
9003         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
9004         using add_cint_operand.
9005         (largetoc_high_plus_aix): Likewise.
9007 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
9009         PR tree-optimization/58364
9010         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
9011         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
9012         the current range can't be an unconditional true or false.
9014 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
9016         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
9018 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
9020         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
9022 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
9024         PR c++/43452
9025         * doc/invoke.texi (-Wdelete-incomplete): Document it.
9027 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
9029         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
9030         NO_REGS for immediate that can't be moved directly into FP_REGS.
9032 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9034         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
9035         comparison with negated operand.
9036         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
9037         RTL form.
9039 2013-09-09  Richard Biener  <rguenther@suse.de>
9041         PR middle-end/58326
9042         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
9043         of a subloop record all its block as affecting loop-closed SSA form.
9045 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
9047         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
9048         of an rtx/bitpos pair.
9049         (store_fixed_bit_field): Update accordingly.
9051 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
9053         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
9054         GEN_INT.
9055         * builtins.c (expand_errno_check): Likewise.
9056         * dwarf2cfi.c (init_return_column_size): Likewise.
9057         * except.c (sjlj_mark_call_sites): Likewise.
9058         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
9059         * lra-constraints.c (emit_inc): Likewise.
9060         * ree.c (combine_set_extension): Likewise.
9061         * regmove.c (fixup_match_2): Likewise.
9062         * reload1.c (inc_for_reload): Likewise.
9064 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
9066         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
9067         (force_to_mode, simplify_shift_const_1, simplify_comparison):
9068         Use gen_int_mode with the mode of the associated simplify_* call.
9069         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
9070         * expmed.c (expand_shift_1): Likewise.
9071         * function.c (instantiate_virtual_regs_in_insn): Likewise.
9072         * loop-iv.c (iv_number_of_iterations): Likewise.
9073         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
9074         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
9076 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
9078         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
9079         of the associated expand_* call.
9080         (asan_emit_stack_protection): Likewise.
9081         * builtins.c (round_trampoline_addr): Likewise.
9082         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
9083         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
9084         (emit_store_flag): Likewise.
9085         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
9086         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
9087         Likewise.
9088         * function.c (instantiate_virtual_regs_in_insn): Likewise.
9089         * ifcvt.c (noce_try_store_flag_constants): Likewise.
9090         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
9091         * modulo-sched.c (generate_prolog_epilog): Likewise.
9092         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
9093         (expand_ctz, expand_ffs, expand_unop): Likewise.
9095 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
9097         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
9098         of the associated gen_rtx_* call.
9099         * caller-save.c (init_caller_save): Likewise.
9100         * combine.c (find_split_point, make_extraction): Likewise.
9101         (make_compound_operation): Likewise.
9102         * dwarf2out.c (mem_loc_descriptor): Likewise.
9103         * explow.c (plus_constant, probe_stack_range): Likewise.
9104         * expmed.c (expand_mult_const): Likewise.
9105         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
9106         * reload1.c (init_reload): Likewise.
9107         * valtrack.c (cleanup_auto_inc_dec): Likewise.
9108         * var-tracking.c (adjust_mems): Likewise.
9109         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
9110         rather than gen_rtx_fmt_ee.
9112 2013-09-09  Jan Hubicka  <jh@suse.cz>
9114         PR middle-end/58294
9115         * value-prof.c (gimple_ic): Copy also abnormal edges.
9117 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
9119         * asan.c (asan_shadow_cst): Use gen_int_mode.
9121 2013-09-08  Jan Hubicka  <jh@suse.cz>
9123         * ipa-profile.c: Add toplevel comment.
9124         (ipa_propagate_frequency_1): Be more conservative when profile is read.
9125         (contains_hot_call_p): New function.
9126         (ipa_propagate_frequency): Set frequencies based on counts when
9127         profile is read.
9128         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
9129         profile; do not tamper with profile after inlining if it is read.
9131 2013-09-08  Jan Hubicka  <jh@suse.cz>
9133         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
9134         speculative edges.
9136 2013-09-08  Jan Hubicka  <jh@suse.cz>
9138         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
9139         summary generation.
9141 2013-09-08  Jeff Law  <law@redhat.com>
9143         PR bootstrap/58340
9144         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
9145         of 'found'.
9147 2013-09-08  Andi Kleen  <ak@linux.intel.com>
9149         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
9151 2013-09-08  Jan Hubicka  <jh@suse.cz>
9153         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
9155 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
9157         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
9158         for non-debug insns.
9159         * lra.c (new_insn_reg): Take the containing insn as a parameter.
9160         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
9161         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
9162         accordingly.
9164 2013-09-08  Jan Hubicka  <jh@suse.cz>
9166         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
9167         targets and devirtualize to BUILT_IN_UNREACHABLE.
9168         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
9169         * ipa.c (walk_polymorphic_call_targets): New function.
9170         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
9171         functions; use the new timevar.
9172         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
9173         was removed from the program.
9174         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
9175         not consider it in the walk when its vtable is dead.
9176         (possible_polymorphic_call_targets_1): Pass anonymous flag to
9177         record_binfo.
9178         (devirt_variable_node_removal_hook): New function.
9179         (possible_polymorphic_call_targets): Also register
9180         devirt_variable_node_removal_hook.
9181         (ipa_devirt): Do not do non-speculative devirtualization.
9182         (gate_ipa_devirt): One execute if devirtualizing speculatively.
9184 2013-09-08  Jan Hubicka  <jh@suse.cz>
9186         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
9187         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
9188         varpool_remove_variable_insertion_hook): Declare.
9189         * varpool.c (varpool_node_hook_list): New structure.
9190         (first_varpool_node_removal_hook,
9191         first_varpool_variable_insertion_hook): New variables.
9192         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
9193         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
9194         varpool_remove_variable_insertion_hook,
9195         varpool_call_variable_insertion_hooks): New functions.
9196         (varpool_remove_node): Use it.
9198 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
9200         PR c++/54941
9201         * diagnostic.c (diagnostic_build_prefix): When s.file is
9202         "<built-in>" don't output line and column numbers.
9204 2013-09-06  Jan Hubicka  <jh@suse.cz>
9206         * cgraphunit.c (expand_thunk): Get body before touching arguments.
9207         * lto-streamer-out.c: Stream thunks, too.
9208         * lto-streamer-in.c (input_function): Pop cfun here
9209         (lto_read_body): Instead of here.
9211 2013-09-06  Caroline Tice  <cmtice@google.com>
9213         * doc/install.texi: Add documentation for the --enable-vtable-verify
9214         and the --disable-libvtv configure options.
9216 2013-09-06  Jeff Law  <law@redhat.com>
9218         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
9219         edge implied equivalences into successor phis.
9221 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
9223         * resource.c (mark_referenced_resources): Handle COND_EXEC.
9225 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
9227         * resource.c (mark_target_live_regs): Compute resources taking
9228         into account if a call is predicated or not.
9230 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
9232         * toplev.c (output_stack_usage): Be prepared for suffixes created by
9233         the compiler in the function names.
9235 2013-09-06  Jan Hubicka  <jh@suse.cz>
9237         PR middle-end/58094
9238         * ipa-inline.c (has_caller_p): New function.
9239         (want_inline_function_to_all_callers_p): Use it.
9240         (sum_callers, inline_to_all_callers): Break out from ...
9241         (ipa_inline): ... here.
9243 2013-09-06  Jan Hubicka  <jh@suse.cz>
9245         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
9246         only when AVX is enabled.
9248 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9250         * config/aarch64/aarch64.md
9251         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
9252         is fpsimd_<load/store>2.
9253         (load_pair<mode>): Likewise.
9254         (store_pair<mode>): Likewise.
9256 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9258         * config/arm/types.md (type): Add "mrs" type.
9259         * config/aarch64/aarch64.md
9260         (aarch64_load_tp_hard): Make type "mrs".
9261         * config/arm/arm.md
9262         (load_tp_hard): Make type "mrs".
9263         * config/arm/cortex-a15.md: Update with new attributes.
9264         * config/arm/cortex-a5.md: Update with new attributes.
9265         * config/arm/cortex-a53.md: Update with new attributes.
9266         * config/arm/cortex-a7.md: Update with new attributes.
9267         * config/arm/cortex-a8.md: Update with new attributes.
9268         * config/arm/cortex-a9.md: Update with new attributes.
9269         * config/arm/cortex-m4.md: Update with new attributes.
9270         * config/arm/cortex-r4.md: Update with new attributes.
9271         * config/arm/fa526.md: Update with new attributes.
9272         * config/arm/fa606te.md: Update with new attributes.
9273         * config/arm/fa626te.md: Update with new attributes.
9274         * config/arm/fa726te.md: Update with new attributes.
9276 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9278         * config/aarch64/aarch64.md
9279         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
9280         (*movtf_aarch64): Likewise.
9281         * config/arm/arm.md
9282         (thumb1_movdi_insn): Use "multiple" for type where more than one
9283         instruction is used for a move.
9284         (*arm32_movhf): Likewise.
9285         (*thumb_movdf_insn): Likewise.
9287 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9289         * config/arm/types.md (type): Rename fcpys to fmov.
9290         * config/arm/vfp.md
9291         (*arm_movsi_vfp): Rename type fcpys as fmov.
9292         (*thumb2_movsi_vfp): Likewise
9293         (*movhf_vfp_neon): Likewise
9294         (*movhf_vfp): Likewise
9295         (*movsf_vfp): Likewise
9296         (*thumb2_movsf_vfp): Likewise
9297         (*movsfcc_vfp): Likewise
9298         (*thumb2_movsfcc_vfp): Likewise
9299         * config/aarch64/aarch64-simd.md
9300         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
9301         * config/aarch64/aarch64.md
9302         (*movsi_aarch64): Replace type mov_reg with fmovs.
9303         (*movdi_aarch64): Likewise
9304         (*movsf_aarch64): Likewise
9305         (*movdf_aarch64): Likewise
9306         * config/arm/arm.c
9307         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
9308         * config/arm/iwmmxt.md
9309         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
9310         * config/arm/arm1020e.md: Update with new attributes.
9311         * config/arm/cortex-a15-neon.md: Update with new attributes.
9312         * config/arm/cortex-a5.md: Update with new attributes.
9313         * config/arm/cortex-a53.md: Update with new attributes.
9314         * config/arm/cortex-a7.md: Update with new attributes.
9315         * config/arm/cortex-a8-neon.md: Update with new attributes.
9316         * config/arm/cortex-a9.md: Update with new attributes.
9317         * config/arm/cortex-m4-fpu.md: Update with new attributes.
9318         * config/arm/cortex-r4f.md: Update with new attributes.
9319         * config/arm/marvell-pj4.md: Update with new attributes.
9320         * config/arm/vfp11.md: Update with new attributes.
9322 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9324         * config/aarch64/aarch64.md
9325         (*madd<mode>): Fix type attribute.
9326         (*maddsi_uxtw): Likewise.
9327         (*msub<mode>): Likewise.
9328         (*msubsi_uxtw): Likewise.
9329         (<su_optab>maddsidi4): Likewise.
9330         (<su_optab>msubsidi4): Likewise.
9332 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9334         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
9335         * config/arm/arm.md (core_cycles): Remove fdiv.
9336         * config/arm/vfp.md:
9337         (*sqrtsf2_vfp): Update for attribute changes.
9338         (*sqrtdf2_vfp): Likewise.
9339         * config/aarch64/aarch64.md:
9340         (sqrt<mode>2): Update for attribute changes.
9341         * config/arm/arm1020e.md: Update with new attributes.
9342         * config/arm/cortex-a15-neon.md: Update with new attributes.
9343         * config/arm/cortex-a5.md: Update with new attributes.
9344         * config/arm/cortex-a53.md: Update with new attributes.
9345         * config/arm/cortex-a7.md: Update with new attributes.
9346         * config/arm/cortex-a8-neon.md: Update with new attributes.
9347         * config/arm/cortex-a9.md: Update with new attributes.
9348         * config/arm/cortex-m4-fpu.md: Update with new attributes.
9349         * config/arm/cortex-r4f.md: Update with new attributes.
9350         * config/arm/marvell-pj4.md: Update with new attributes.
9351         * config/arm/vfp11.md: Update with new attributes.
9353 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9355         * config/arm/types.md
9356         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
9357         * config/aarch64/aarch64.md
9358         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
9359         new attributes.
9360         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
9361         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
9362         (float<GPI:mode><GPF:mode>2): Likewise.
9363         * config/arm/vfp.md
9364         (*truncsisf2_vfp): Update with new attributes.
9365         (*truncsidf2_vfp): Likewise.
9366         (fixuns_truncsfsi2): Likewise.
9367         (fixuns_truncdfsi2): Likewise.
9368         (*floatsisf2_vfp): Likewise.
9369         (*floatsidf2_vfp): Likewise.
9370         (floatunssisf2): Likewise.
9371         (floatunssidf2): Likewise.
9372         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
9373         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
9374         * config/arm/arm1020e.md: Update with new attributes.
9375         * config/arm/cortex-a15-neon.md: Update with new attributes.
9376         * config/arm/cortex-a5.md: Update with new attributes.
9377         * config/arm/cortex-a53.md: Update with new attributes.
9378         * config/arm/cortex-a7.md: Update with new attributes.
9379         * config/arm/cortex-a8-neon.md: Update with new attributes.
9380         * config/arm/cortex-a9.md: Update with new attributes.
9381         * config/arm/cortex-m4-fpu.md: Update with new attributes.
9382         * config/arm/cortex-r4f.md: Update with new attributes.
9383         * config/arm/marvell-pj4.md: Update with new attributes.
9384         * config/arm/vfp11.md: Update with new attributes.
9386 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9388         * config/aarch64/arm_neon.h
9389         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
9390         (vqtbx<1,2,3,4><q>_s8): Likewise.
9392 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9394         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
9395         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
9396         (add<mode>3): Add type attribute.
9397         (add<mode>3): Likewise.
9398         (usadd<mode>3): Likewise.
9399         (ssadd<mode>3): Likewise.
9400         (sub<mode>3): Likewise.
9401         (sub<mode>3): Likewise.
9402         (ussub<mode>3): Likewise.
9403         (sssub<mode>3): Likewise.
9404         (ssmulsa3): Likewise.
9405         (usmulusa3): Likewise.
9406         (arm_usatsihi): Likewise.
9407         * config/arm/vfp.md
9408         (*movdi_vfp): Add types for all instructions.
9409         (*movdi_vfp_cortexa8): Likewise.
9410         (*movhf_vfp_neon): Likewise.
9411         (*movhf_vfp): Likewise.
9412         (*movdf_vfp): Likewise.
9413         (*thumb2_movdf_vfp): Likewise.
9414         (*thumb2_movdfcc_vfp): Likewise.
9415         * config/arm/arm.md: Add type attribute to all insn patterns.
9416         (*thumb1_adddi3): Add type attribute.
9417         (*arm_adddi3): Likewise.
9418         (*adddi_sesidi_di): Likewise.
9419         (*adddi_zesidi_di): Likewise.
9420         (*thumb1_addsi3): Likewise.
9421         (addsi3_compare0): Likewise.
9422         (*addsi3_compare0_scratch): Likewise.
9423         (*compare_negsi_si): Likewise.
9424         (cmpsi2_addneg): Likewise.
9425         (*addsi3_carryin_<optab>): Likewise.
9426         (*addsi3_carryin_alt2_<optab>): Likewise.
9427         (*addsi3_carryin_clobercc_<optab>): Likewise.
9428         (*subsi3_carryin): Likewise.
9429         (*subsi3_carryin_const): Likewise.
9430         (*subsi3_carryin_compare): Likewise.
9431         (*subsi3_carryin_compare_const): Likewise.
9432         (*arm_subdi3): Likewise.
9433         (*thumb_subdi3): Likewise.
9434         (*subdi_di_zesidi): Likewise.
9435         (*subdi_di_sesidi): Likewise.
9436         (*subdi_zesidi_di): Likewise.
9437         (*subdi_sesidi_di): Likewise.
9438         (*subdi_zesidi_ze): Likewise.
9439         (thumb1_subsi3_insn): Likewise.
9440         (*arm_subsi3_insn): Likewise.
9441         (*anddi3_insn): Likewise.
9442         (*anddi_zesidi_di): Likewise.
9443         (*anddi_sesdi_di): Likewise.
9444         (*ne_zeroextracts): Likewise.
9445         (*ne_zeroextracts): Likewise.
9446         (*ite_ne_zeroextr): Likewise.
9447         (*ite_ne_zeroextr): Likewise.
9448         (*anddi_notdi_di): Likewise.
9449         (*anddi_notzesidi): Likewise.
9450         (*anddi_notsesidi): Likewise.
9451         (andsi_notsi_si): Likewise.
9452         (thumb1_bicsi3): Likewise.
9453         (*iordi3_insn): Likewise.
9454         (*iordi_zesidi_di): Likewise.
9455         (*iordi_sesidi_di): Likewise.
9456         (*thumb1_iorsi3_insn): Likewise.
9457         (*xordi3_insn): Likewise.
9458         (*xordi_zesidi_di): Likewise.
9459         (*xordi_sesidi_di): Likewise.
9460         (*arm_xorsi3): Likewise.
9461         (*andsi_iorsi3_no): Likewise.
9462         (*smax_0): Likewise.
9463         (*smax_m1): Likewise.
9464         (*arm_smax_insn): Likewise.
9465         (*smin_0): Likewise.
9466         (*arm_smin_insn): Likewise.
9467         (*arm_umaxsi3): Likewise.
9468         (*arm_uminsi3): Likewise.
9469         (*minmax_arithsi): Likewise.
9470         (*minmax_arithsi_): Likewise.
9471         (*satsi_<SAT:code>): Likewise.
9472         (arm_ashldi3_1bit): Likewise.
9473         (arm_ashrdi3_1bit): Likewise.
9474         (arm_lshrdi3_1bit): Likewise.
9475         (*arm_negdi2): Likewise.
9476         (*thumb1_negdi2): Likewise.
9477         (*arm_negsi2): Likewise.
9478         (*thumb1_negsi2): Likewise.
9479         (*negdi_extendsid): Likewise.
9480         (*negdi_zero_extend): Likewise.
9481         (*arm_abssi2): Likewise.
9482         (*thumb1_abssi2): Likewise.
9483         (*arm_neg_abssi2): Likewise.
9484         (*thumb1_neg_abss): Likewise.
9485         (one_cmpldi2): Likewise.
9486         (extend<mode>di2): Likewise.
9487         (*compareqi_eq0): Likewise.
9488         (*arm_extendhisi2addsi): Likewise.
9489         (*arm_movdi): Likewise.
9490         (*thumb1_movdi_insn): Likewise.
9491         (*arm_movt): Likewise.
9492         (*thumb1_movsi_insn): Likewise.
9493         (pic_add_dot_plus_four): Likewise.
9494         (pic_add_dot_plus_eight): Likewise.
9495         (tls_load_dot_plus_eight): Likewise.
9496         (*thumb1_movhi_insn): Likewise.
9497         (*thumb1_movsf_insn): Likewise.
9498         (*movdf_soft_insn): Likewise.
9499         (*thumb_movdf_insn): Likewise.
9500         (cbranchsi4_insn): Likewise.
9501         (cbranchsi4_scratch): Likewise.
9502         (*negated_cbranchsi4): Likewise.
9503         (*tbit_cbranch): Likewise.
9504         (*tlobits_cbranch): Likewise.
9505         (*tstsi3_cbranch): Likewise.
9506         (*cbranchne_decr1): Likewise.
9507         (*addsi3_cbranch): Likewise.
9508         (*addsi3_cbranch_scratch): Likewise.
9509         (*arm_cmpdi_insn): Likewise.
9510         (*arm_cmpdi_unsig): Likewise.
9511         (*arm_cmpdi_zero): Likewise.
9512         (*thumb_cmpdi_zero): Likewise.
9513         (*deleted_compare): Likewise.
9514         (*mov_scc): Likewise.
9515         (*mov_negscc): Likewise.
9516         (*mov_notscc): Likewise.
9517         (*cstoresi_eq0_thumb1_insn): Likewise.
9518         (cstoresi_nltu_thumb1): Likewise.
9519         (cstoresi_ltu_thu): Likewise.
9520         (thumb1_addsi3_addgeu): Likewise.
9521         (*arm_jump): Likewise.
9522         (*thumb_jump): Likewise.
9523         (*check_arch2): Likewise.
9524         (arm_casesi_internal): Likewise.
9525         (thumb1_casesi_dispatch): Likewise.
9526         (*arm_indirect_jump): Likewise.
9527         (*thumb1_indirect_jump): Likewise.
9528         (nop): Likewise.
9529         (*and_scc): Likewise.
9530         (*ior_scc): Likewise.
9531         (*compare_scc): Likewise.
9532         (*cond_move): Likewise.
9533         (*cond_arith): Likewise.
9534         (*cond_sub): Likewise.
9535         (*cmp_ite0): Likewise.
9536         (*cmp_ite1): Likewise.
9537         (*cmp_and): Likewise.
9538         (*cmp_ior): Likewise.
9539         (*ior_scc_scc): Likewise.
9540         (*ior_scc_scc_cmp): Likewise.
9541         (*and_scc_scc): Likewise.
9542         (*and_scc_scc_cmp): Likewise.
9543         (*and_scc_scc_nod): Likewise.
9544         (*negscc): Likewise.
9545         (movcond_addsi): Likewise.
9546         (movcond): Likewise.
9547         (*ifcompare_plus_move): Likewise.
9548         (*if_plus_move): Likewise.
9549         (*ifcompare_move_plus): Likewise.
9550         (*if_move_plus): Likewise.
9551         (*ifcompare_arith_arith): Likewise.
9552         (*if_arith_arith): Likewise.
9553         (*ifcompare_arith_move): Likewise.
9554         (*if_arith_move): Likewise.
9555         (*ifcompare_move_arith): Likewise.
9556         (*if_move_arith): Likewise.
9557         (*ifcompare_move_not): Likewise.
9558         (*if_move_not): Likewise.
9559         (*ifcompare_not_move): Likewise.
9560         (*if_not_move): Likewise.
9561         (*ifcompare_shift_move): Likewise.
9562         (*if_shift_move): Likewise.
9563         (*ifcompare_move_shift): Likewise.
9564         (*if_move_shift): Likewise.
9565         (*ifcompare_shift_shift): Likewise.
9566         (*ifcompare_not_arith): Likewise.
9567         (*ifcompare_arith_not): Likewise.
9568         (*if_arith_not): Likewise.
9569         (*ifcompare_neg_move): Likewise.
9570         (*if_neg_move): Likewise.
9571         (*ifcompare_move_neg): Likewise.
9572         (*if_move_neg): Likewise.
9573         (prologue_thumb1_interwork): Likewise.
9574         (*cond_move_not): Likewise.
9575         (*sign_extract_onebit): Likewise.
9576         (*not_signextract_onebit): Likewise.
9577         (stack_tie): Likewise.
9578         (align_4): Likewise.
9579         (align_8): Likewise.
9580         (consttable_end): Likewise.
9581         (consttable_1): Likewise.
9582         (consttable_2): Likewise.
9583         (consttable_4): Likewise.
9584         (consttable_8): Likewise.
9585         (consttable_16): Likewise.
9586         (*thumb1_tablejump): Likewise.
9587         (prefetch): Likewise.
9588         (force_register_use): Likewise.
9589         (thumb_eh_return): Likewise.
9590         (load_tp_hard): Likewise.
9591         (load_tp_soft): Likewise.
9592         (tlscall): Likewise.
9593         (*arm_movtas_ze): Likewise.
9594         (*arm_rev): Likewise.
9595         (*arm_revsh): Likewise.
9596         (*arm_rev16): Likewise.
9597         * config/arm/thumb2.md
9598         (*thumb2_smaxsi3): Likewise.
9599         (*thumb2_sminsi3): Likewise.
9600         (*thumb32_umaxsi3): Likewise.
9601         (*thumb2_uminsi3): Likewise.
9602         (*thumb2_negdi2): Likewise.
9603         (*thumb2_abssi2): Likewise.
9604         (*thumb2_neg_abss): Likewise.
9605         (*thumb2_movsi_insn): Likewise.
9606         (tls_load_dot_plus_four): Likewise.
9607         (*thumb2_movhi_insn): Likewise.
9608         (*thumb2_mov_scc): Likewise.
9609         (*thumb2_mov_negs): Likewise.
9610         (*thumb2_mov_negs): Likewise.
9611         (*thumb2_mov_nots): Likewise.
9612         (*thumb2_mov_nots): Likewise.
9613         (*thumb2_movsicc_): Likewise.
9614         (*thumb2_movsfcc_soft_insn): Likewise.
9615         (*thumb2_indirect_jump): Likewise.
9616         (*thumb2_and_scc): Likewise.
9617         (*thumb2_ior_scc): Likewise.
9618         (*thumb2_ior_scc_strict_it): Likewise.
9619         (*thumb2_cond_move): Likewise.
9620         (*thumb2_cond_arith): Likewise.
9621         (*thumb2_cond_ari): Likewise.
9622         (*thumb2_cond_sub): Likewise.
9623         (*thumb2_negscc): Likewise.
9624         (*thumb2_movcond): Likewise.
9625         (thumb2_casesi_internal): Likewise.
9626         (thumb2_casesi_internal_pic): Likewise.
9627         (*thumb2_alusi3_short): Likewise.
9628         (*thumb2_mov<mode>_shortim): Likewise.
9629         (*thumb2_addsi_short): Likewise.
9630         (*thumb2_subsi_short): Likewise.
9631         (thumb2_addsi3_compare0): Likewise.
9632         (*thumb2_cbz): Likewise.
9633         (*thumb2_cbnz): Likewise.
9634         (*thumb2_one_cmplsi2_short): Likewise.
9635         (*thumb2_negsi2_short): Likewise.
9636         (*orsi_notsi_si): Likewise.
9637         * config/arm/arm1020e.md: Update with new attributes.
9638         * config/arm/arm1026ejs.md: Update with new attributes.
9639         * config/arm/arm1136jfs.md: Update with new attributes.
9640         * config/arm/arm926ejs.md: Update with new attributes.
9641         * config/arm/cortex-a15.md: Update with new attributes.
9642         * config/arm/cortex-a5.md: Update with new attributes.
9643         * config/arm/cortex-a53.md: Update with new attributes.
9644         * config/arm/cortex-a7.md: Update with new attributes.
9645         * config/arm/cortex-a8.md: Update with new attributes.
9646         * config/arm/cortex-a9.md: Update with new attributes.
9647         * config/arm/cortex-m4.md: Update with new attributes.
9648         * config/arm/cortex-r4.md: Update with new attributes.
9649         * config/arm/fa526.md: Update with new attributes.
9650         * config/arm/fa606te.md: Update with new attributes.
9651         * config/arm/fa626te.md: Update with new attributes.
9652         * config/arm/fa726te.md: Update with new attributes.
9654 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9656         * config/aarch64/aarch64-simd.md
9657         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
9658         <vwx> iterator to ensure correct register choice.
9659         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
9660         (aarch64_sqdmull_n<mode>): Likewise.
9661         (aarch64_sqdmull2_n<mode>_internal): Likewise.
9662         * config/aarch64/arm_neon.h
9663         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
9664         (vml<as><q>_n_<su>16): Likewise.
9665         (vml<as>l_high_lane<q>_<su>16): Likewise.
9666         (vml<as>l_high_n_<su>16): Likewise.
9667         (vml<as>l_lane<q>_<su>16): Likewise.
9668         (vml<as>l_n_<su>16): Likewise.
9669         (vmul<q>_lane<q>_<su>16): Likewise.
9670         (vmul<q>_n_<su>16): Likewise.
9671         (vmull_lane<q>_<su>16): Likewise.
9672         (vmull_n_<su>16): Likewise.
9673         (vmull_high_lane<q>_<su>16): Likewise.
9674         (vmull_high_n_<su>16): Likewise.
9675         (vqrdmulh<q>_n_s16): Likewise.
9677 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
9679         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
9680         have the correct lane parameter.
9682 2013-09-06  Richard Biener <rguenther@suse.de>
9684         * cfganal.c (control_dependences::~control_dependences):
9685         Properly free all of the vector.
9687 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
9689         PR target/58269
9690         * config/i386/i386.c (ix86_conditional_register_usage):
9691         Proper initialize extended SSE registers.
9693 2013-09-06  Jan Hubicka  <jh@suse.cz>
9695         PR tree-optimization/58311
9696         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
9698 2013-09-06  Jan Hubicka  <jh@suse.cz>
9700         * Makefile.in (tree-sra.o): Update dependencies.
9701         * tree-sra.c: Include ipa-utils.h
9702         (scan_function): Use recursive_call_p.
9703         (has_caller_p): New function.
9704         (cgraph_for_node_and_aliases): Count also callers of aliases.
9706 2013-09-06  Jan Hubicka  <jh@suse.cz>
9708         PR middle-end/58094
9709         * cgraph.h (symtab_semantically_equivalent_p): Declare.
9710         * tree-tailcall.c: Include ipa-utils.h.
9711         (find_tail_calls): Use it.
9712         * ipa-pure-const.c (check_call): Likewise.
9713         * ipa-utils.c (recursive_call_p): New function.
9714         * ipa-utils.h (recursive_call_p): Dclare.
9715         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
9716         (symtab_semantically_equivalent_p): New function.
9717         * Makefile.in (tree-tailcall.o): Update dependencies.
9719 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
9721         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
9722         non-inlinable part.
9724 2013-09-06  Richard Biener  <rguenther@suse.de>
9726         * lto-streamer.h (lto_global_var_decls): Remove.
9727         * Makefile.in (OBJS): Remove lto-symtab.o.
9728         (lto-symtab.o): Remove.
9729         (GTFILES): Remove lto-symtab.c
9730         * lto-symtab.c: Move to lto/
9732 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9734         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
9735         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
9736         (UNSPEC_FPINT_RINT): New constant definitions.
9737         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
9738         definition with 2 attributes.
9739         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
9740         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
9741         definitions.
9743 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9745         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
9746         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
9747         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
9748         attribute to "z196_alone".
9749         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
9750         "zEC12_simple".
9752 2013-09-06  Richard Biener  <rguenther@suse.de>
9754         * basic-block.h (class control_dependences): New.
9755         * tree-ssa-dce.c (control_dependence_map): Remove.
9756         (cd): New global.
9757         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
9758         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
9759         find_pdom, find_control_dependence, find_all_control_dependences):
9760         Move to cfganal.c.
9761         (mark_control_dependent_edges_necessary,
9762         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
9763         tree_dce_done, perform_tree_ssa_dce): Adjust.
9764         * cfganal.c (set_control_dependence_map_bit,
9765         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
9766         find_all_control_dependences): Move from tree-ssa-dce.c and
9767         implement as methods of control_dependences class.
9768         (control_dependences::control_dependences): New.
9769         (control_dependences::~control_dependences): Likewise.
9770         (control_dependences::get_edges_dependent_on): Likewise.
9771         (control_dependences::get_edge): Likewise.
9773 2013-09-04  Jan Hubicka  <jh@suse.cz>
9775         * tree.c (types_same_for_odr): Drop overactive check.
9776         * ipa-devirt.c (hash_type_name): Likewise.
9778 2013-09-04  Jan Hubicka  <jh@suse.cz>
9780         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
9781         (analyze_functions): ... here.
9783 2013-09-04  Jan Hubicka  <jh@suse.cz>
9785         PR middle-end/58201
9786         * cgraphunit.c (analyze_functions): Clear AUX fields
9787         after processing; initialize assembler name has.
9789 2013-09-05  Jeff Law  <law@redhat.com>
9791         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
9792         from thread_around_empty_block.  Record threading path into PATH.
9793         Recurse if threading through the initial block is successful.
9794         (thread_across_edge): Corresponding changes to slightly simplify.
9796 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
9798         * config/aarch64/aarch64.md
9799         (type): Remove frecpe, frecps, frecpx.
9800         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
9801         fix to be a TARGET_SIMD instruction.
9802         (aarch64_frecps): Remove.
9803         * config/aarch64/aarch64-simd.md
9804         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
9805         (aarch64_frecps<mode>): Handle all float/vector of float modes.
9807 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
9808             Sofiane Naci  <sofiane.naci@arm.com>
9810         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
9811         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
9812         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
9813         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
9814         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
9815         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
9816         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
9817         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
9818         into "clz, "rbit".  Rename "shift" to "shift_imm".
9819         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
9820         changes.  Update for attribute changes all occurrences of arlo_* and
9821         shift* types.
9822         * config/arm/arm-fixed.md: Update for attribute changes
9823         all occurrences of arlo_* types.
9824         * config/arm/thumb2.md: Update for attribute changes all occurrences
9825         of arlo_* types.
9826         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
9827         (cortexa7_older_only): Likewise.
9828         (cortexa7_younger):  Likewise.
9829         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
9830         (1020alu_shift_op): Likewise.
9831         (1020alu_shift_reg_op): Likewise.
9832         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
9833         (alu_shift_op): Likewise.
9834         (alu_shift_reg_op): Likewise.
9835         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
9836         (11_alu_shift_op): Likewise.
9837         (11_alu_shift_reg_op): Likewise.
9838         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
9839         (9_alu_shift_reg_op): Likewise.
9840         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
9841         attribute changes.
9842         (cortex_a15_alu_shift): Likewise.
9843         (cortex_a15_alu_shift_reg): Likewise.
9844         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
9845         attribute changes.
9846         (cortex_a5_alu_shift): Likewise.
9847         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
9848         attribute changes.
9849         (cortex_a53_alu_shift): Likewise.
9850         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
9851         attribute changes.
9852         (cortex_a7_alu_reg): Likewise.
9853         (cortex_a7_alu_shift): Likewise.
9854         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
9855         attribute changes.
9856         (cortex_a8_alu_shift): Likewise.
9857         (cortex_a8_alu_shift_reg): Likewise.
9858         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
9859         (cortex_a9_dp_shift): Likewise.
9860         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
9861         attribute changes.
9862         * config/arm/cortex-r4.md
9863         (cortex_r4_alu): Update for attribute changes.
9864         (cortex_r4_mov): Likewise.
9865         (cortex_r4_alu_shift_reg): Likewise.
9866         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
9867         (526_alu_shift_op): Likewise.
9868         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
9869         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
9870         (626te_alu_shift_op): Likewise.
9871         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
9872         (726te_alu_shift_op): Likewise.
9873         (726te_alu_shift_reg_op): Likewise.
9874         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
9875         (mp626_alu_shift_op): Likewise.
9876         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
9877         (pj4_alu_conds): Likewise.
9878         (pj4_shift): Likewise.
9879         (pj4_shift_conds): Likewise.
9880         (pj4_alu_shift): Likewise.
9881         (pj4_alu_shift_conds): Likewise.
9882         * config/aarch64/aarch64.md: Update for attribute change
9883         all occurrences of arlo_* and shift* types.
9885 2013-09-05  Mike Stump  <mikestump@comcast.net>
9887         * tree.h: Move documentation for tree_function_decl to tree-core.h
9888         with the declaration.
9890 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
9892         PR target/58139
9893         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
9894         looking for widest mode.
9896 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
9898         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
9900 2013-09-05  Richard Biener  <rguenther@suse.de>
9902         PR tree-optimization/58137
9903         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
9904         Do not create vectors of pointers.
9905         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
9906         types for the components of the vector initializer.
9907         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
9908         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
9910 2013-09-05  Martin Jambor  <mjambor@suse.cz>
9912         * ipa-prop.c (remove_described_reference): Accept missing references,
9913         return false if that hppens, otherwise return true.
9914         (cgraph_node_for_jfunc): New function.
9915         (try_decrement_rdesc_refcount): Likewise.
9916         (try_make_edge_direct_simple_call): Use them.
9917         (ipa_edge_removal_hook): Remove references from rdescs.
9918         (ipa_edge_duplication_hook): Clone rdescs and their references
9919         when the new edge has the same caller as the old one.
9920         * cgraph.c (cgraph_resolve_speculation): Remove speculative
9921         reference before removing any edges.
9923 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
9925         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
9926         initial store.
9927         * thumb2.md (thumb2_storewb_parisi): New pattern.
9929 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
9931         * config/aarch64/aarch64-option-extensions.def: Add
9932         AARCH64_OPT_EXTENSION of 'crc'.
9933         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
9934         (AARCH64_ISA_CRC): Ditto.
9935         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
9936         description of the CRC extension.
9938 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
9940         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
9941         * config/rs6000/linux.h: Ditto.
9942         * alpha/linux.h: Ditto.
9943         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
9944         no_c99_libc_has_function.
9945         * config/c6x/uclinux-elf.h: Ditto.
9946         * config/lm32/uclinux-elf.h: Ditto.
9947         * config/m68k/uclinux.h: Ditto.
9948         * config/moxie/uclinux.h: Ditto.
9949         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
9950         (crisv32-*-linux*, cris-*-linux*): Ditto.
9951         * config/bfin/bfin.c: Include "tm_p.h".
9953 2013-09-05  Richard Biener  <rguenther@suse.de>
9955         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
9956         check for a definition without a basic-block.
9958 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
9959             Sofiane Naci <sofiane.naci@arm.com>
9961         * config/aarch64/aarch64.md
9962         (*movti_aarch64): Rename r_2_f and f_2_r.
9963         (*movsf_aarch64): Likewise.
9964         (*movdf_aarch64): Likewise.
9965         (*movtf_aarch64): Likewise.
9966         (aarch64_movdi_<mode>low): Likewise.
9967         (aarch64_movdi_<mode>high): Likewise.
9968         (aarch64_mov<mode>high_di): Likewise.
9969         (aarch64_mov<mode>low_di): Likewise.
9970         (aarch64_movtilow_tilow): Likewise.
9971         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
9972         values to config/arm/types.md
9973         (attribute "conds"): Update for attribute change.
9974         (anddi3_insn): Likewise.
9975         (iordi3_insn): Likewise.
9976         (xordi3_insn): Likewise.
9977         (one_cmpldi2): Likewise.
9978         * config/arm/types.md (type): Add Neon types.
9979         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
9980         use "type" attribute.
9981         (movmisalign<mode>_neon_store): Likewise.
9982         (movmisalign<mode>_neon_load): Likewise.
9983         (vec_set<mode>_internal): Likewise.
9984         (vec_setv2di_internal): Likewise.
9985         (vec_extract<mode>): Likewise.
9986         (vec_extractv2di): Likewise.
9987         (add<mode>3_neon): Likewise.
9988         (adddi3_neon): Likewise.
9989         (sub<mode>3_neon): Likewise.
9990         (subdi3_neon): Likewise.
9991         (mul<mode>3_neon): Likewise.
9992         (mul<mode>3add<mode>_neon): Likewise.
9993         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
9994         (fma<VCVTF:mode>4)): Likewise.
9995         (fma<VCVTF:mode>4_intrinsic): Likewise.
9996         (fmsub<VCVTF:mode>4)): Likewise.
9997         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
9998         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
9999         (ior<mode>3): Likewise.
10000         (and<mode>3): Likewise.
10001         (anddi3_neon): Likewise.
10002         (orn<mode>3_neon): Likewise.
10003         (orndi3_neon): Likewise.
10004         (bic<mode>3_neon): Likewise.
10005         (bicdi3_neon): Likewise.
10006         (xor<mode>3): Likewise.
10007         (one_cmpl<mode>2): Likewise.
10008         (abs<mode>2): Likewise.
10009         (neg<mode>2): Likewise.
10010         (umin<mode>3_neon): Likewise.
10011         (umax<mode>3_neon): Likewise.
10012         (smin<mode>3_neon): Likewise.
10013         (smax<mode>3_neon): Likewise.
10014         (vashl<mode>3): Likewise.
10015         (vashr<mode>3_imm): Likewise.
10016         (vlshr<mode>3_imm): Likewise.
10017         (ashl<mode>3_signed): Likewise.
10018         (ashl<mode>3_unsigned): Likewise.
10019         (neon_load_count): Likewise.
10020         (ashldi3_neon_noclobber): Likewise.
10021         (signed_shift_di3_neon): Likewise.
10022         (unsigned_shift_di3_neon): Likewise.
10023         (ashrdi3_neon_imm_noclobber): Likewise.
10024         (lshrdi3_neon_imm_noclobber): Likewise.
10025         (widen_ssum<mode>3): Likewise.
10026         (widen_usum<mode>3): Likewise.
10027         (quad_halves_<code>v4si): Likewise.
10028         (quad_halves_<code>v4sf): Likewise.
10029         (quad_halves_<code>v8hi): Likewise.
10030         (quad_halves_<code>v16qi): Likewise.
10031         (reduc_splus_v2di): Likewise.
10032         (neon_vpadd_internal<mode>): Likewise.
10033         (neon_vpsmin<mode>): Likewise.
10034         (neon_vpsmax<mode>): Likewise.
10035         (neon_vpumin<mode>): Likewise.
10036         (neon_vpumax<mode>): Likewise.
10037         (ss_add<mode>_neon): Likewise.
10038         (us_add<mode>_neon): Likewise.
10039         (ss_sub<mode>_neon): Likewise.
10040         (us_sub<mode>_neon): Likewise.
10041         (neon_vadd<mode>_unspec): Likewise.
10042         (neon_vaddl<mode>): Likewise.
10043         (neon_vaddw<mode>): Likewise.
10044         (neon_vhadd<mode>): Likewise.
10045         (neon_vqadd<mode>): Likewise.
10046         (neon_vaddhn<mode>): Likewise.
10047         (neon_vmul<mode>): Likewise.
10048         (neon_vmla<mode>): Likewise.
10049         (neon_vmlal<mode>): Likewise.
10050         (neon_vmls<mode>): Likewise.
10051         (neon_vmlsl<mode>): Likewise.
10052         (neon_vqdmulh<mode>): Likewise.
10053         (neon_vqdmlal<mode>): Likewise.
10054         (neon_vqdmlsl<mode>): Likewise.
10055         (neon_vmull<mode>): Likewise.
10056         (neon_vqdmull<mode>): Likewise.
10057         (neon_vsub<mode>_unspec): Likewise.
10058         (neon_vsubl<mode>): Likewise.
10059         (neon_vsubw<mode>): Likewise.
10060         (neon_vqsub<mode>): Likewise.
10061         (neon_vhsub<mode>): Likewise.
10062         (neon_vsubhn<mode>): Likewise.
10063         (neon_vceq<mode>): Likewise.
10064         (neon_vcge<mode>): Likewise.
10065         (neon_vcgeu<mode>): Likewise.
10066         (neon_vcgt<mode>): Likewise.
10067         (neon_vcgtu<mode>): Likewise.
10068         (neon_vcle<mode>): Likewise.
10069         (neon_vclt<mode>): Likewise.
10070         (neon_vcage<mode>): Likewise.
10071         (neon_vcagt<mode>): Likewise.
10072         (neon_vtst<mode>): Likewise.
10073         (neon_vabd<mode>): Likewise.
10074         (neon_vabdl<mode>): Likewise.
10075         (neon_vaba<mode>): Likewise.
10076         (neon_vabal<mode>): Likewise.
10077         (neon_vmax<mode>): Likewise.
10078         (neon_vmin<mode>): Likewise.
10079         (neon_vpaddl<mode>): Likewise.
10080         (neon_vpadal<mode>): Likewise.
10081         (neon_vpmax<mode>): Likewise.
10082         (neon_vpmin<mode>): Likewise.
10083         (neon_vrecps<mode>): Likewise.
10084         (neon_vrsqrts<mode>): Likewise.
10085         (neon_vqabs<mode>): Likewise.
10086         (neon_vqneg<mode>): Likewise.
10087         (neon_vcls<mode>): Likewise.
10088         (clz<mode>2): Likewise.
10089         (popcount<mode>2): Likewise.
10090         (neon_vrecpe): Likewise.
10091         (neon_vrsqrte): Likewise.
10092         (neon_vget_lane<mode>_sext_internal): Likewise.
10093         (neon_vget_lane<mode>_zext_internal): Likewise.
10094         (neon_vdup_n<mode>): Likewise.
10095         (neon_vdup_nv2di): Likewise.
10096         (neon_vdpu_lane<mode>_internal): Likewise.
10097         (neon_vswp<mode>): Likewise.
10098         (float<mode><V_cvtto>2): Likewise.
10099         (floatuns<mode><V_cvtto>2): Likewise.
10100         (fix_trunc<mode><V_cvtto>)2): Likewise
10101         (fixuns_trunc<mode><V_cvtto)2): Likewise.
10102         (neon_vcvt<mode>): Likewise.
10103         (neon_vcvtv4sfv4hf): Likewise.
10104         (neon_vcvtv4hfv4sf): Likewise.
10105         (neon_vcvt_n<mode>): Likewise.
10106         (neon_vmovn<mode>): Likewise.
10107         (neon_vqmovn<mode>): Likewise.
10108         (neon_vqmovun<mode>): Likewise.
10109         (neon_vmovl<mode>): Likewise.
10110         (neon_vmul_lane<mode>): Likewise.
10111         (neon_vmull_lane<mode>): Likewise.
10112         (neon_vqdmull_lane<mode>): Likewise.
10113         (neon_vqdmulh_lane<mode>): Likewise.
10114         (neon_vmla_lane<mode>): Likewise.
10115         (neon_vmlal_lane<mode>): Likewise.
10116         (neon_vqdmlal_lane<mode>): Likewise.
10117         (neon_vmls_lane<mode>): Likewise.
10118         (neon_vmlsl_lane<mode>): Likewise.
10119         (neon_vqdmlsl_lane<mode>): Likewise.
10120         (neon_vext<mode>): Likewise.
10121         (neon_vrev64<mode>): Likewise.
10122         (neon_vrev32<mode>): Likewise.
10123         (neon_vrev16<mode>): Likewise.
10124         (neon_vbsl<mode>_internal): Likewise.
10125         (neon_vshl<mode>): Likewise.
10126         (neon_vqshl<mode>): Likewise.
10127         (neon_vshr_n<mode>): Likewise.
10128         (neon_vshrn_n<mode>): Likewise.
10129         (neon_vqshrn_n<mode>): Likewise.
10130         (neon_vqshrun_n<mode>): Likewise.
10131         (neon_vshl_n<mode>): Likewise.
10132         (neon_vqshl_n<mode>): Likewise.
10133         (neon_vqshlu_n<mode>): Likewise.
10134         (neon_vshll_n<mode>): Likewise.
10135         (neon_vsra_n<mode>): Likewise.
10136         (neon_vsri_n<mode>): Likewise.
10137         (neon_vsli_n<mode>): Likewise.
10138         (neon_vtbl1v8qi): Likewise.
10139         (neon_vtbl2v8qi): Likewise.
10140         (neon_vtbl3v8qi): Likewise.
10141         (neon_vtbl4v8qi): Likewise.
10142         (neon_vtbx1v8qi): Likewise.
10143         (neon_vtbx2v8qi): Likewise.
10144         (neon_vtbx3v8qi): Likewise.
10145         (neon_vtbx4v8qi): Likewise.
10146         (neon_vtrn<mode>_internal): Likewise.
10147         (neon_vzip<mode>_internal): Likewise.
10148         (neon_vuzp<mode>_internal): Likewise.
10149         (neon_vld1<mode>): Likewise.
10150         (neon_vld1_lane<mode>): Likewise.
10151         (neon_vld1_dup<mode>): Likewise.
10152         (neon_vld1_dupv2di): Likewise.
10153         (neon_vst1<mode>): Likewise.
10154         (neon_vst1_lane<mode>): Likewise.
10155         (neon_vld2<mode>): Likewise.
10156         (neon_vld2_lane<mode>): Likewise.
10157         (neon_vld2_dup<mode>): Likewise.
10158         (neon_vst2<mode>): Likewise.
10159         (neon_vst2_lane<mode>): Likewise.
10160         (neon_vld3<mode>): Likewise.
10161         (neon_vld3qa<mode>): Likewise.
10162         (neon_vld3qb<mode>): Likewise.
10163         (neon_vld3_lane<mode>): Likewise.
10164         (neon_vld3_dup<mode>): Likewise.
10165         (neon_vst3<mode>): Likewise.
10166         (neon_vst3qa<mode>): Likewise.
10167         (neon_vst3qb<mode>): Likewise.
10168         (neon_vst3_lane<mode>): Likewise.
10169         (neon_vld4<mode>): Likewise.
10170         (neon_vld4qa<mode>): Likewise.
10171         (neon_vld4qb<mode>): Likewise.
10172         (neon_vld4_lane<mode>): Likewise.
10173         (neon_vld4_dup<mode>): Likewise.
10174         (neon_vst4<mode>): Likewise.
10175         (neon_vst4qa<mode>): Likewise.
10176         (neon_vst4qb<mode>): Likewise.
10177         (neon_vst4_lane<mode>): Likewise.
10178         (neon_vec_unpack<US>_lo_<mode>): Likewise.
10179         (neon_vec_unpack<US>_hi_<mode>): Likewise.
10180         (neon_vec_<US>mult_lo_<mode>): Likewise.
10181         (neon_vec_<US>mult_hi_<mode>): Likewise.
10182         (neon_vec_<US>shiftl_<mode>): Likewise.
10183         (neon_unpack<US>_<mode>): Likewise.
10184         (neon_vec_<US>mult_<mode>): Likewise.
10185         (vec_pack_trunc_<mode>): Likewise.
10186         (neon_vec_pack_trunk_<mode>): Likewise.
10187         (neon_vabd<mode>_2): Likewise.
10188         (neon_vabd<mode>_3): Likewise.
10189         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
10190         (thumb2_movsi_vfp): Likewise.
10191         (movdi_vfp): Likewise.
10192         (movdi_vfp_cortexa8): Likewise.
10193         (movhf_vfp_neon): Likewise.
10194         (movhf_vfp): Likewiwse.
10195         (movsf_vfp): Likewiwse.
10196         (thumb2_movsf_vfp): Likewiwse.
10197         (movdf_vfp): Likewise.
10198         (thumb2_movdf_vfp): Likewise.
10199         (movsfcc_vfp): Likewise.
10200         (thumb2_movsfcc_vfp): Likewise.
10201         (movdfcc_vfp): Likewise.
10202         (thumb2_movdfcc_vfp): Likewise.
10203         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
10204         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
10205         (v10_v2c): Likewise.
10206         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
10207         attribute change.
10208         (cortex_a15_neon_int_2): Likewise.
10209         (cortex_a15_neon_int_3): Likewise.
10210         (cortex_a15_neon_int_4): Likewise.
10211         (cortex_a15_neon_int_5): Likewise.
10212         (cortex_a15_neon_vqneg_vqabs): Likewise.
10213         (cortex_a15_neon_vmov): Likewise.
10214         (cortex_a15_neon_vaba): Likewise.
10215         (cortex_a15_neon_vaba_qqq): Likewise.
10216         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
10217         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
10218         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
10219         scalar_64_32_long_scalar): Likewise.
10220         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
10221         (cortex_a15_neon_mla_qqq_8_16): Likewise.
10222         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
10223         lotype_qdd_64_32_long): Likewise.
10224         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
10225         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
10226         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
10227         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
10228         (cortex_a15_neon_shift_1): Likewise.
10229         (cortex_a15_neon_shift_2): Likewise.
10230         (cortex_a15_neon_shift_3): Likewise.
10231         (cortex_a15_neon_vshl_ddd): Likewise.
10232         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
10233         (cortex_a15_neon_vsra_vrsra): Likewise.
10234         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
10235         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
10236         (cortex_a15_neon_fp_vmul_ddd): Likewise.
10237         (cortex_a15_neon_fp_vmul_qqd): Likewise.
10238         (cortex_a15_neon_fp_vmla_ddd): Likewise.
10239         (cortex_a15_neon_fp_vmla_qqq): Likewise.
10240         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
10241         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
10242         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
10243         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
10244         (cortex_a15_neon_bp_simple): Likewise.
10245         (cortex_a15_neon_bp_2cycle): Likewise.
10246         (cortex_a15_neon_bp_3cycle): Likewise.
10247         (cortex_a15_neon_vld1_1_2_regs): Likewise.
10248         (cortex_a15_neon_vld1_3_4_regs): Likewise.
10249         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
10250         (cortex_a15_neon_vld2_4_regs): Likewise.
10251         (cortex_a15_neon_vld3_vld4): Likewise.
10252         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
10253         (cortex_a15_neon_vst1_3_4_regs): Likewise.
10254         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
10255         (cortex_a15_neon_vst3_vst4): Likewise.
10256         (cortex_a15_neon_vld1_vld2_lane): Likewise.
10257         (cortex_a15_neon_vld3_vld4_lane" 10
10258         (cortex_a15_neon_vst1_vst2_lane): Likewise.
10259         (cortex_a15_neon_vst3_vst4_lane): Likewise.
10260         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
10261         (cortex_a15_neon_ldm_2): Likewise.0
10262         (cortex_a15_neon_stm_2): Likewise.
10263         (cortex_a15_neon_mcr): Likewise.
10264         (cortex_a15_neon_mcr_2_mcrr): Likewise.
10265         (cortex_a15_neon_mrc): Likewise.
10266         (cortex_a15_neon_mrrc): Likewise.
10267         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
10268         change.
10269         (cortex_a15_alu_shift): Likewise.
10270         (cortex_a15_alu_shift_reg): Likewise.
10271         (cortex_a15_mult32): Likewise.
10272         (cortex_a15_mult64): Likewise.
10273         (cortex_a15_block): Likewise.
10274         (cortex_a15_branch): Likewise.
10275         (cortex_a15_load1): Likewise.
10276         (cortex_a15_load3): Likewise.
10277         (cortex_a15_store1): Likewise.
10278         (cortex_a15_store3): Likewise.
10279         (cortex_a15_call): Likewise.
10280         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
10281         (cortex_a5_f2r): Likewise.
10282         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
10283         change.
10284         (cortex_a53_f2r): Likewise.
10285         * config/arm/cortex-a7.md
10286         (cortex_a7_branch): Update for attribute change.
10287         (cortex_a7_call): Likewise.
10288         (cortex_a7_alu_imm): Likewise.
10289         (cortex_a7_alu_reg): Likewise.
10290         (cortex_a7_alu_shift): Likewise.
10291         (cortex_a7_mul): Likewise.
10292         (cortex_a7_load1): Likewise.
10293         (cortex_a7_store1): Likewise.
10294         (cortex_a7_load2): Likewise.
10295         (cortex_a7_store2): Likewise.
10296         (cortex_a7_load3): Likewise.
10297         (cortex_a7_store3): Likewise.
10298         (cortex_a7_load4): Likewise.
10299         (cortex_a7_store4): Likewise.
10300         (cortex_a7_fpalu): Likewise.
10301         (cortex_a7_fconst): Likewise.
10302         (cortex_a7_fpmuls): Likewise.
10303         (cortex_a7_neon_mul): Likewise.
10304         (cortex_a7_fpmacs): Likewise.
10305         (cortex_a7_neon_mla: Likewise.
10306         (cortex_a7_fpmuld: Likewise.
10307         (cortex_a7_fpmacd: Likewise.
10308         (cortex_a7_fpfmad: Likewise.
10309         (cortex_a7_fdivs: Likewise.
10310         (cortex_a7_fdivd: Likewise.
10311         (cortex_a7_r2f: Likewise.
10312         (cortex_a7_f2r: Likewise.
10313         (cortex_a7_f_flags: Likewise.
10314         (cortex_a7_f_loads: Likewise.
10315         (cortex_a7_f_loadd: Likewise.
10316         (cortex_a7_f_stores: Likewise.
10317         (cortex_a7_f_stored: Likewise.
10318         (cortex_a7_neon): Likewise.
10319         * config/arm/cortex-a8-neon.md
10320         (cortex_a8_neon_mrc): Update for attribute change.
10321         (cortex_a8_neon_mrrc): Likewise.
10322         (cortex_a8_neon_int_1): Likewise.
10323         (cortex_a8_neon_int_2): Likewise.
10324         (cortex_a8_neon_int_3): Likewise.
10325         (cortex_a8_neon_int_4): Likewise.
10326         (cortex_a8_neon_int_5): Likewise.
10327         (cortex_a8_neon_vqneg_vqabs): Likewise.
10328         (cortex_a8_neon_vmov): Likewise.
10329         (cortex_a8_neon_vaba): Likewise.
10330         (cortex_a8_neon_vaba_qqq): Likewise.
10331         (cortex_a8_neon_vsma): Likewise.
10332         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
10333         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
10334         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
10335         long_scalar): Likewise.
10336         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
10337         (cortex_a8_neon_mla_qqq_8_16): Likewise.
10338         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
10339         long_scalar_qdd_64_32_long): Likewise.
10340         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
10341         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
10342         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
10343         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
10344         (cortex_a8_neon_shift_1): Likewise.
10345         (cortex_a8_neon_shift_2): Likewise.
10346         (cortex_a8_neon_shift_3): Likewise.
10347         (cortex_a8_neon_vshl_ddd): Likewise.
10348         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
10349         (cortex_a8_neon_vsra_vrsra): Likewise.
10350         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
10351         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
10352         (cortex_a8_neon_fp_vsum): Likewise.
10353         (cortex_a8_neon_fp_vmul_ddd): Likewise.
10354         (cortex_a8_neon_fp_vmul_qqd): Likewise.
10355         (cortex_a8_neon_fp_vmla_ddd): Likewise.
10356         (cortex_a8_neon_fp_vmla_qqq): Likewise.
10357         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
10358         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
10359         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
10360         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
10361         (cortex_a8_neon_bp_simple): Likewise.
10362         (cortex_a8_neon_bp_2cycle): Likewise.
10363         (cortex_a8_neon_bp_3cycle): Likewise.
10364         (cortex_a8_neon_ldr): Likewise.
10365         (cortex_a8_neon_str): Likewise.
10366         (cortex_a8_neon_vld1_1_2_regs): Likewise.
10367         (cortex_a8_neon_vld1_3_4_regs): Likewise.
10368         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
10369         (cortex_a8_neon_vld2_4_regs): Likewise.
10370         (cortex_a8_neon_vld3_vld4): Likewise.
10371         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
10372         (cortex_a8_neon_vst1_3_4_regs): Likewise.
10373         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
10374         (cortex_a8_neon_vst3_vst4): Likewise.
10375         (cortex_a8_neon_vld1_vld2_lane): Likewise.
10376         (cortex_a8_neon_vld3_vld4_lane): Likewise.
10377         (cortex_a8_neon_vst1_vst2_lane): Likewise.
10378         (cortex_a8_neon_vst3_vst4_lane): Likewise.
10379         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
10380         (cortex_a8_neon_mcr): Likewise.
10381         (cortex_a8_neon_mcr_2_mcrr): Likewise.
10382         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
10383         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
10384         change.
10385         (ca9_neon_mrrc): Likewise.
10386         (cortex_a9_neon_int_1): Likewise.
10387         (cortex_a9_neon_int_2): Likewise.
10388         (cortex_a9_neon_int_3): Likewise.
10389         (cortex_a9_neon_int_4): Likewise.
10390         (cortex_a9_neon_int_5): Likewise.
10391         (cortex_a9_neon_vqneg_vqabs): Likewise.
10392         (cortex_a9_neon_vmov): Likewise.
10393         (cortex_a9_neon_vaba): Likewise.
10394         (cortex_a9_neon_vaba_qqq): Likewise.
10395         (cortex_a9_neon_vsma): Likewise.
10396         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
10397         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
10398         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
10399         long_scalar): Likewise.
10400         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
10401         (cortex_a9_neon_mla_qqq_8_16): Likewise.
10402         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
10403         long_scalar_qdd_64_32_long): Likewise.
10404         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
10405         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
10406         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
10407         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
10408         (cortex_a9_neon_shift_1): Likewise.
10409         (cortex_a9_neon_shift_2): Likewise.
10410         (cortex_a9_neon_shift_3): Likewise.
10411         (cortex_a9_neon_vshl_ddd): Likewise.
10412         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
10413         (cortex_a9_neon_vsra_vrsra): Likewise.
10414         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
10415         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
10416         (cortex_a9_neon_fp_vsum): Likewise.
10417         (cortex_a9_neon_fp_vmul_ddd): Likewise.
10418         (cortex_a9_neon_fp_vmul_qqd): Likewise.
10419         (cortex_a9_neon_fp_vmla_ddd): Likewise.
10420         (cortex_a9_neon_fp_vmla_qqq): Likewise.
10421         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
10422         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
10423         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
10424         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
10425         (cortex_a9_neon_bp_simple): Likewise.
10426         (cortex_a9_neon_bp_2cycle): Likewise.
10427         (cortex_a9_neon_bp_3cycle): Likewise.
10428         (cortex_a9_neon_ldr): Likewise.
10429         (cortex_a9_neon_str): Likewise.
10430         (cortex_a9_neon_vld1_1_2_regs): Likewise.
10431         (cortex_a9_neon_vld1_3_4_regs): Likewise.
10432         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
10433         (cortex_a9_neon_vld2_4_regs): Likewise.
10434         (cortex_a9_neon_vld3_vld4): Likewise.
10435         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
10436         (cortex_a9_neon_vst1_3_4_regs): Likewise.
10437         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
10438         (cortex_a9_neon_vst3_vst4): Likewise.
10439         (cortex_a9_neon_vld1_vld2_lane): Likewise.
10440         (cortex_a9_neon_vld3_vld4_lane): Likewise.
10441         (cortex_a9_neon_vst1_vst2_lane): Likewise.
10442         (cortex_a9_neon_vst3_vst4_lane): Likewise.
10443         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
10444         (cortex_a9_neon_mcr): Likewise.
10445         (cortex_a9_neon_mcr_2_mcrr): Likewise.
10446         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
10447         (cortex_a9_fps): Likewise.
10448         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
10449         change.
10450         (cortex_m4_fmuls): Likewise.
10451         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
10452         change.
10453         (cortex_r4_mrc): Likewise.
10454         * config/arm/iterators.md: Update comment referring to neon_type.
10455         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
10456         (iwmmxt_movsi_insn): Likewise.
10457         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
10458         attribute change.
10459         (pj4_core_to_vfp): Likewise.
10460         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
10461         attribute change.
10462         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
10463         (vfp_fstore): Likewise.
10464         * doc/md.texi: Change references to neon_type to refer to type.
10466 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
10468         * tree.h (DECL_BUILT_IN): Fix typo in comment.
10470 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
10472         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
10473         lglobl if not weak.
10475 2013-09-04  Easwaran Raman  <eraman@google.com>
10477         PR middle-end/57370
10478         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
10479         (build_and_add_sum): Use it.
10480         (appears_later_in_bb): Simplify code.
10482 2013-09-04  Teresa Johnson  <tejohnson@google.com>
10484         * dumpfile.c (dump_finish): Don't close stderr/stdout.
10486 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
10488         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
10490 2013-09-04  Jan Hubicka  <jh@suse.cz>
10492         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
10493         * ipa-devirt.c: Include diganostic.h
10494         (odr_type_d): Add types and types_set.
10495         (hash_type_name): Work for types with vtables during LTO.
10496         (odr_hasher::remove): Fix comment; destroy types_set.
10497         (add_type_duplicate): New function,
10498         (get_odr_type): Use it.
10499         (dump_type_inheritance_graph): Dump type duplicates.
10500         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
10501         graph.
10502         * tree.c (types_same_for_odr): Give exact answers on types with
10503         virtual tables.
10505 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
10507         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
10508         explaining their differences.
10510 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
10512         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
10514 2013-09-03  Jeff Law  <law@redhat.com>
10516         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
10517         when not threading through a joiner block.  Pass joiner/no joiner
10518         state to register_jump_thread.
10519         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
10520         state from argument rather than implying on path length.
10521         Dump the entire jump thread path into debugging dump.
10522         * tree-flow.h (register_jump_thread): Update prototype.
10524 2013-08-29  Xinliang David Li  <davidxl@google.com>
10526         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
10527         Remove a trivial gcc_assert.
10529 2013-08-29  Xinliang David Li  <davidxl@google.com>
10531         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
10532         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
10533         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
10534         Delay base decl alignment adjustment.
10535         * tree-vectorizer.c (vect_destroy_datarefs): New function.
10536         * tree-vectorizer.h: New data structure.
10537         (set_dr_misalignment): New function.
10538         (dr_misalignment): Ditto.
10539         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
10540         (vectorizable_load): Ditto.
10541         (ensure_base_align): New function.
10542         (vectorize_loops): Add dbg_cnt support.
10543         (execute_vect_slp): Ditto.
10544         * dbgcnt.def: New debug counter.
10545         * Makefile: New dependency.
10547 2013-09-03  Meador Inge  <meadori@codesourcery.com>
10549         Revert:
10551         2013-08-30  Meador Inge  <meadori@codesourcery.com>
10553         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
10555 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
10557         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
10558         function descriptor.
10560 2013-09-03  Richard Biener  <rguenther@suse.de>
10562         * tree-affine.c (add_elt_to_tree): Fix association issue,
10563         avoid useless converts and make sure to always return a
10564         properly typed result.
10566 2013-09-03  Richard Biener  <rguenther@suse.de>
10568         PR middle-end/57656
10569         * fold-const.c (negate_expr_p): Fix division case.
10570         (negate_expr): Likewise.
10572 2013-09-03  Richard Biener  <rguenther@suse.de>
10574         PR lto/58285
10575         * tree-streamer-out.c: Include tm.h.
10576         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
10578 2013-09-03  Jan Hubicka  <jh@suse.cz>
10580         * tree-profile.c (tree_profiling): Cleanup CFG when done.
10582 2013-09-03  Alan Modra  <amodra@gmail.com>
10584         * config.gcc (powerpc*-*-linux*): Add support for little-endian
10585         multilibs to big-endian target and vice versa.
10586         * config/rs6000/t-linux64: Use := assignment on all vars.
10587         (MULTILIB_EXTRA_OPTS): Remove fPIC.
10588         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
10589         * config/rs6000/t-linux64le: New file.
10590         * config/rs6000/t-linux64bele: New file.
10591         * config/rs6000/t-linux64lebe: New file.
10593 2013-09-02  Jan Hubicka  <jh@suse.cz>
10595         * ipa-inline-transform.c (inline_transform): Do not
10596         optimize_inline_calls when not optimizing.
10598 2013-09-02  Jan Hubicka  <jh@suse.cz>
10600         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
10601         duplicated nodes for assembler names.
10602         * symtab.c (symtab_unregister_node): Do not attempt to unlink
10603         hard registers from assembler name hash.
10605 2013-09-02  Jan Hubicka  <jh@suse.cz>
10607         * ipa-split.c (execute_split_functions): Split externally visible
10608         functions called once.
10610 2013-09-02  Martin Jambor  <mjambor@suse.cz>
10612         PR ipa/58106
10613         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
10614         linked list.  When finding the correct duplicate, also consider also
10615         the caller in additon to its inlined_to node.
10617 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
10619         * config/aarch64/aarch64-simd-builtins.def
10620         (dup_lane_scalar): Remove.
10621         * config/aarch64/aarch64-simd.md
10622         (aarch64_simd_dup): Add 'w->w' alternative.
10623         (aarch64_dup_lane<mode>): Allow for VALL.
10624         (aarch64_dup_lane_scalar<mode>): Remove.
10625         (aarch64_dup_lane_<vswap_width_name><mode>): New.
10626         (aarch64_get_lane_signed<mode>): Add w->w altenative.
10627         (aarch64_get_lane_unsigned<mode>): Likewise.
10628         (aarch64_get_lane<mode>): Likewise.
10629         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
10630         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
10631         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
10632         (VCON): Change container of V2SF.
10633         (vswap_width_name): Likewise.
10634         * config/aarch64/arm_neon.h
10635         (__aarch64_vdup_lane_any): New.
10636         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
10637         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
10638         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
10640 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
10642         PR middle-end/56382
10643         * expr.c (emit_move_complex): Do not move complex FP values as parts if
10644         the source or the destination is a single hard register.
10646 2013-09-02  Richard Biener  <rguenther@suse.de>
10648         PR middle-end/57511
10649         * tree-scalar-evolution.c (instantiate_scev_name): Allow
10650         non-linear SCEVs.
10652 2013-09-02  Richard Biener  <rguenther@suse.de>
10654         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
10655         arithmetic to sizetype.
10657 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
10659         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
10660         Find auto-increment use both before and after candidate.
10662 2013-09-02  Marek Polacek  <polacek@redhat.com>
10664         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
10666 2013-09-01  Jan Hubicka  <jh@suse.cz>
10668         * Makefile.in: Add ipa-profile.o
10669         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
10670         * cgraph.c (struct cgraph_propagate_frequency_data,
10671         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
10672         ipa-profile.c; replace cgraph_ by ipa_ prefix.
10673         * cgraph.h (cgraph_propagate_frequency): Remove.
10674         * ipa-inline-analysis.c: Include ipa-utils.h;
10675         drop duplicated cfgloop.h.
10676         (inline_update_callee_summaries): Update.
10677         * ipa-profile.c: New file.
10678         * ipa-utils.h (ipa_propagate_frequency): Declare.
10679         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
10680         data-streamer.h, value-prof.h.
10681         (symtab_remove_unreachable_nodes): Update profile.
10682         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
10683         account_time_size, cmp_counts, dump_histogram,
10684         ipa_profile_generate_summary, ipa_profile_write_summary,
10685         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
10686         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
10687         Move to ipa-profile.c.
10689 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
10691         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
10693 2013-09-01  Jan Hubicka  <jh@suse.cz>
10695         * common.opt (fdevirtualize-speculatively): New function.
10696         * invoke.texi (fdevirtualize-speculatively): Document.
10697         * ipa-devirt.c: Include ipa-inline.h
10698         (likely_target_p): New function.
10699         (ipa_devirt): New function.
10700         (gate_ipa_devirt): New function.
10701         (pass_data_ipa_devirt): New static var.
10702         (pass_ipa_devirt): Likewise.
10703         (make_pass_ipa_devirt): New function.
10704         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
10705         (common_handle_option): Disable devirtualization when
10706         value range profiling is available.
10707         * passes.def (pass_ipa_devirt): Add.
10708         * timever.def (TV_IPA_DEVIRT): New timevar.
10709         * tree-pass.h (make_pass_ipa_devirt):
10711 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
10713         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
10714         include sanitize(undefined).
10716 2013-08-31  Diego Novillo  <dnovillo@google.com>
10718         * Makefile.in (TREE_CORE_H): Define.
10719         (TREE_H): Use.
10720         (GTFILES): Add tree-core.h.
10721         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
10722         size the array.
10723         * tree-core.h: New file.
10724         Move all data structures, enum, typedefs, global
10725         declarations and constants from ...
10726         * tree.h: ... here.
10728 2013-08-31  Jan Hubicka  <jh@suse.cz>
10730         * bulitins.c (expand_builtin): Do not early exit for gcov
10731         instrumented functions.
10733 2013-08-31  Marek Polacek  <polacek@redhat.com>
10735         * ubsan.c: Include tm_p.h.
10737 2013-08-31  Jan Hubicka  <jh@suse.cz>
10739         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
10740         warning.
10742         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
10743         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
10744         * tree-cfg.c (verify_gimple_label): ... here.
10745         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
10746         (ipa_merge_profiles): New function.
10747         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
10748         (lto_input_function_body): Likewise.
10749         * ipa-utils.h (ipa_merge_profiles): Declare.
10750         * lto-streamer.h (lto_input_function_body): Update prototype.
10751         (emit_label_in_global_context_p): Remove.
10752         * lto-symtab.c: Include ipa-utils.h
10753         (lto_cgraph_replace_node): Use ipa_merge_profiles.
10755 2013-08-31  Jan Hubicka  <jh@suse.cz>
10757         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
10759 2013-08-31  Jan Hubicka  <jh@suse.cz>
10761         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
10763 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
10765         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
10766         "cmp" RTX before signed_comparison_operator check to account
10767         for "code" changes.
10769 2013-08-30  Jan Hubicka  <jh@suse.cz>
10771         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
10772         (detect_type_change_1): Rename to ...
10773         (detect_type_change): ... this one; early return on non-polymorphic
10774         types.
10775         (detect_type_change_ssa): Add comp_type parameter; update
10776         use of detect_type_change.
10777         (compute_complex_assign_jump_func): Add param_type parameter;
10778         update use of detect_type_change_ssa.
10779         (compute_complex_ancestor_jump_func): Likewise.
10780         (ipa_get_callee_param_type): New function.
10781         (ipa_compute_jump_functions_for_edge): Compute parameter type;
10782         update calls to the jump function computation functions.
10784 2013-08-30  Teresa Johnson  <tejohnson@google.com>
10785             Steven Bosscher  <steven@gcc.gnu.org>
10787         * cfgrtl.c (fixup_new_cold_bb): New routine.
10788         (commit_edge_insertions): Invoke fixup_partitions.
10789         (find_partition_fixes): New routine.
10790         (fixup_partitions): Ditto.
10791         (verify_hot_cold_block_grouping): Update comments.
10792         (rtl_verify_edges): Invoke find_partition_fixes.
10793         (rtl_verify_bb_pointers): Update comments.
10794         (rtl_verify_bb_layout): Ditto.
10795         * basic-block.h (probably_never_executed_edge_p): Declare.
10796         (fixup_partitions): Ditto.
10797         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
10798         * bb-reorder.c (sanitize_hot_paths): New function.
10799         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
10800         sanitize_hot_paths.
10801         * predict.c (probably_never_executed_edge_p): New routine.
10802         * cfg.c (check_bb_profile): Add partition insanity warnings.
10804 2013-08-30  Meador Inge  <meadori@codesourcery.com>
10806         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
10808 2013-08-30  Marek Polacek  <polacek@redhat.com>
10810         * Makefile.in (ubsan.o): Add.
10811         (c-family/c-ubsan.o): Add.
10812         (builtins.o): Add ubsan.h dependency.
10813         * ubsan.h: New file.
10814         * ubsan.c: New file.
10815         * common.opt: Add -fsanitize=undefined option.
10816         (flag_sanitize): Add variable.
10817         (fsanitize=): Add option.  Add Driver.
10818         (fsanitize=thread): Remove option.
10819         (fsanitize=address): Likewise.
10820         (static-libubsan): New option.
10821         * doc/invoke.texi: Document the new flag and -static-libubsan.
10822         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
10823         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
10824         * builtin-attrs.def (ATTR_COLD): Define.
10825         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
10826         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
10827         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
10828         * flag-types.h (sanitize_code): New enum.
10829         * opts.c (common_handle_option): Parse command line arguments
10830         of -fsanitize=.  Add -fsanitize=unreachable option.
10831         * varasm.c (get_variable_section): Adjust.
10832         (assemble_noswitch_variable): Likewise.
10833         (assemble_variable): Likewise.
10834         (output_constant_def_contents): Likewise.
10835         (categorize_decl_for_section): Likewise.
10836         (place_block_symbol): Likewise.
10837         (output_object_block): Likewise.
10838         * builtins.def: Likewise.
10839         * toplev.c (compile_file): Likewise.
10840         (process_options): Likewise.
10841         * cppbuiltin.c: Likewise.
10842         * tsan.c (tsan_pass): Likewise.
10843         (tsan_gate): Likewise.
10844         (tsan_gate_O0): Likewise.
10845         * cfgexpand.c (partition_stack_vars): Likewise.
10846         (expand_stack_vars): Likewise.
10847         (defer_stack_allocation): Likewise.
10848         (expand_used_vars): Likewise.
10849         * cfgcleanup.c (old_insns_match_p): Likewise.
10850         * asan.c (asan_finish_file): Likewise.
10851         (asan_instrument): Likewise.
10852         (gate_asan): Likewise.
10853         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
10854         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
10855         (asan_global_struct): Use pointer_sized_int_node instead
10856         calling build_nonstandard_integer_type.
10857         (initialize_sanitizer_builtins): Likewise.
10858         (asan_finish_file): Likewise.
10859         * gcc.c: Document %{%:function(args):X}.
10860         (static_spec_functions): Add sanitize.
10861         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
10862         store funcval != NULL there.
10863         (do_spec_1): Adjust handle_spec_function caller.
10864         (handle_braces): Allow %:function(args) as condition.
10865         (sanitize_spec_function): New function.
10866         (ADD_STATIC_LIBUBSAN_LIBS): Define.
10867         (LIBUBSAN_SPEC): Likewise.
10868         (LIBUBSAN_EARLY_SPEC): Likewise.
10869         (SANITIZER_SPEC): Handle libubsan.
10870         (SANITIZER_EARLY_SPEC): Likewise.
10871         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
10872         instead of fsanitize=address.
10873         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
10874         instead of fsanitize=address*.
10875         * builtins.c: Include ubsan.h.
10876         (fold_builtin_0): Instrument __builtin_unreachable.
10877         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
10878         instead of flag_asan.
10879         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
10880         (pointer_sized_int_node): Define.
10881         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
10883 2013-08-30  Mike Stump  <mikestump@comcast.net>
10885         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
10886         with RE patterns.
10888 2013-08-29  Jan Hubicka  <jh@suse.cz>
10890         * cgraph.c (cgraph_function_body_availability): Handle weakref
10891         correctly.
10892         * passes.def: Remove pass_fixup_cfg.
10893         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
10894         track when we need to remove functions.
10895         (gate_ipa_inline): Execute inlining always; add comment why.
10896         (pass_data_ipa_inline): Remove TODO_remove_functions.
10897         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
10898         do not produce summaries.
10899         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
10900         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
10901         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
10902         constant pool and vtable.
10904 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
10906         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
10907         New arm_neon.h's internal macros to specify 64-bit constants.
10908         Avoid using stdint.h's macros.
10910 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
10912         * recog.c (verify_changes): Verify that changes[i].old is non-zero
10913         before applying REG_P.
10915 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
10917         PR tree-optimization/58277
10918         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
10919         after seeing too many stmts with vdef in between dombb and current
10920         bb, invalidate everything.
10922 2013-08-30  Richard Biener  <rguenther@suse.de>
10924         * fold-const.c (fold_single_bit_test): Fix overflow test.
10926 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
10928         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
10929         and which can live in a register, always retrieve the value on entry.
10930         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
10931         passed by invisible reference specially.
10932         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
10933         (vt_add_function_parameter): Correctly deal with a parameter passed by
10934         invisible reference.
10936 2013-08-30  Jan Hubicka  <jh@suse.cz>
10938         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
10940 2013-08-30  Richard Biener  <rguenther@suse.de>
10942         PR tree-optimization/58228
10943         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
10944         allow invariant loads in nested loop vectorization.
10946 2013-08-30  Richard Biener  <rguenther@suse.de>
10948         PR tree-optimization/58223
10949         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
10950         (has_anti_or_output_dependence): ... this and adjust to also
10951         look for output dependences.
10952         (mark_nodes_having_upstream_mem_writes): Adjust.
10953         (rdg_flag_uses): Likewise.
10955 2013-08-30  Richard Biener  <rguenther@suse.de>
10957         PR tree-optimization/58010
10958         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
10959         assert that we have a loop-closed PHI.
10961 2013-08-29  Jan Hubicka  <jh@suse.cz>
10963         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
10964         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
10965         * lto-section-in.c (lto_free_function_in_decl_state): New function.
10966         (lto_free_function_in_decl_state_for_node): New function.
10968 2013-08-29  Xinliang David Li  <davidxl@google.com>
10970         * loop-unroll.c (report_unroll_peel): Minor message change.
10971         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
10972         Emit alignment peeling message with default -fopt-info.
10973         (vect_loop_versioning): Emit loop version info message.
10974         * tree-vectorizer.c (vectorize_loops): Minor message change.
10975         (execute_vect_slp): Ditto.
10977 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
10979         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
10980         of the clone from the DECL_NAME of the original function.
10982 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
10984         * passes.c (register_pass): Add overload.
10985         * tree-pass.h (register_pass): Forward declare it.  Add comment.
10987 2013-08-29  Jan Hubicka  <jh@suse.cz>
10989         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
10990         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
10991         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
10992         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
10993         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
10994         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
10995         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
10996         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
10998 2013-08-29  Teresa Johnson  <tejohnson@google.com>
11000         * dumpfile.c (dump_loc): Output column number.
11001         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
11002         * doc/invoke.texi: Document optall -fopt-info flag.
11003         * profile.c (read_profile_edge_counts): Use new dump framework.
11004         (compute_branch_probabilities): Ditto.
11005         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
11006         when pass not in any opt group.
11007         * pass_manager.h (pass_manager::get_pass_profile): New method.
11008         * value-prof.c (check_counter): Use new dump framework.
11009         (check_ic_target): Ditto.
11010         * coverage.c (get_coverage_counts): Ditto.
11011         (coverage_init): Setup new dump framework.
11013 2013-08-29  Richard Biener  <rguenther@suse.de>
11015         PR tree-optimization/58246
11016         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
11017         handle the dominance check inside a basic-block.
11019 2013-08-29  Richard Biener  <rguenther@suse.de>
11021         PR middle-end/57287
11022         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
11023         of default defs that appear in abnormal PHI nodes.
11025 2013-08-29  Richard Biener  <rguenther@suse.de>
11027         PR tree-optimization/57685
11028         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
11029         single-use operands to avoid exponential complexity.
11031 2013-08-28  Dehao Chen  <dehao@google.com>
11033         * ipa-inline.c (edge_badness): Fix integer underflow.
11035 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
11037         * gtm-builtins.def (_ITM_free): Declare leaf.
11039 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
11041         PR target/58067
11042         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
11043         (*tls_local_dynamic_base_64_largepic): Likewise.
11044         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
11045         Remove predicate from call operand.
11046         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
11047         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
11049 2013-08-28  Jeff Law  <law@redhat.com>
11051         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
11052         checks for the number of predecessors and successors allowed.
11053         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
11054         which require copying a joiner block if there is a request which
11055         is a subpath that requires no joiner block copying.
11057 2013-08-28  Jan Hubicka  <jh@suse.cz>
11059         * lto-streamer-out.c (DFS_write_tree_body): Drop
11060         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
11061         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
11062         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
11063         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
11064         Do not read DECL_ERROR_ISSUED.
11065         (unpack_ts_decl_with_vis_value_fields): Do not read
11066         DECL_DEFER_OUTPUT.
11067         (lto_input_ts_binfo_tree_pointers): Do not read
11068         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
11069         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
11070         write DECL_ERROR_ISSUED..
11071         (pack_ts_decl_with_vis_value_fields): Do not write
11072         DECL_DEFER_OUTPUT.
11073         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
11074         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
11075         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
11076         * tree.h (tree_decl_common): Update comment.
11077         (DECL_ERROR_ISSUED): Remove.
11079 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
11081         PR middle-end/58257
11082         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
11084 2013-08-28  Jan Hubicka  <jh@suse.cz>
11086         * builtins.def (free): Declare leaf.
11088 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
11090         * gdbhooks.py: New.
11091         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
11092         * configure: Regenerate.
11094 2013-08-27  Martin Jambor  <mjambor@suse.cz>
11096         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
11097         (ipa_ancestor_jf_data): Likewise.
11098         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
11099         (ipa_get_jf_pass_through_type_preserved): New function.
11100         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
11101         (ipa_get_jf_ancestor_type_preserved): New function.
11102         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
11103         (ipa_get_jf_ancestor_result): Likewise.
11104         (propagate_vals_accross_pass_through): Use
11105         ipa_get_jf_pass_through_result to do all the value mappings.
11106         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
11107         type_preserved flag.
11108         (ipa_set_jf_cst_copy): New function.
11109         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
11110         (ipa_set_jf_arith_pass_through): Likewise.
11111         (ipa_set_ancestor_jf): Likewise.
11112         (compute_complex_assign_jump_func): Set type_preserved instead of
11113         punting.
11114         (ipa_compute_jump_functions_for_edge): Likewise.
11115         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
11116         (update_jump_functions_after_inlining): Update type_preserved.
11117         Explicitely create jump functions when combining one with pass_through.
11118         (ipa_write_jump_function): Stream the type_preserved flags.
11119         (ipa_read_jump_function): Likewise.
11121 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
11122             Aldy Hernandez  <aldyh@redhat.com>
11124         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
11125         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
11126         * function.h (struct function): Add has_force_vect_loops and
11127         has_simduid_loops.
11128         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
11129         * gimple.c (gimple_build_omp_critical): Add KIND argument and
11130         handle it.
11131         * gimple.def: Update CLAUSES comments.
11132         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
11133         (gimple_build_omp_for): Add argument to prototype.
11134         (gimple_omp_for_kind): New.
11135         (gimple_omp_for_set_kind): New.
11136         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
11137         GOVD_DATA_SHARE_CLASS.
11138         (enum omp_region_type): Add ORT_SIMD.
11139         (gimple_add_tmp_var): Handle ORT_SIMD.
11140         (gimplify_var_or_parm_decl): Same.
11141         (is_gimple_stmt): Same.
11142         (omp_firstprivatize_variable): Same.
11143         (omp_add_variable): Only use splay_tree_insert if lookup failed.
11144         (omp_notice_variable): Handle ORT_SIMD.
11145         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
11146         (omp_check_private): Handle ORT_SIMD.
11147         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
11148         OMP_CLAUSE_SAFELEN.
11149         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
11150         Handle OMP_CLAUSE_LASTPRIVATE.
11151         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
11152         OMP_CLAUSE_SAFELEN.
11153         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
11154         (gimplify_expr): Handle OMP_SIMD.
11155         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
11156         (expand_GOMP_SIMD_VF): New.
11157         (expand_GOMP_SIMD_LAST_LANE): New.
11158         * internal-fn.def (GOMP_SIMD_LANE): New.
11159         (GOMP_SIMD_VF): New.
11160         (GOMP_SIMD_LAST_LANE): New.
11161         * omp-low.c: Include target.h.
11162         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
11163         OMP_CLAUSE_SAFELEN.
11164         (check_omp_nesting_restrictions): Same.
11165         (omp_max_vf): New.
11166         (lower_rec_simd_input_clauses): New.
11167         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
11168         OMP_CLAUSE_LINEAR.
11169         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
11170         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
11171         (expand_omp_build_assign): New.
11172         (expand_omp_for_init_counts): New.
11173         (expand_omp_for_init_vars): New.
11174         (extract_omp_for_update_vars): New.
11175         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
11176         and rewrite accordingly.
11177         (expand_omp_simd): New.
11178         (expand_omp_for): Use expand_omp_simd.
11179         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
11180         (lower_omp_for): Do not lower the body.
11181         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
11182         in their own loops.
11183         * tree-flow.h (find_omp_clause): Remove prototype.
11184         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
11185         forcing vectorization of the loop, or if flag_tree_vectorize.
11186         (gate_tree_if_conversion): Similarly.
11187         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
11188         gimple_build_omp_for.
11189         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
11190         * tree-parloops (create_parallel_loop): Pass kind argument to
11191         gimple_build_omp_for.
11192         * tree-pretty-print.c (dump_omp_clause): Add cases for
11193         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
11194         OMP_CLAUSE__SIMDUID_.
11195         (dump_generic_node): Handle OMP_SIMD.
11196         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
11197         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
11198         unroll OMP_SIMD loops here.
11199         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
11200         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
11201         loop->safelen.
11202         (vect_analyze_data_refs): Handle simd loops.
11203         * tree-vect-loop.c (vectorizable_live_operation): Handle
11204         IFN_GOMP_SIMD*.
11205         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
11206         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
11207         (vectorizable_load): Same.
11208         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
11209         (struct simduid_to_vf): New.
11210         (simduid_to_vf::hash): New.
11211         (simduid_to-vf::equal): New.
11212         (struct simd_array_to_simduid): New.
11213         (simd_array_to_simduid::hash): New.
11214         (simd_array_to_simduid::equal): New.
11215         (adjust_simduid_builtins): New.
11216         (struct note_simd_array_uses_struct): New.
11217         (note_simd_array_uses_cb): New.
11218         (note_simd_array_uses): New.
11219         (vectorize_loops): Handle simd hints and adjust simd builtins
11220         accordingly.
11221         * tree-vectorizer.h (struct _stmt_vec_info): Add
11222         simd_lane_access_p field.
11223         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
11224         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
11225         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
11226         (omp_clause_code_name): Same.
11227         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
11228         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
11229         * tree.def (OMP_SIMD): New entry.
11230         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
11231         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
11232         (OMP_CLAUSE_DECL): Adjust range for new clauses.
11233         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
11234         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
11235         (OMP_CLAUSE_LINEAR_STEP): New.
11236         (OMP_CLAUSE_SAFELEN_EXPR): New.
11237         (OMP_CLAUSE__SIMDUID__DECL): New.
11238         (find_omp_clause): New prototype.
11240 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
11242         * config/i386/driver-i386.c (host_detect_local_cpu): Update
11243         Haswell processor detection.
11245 2013-08-27  Christian Widmer  <shadow@umbrox.de>
11247         PR target/57927
11248         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
11249         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
11250         AVX2 capable processors.
11252 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
11254         * config/aarch64/arm_neon.h: Replace all inline asm implementations
11255         of vget_low_* with implementations in terms of other intrinsics.
11257 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
11259         PR middle-end/57219
11260         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
11261         values to -1, 0 and 1.
11263 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
11265         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
11266         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
11267         (<optab><mode>3_insn): Remove.
11268         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
11269         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
11270         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
11271         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
11272         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
11273         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
11274         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
11275         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
11276         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
11277         (ror<mode>3_insn): Likewise.
11278         * config/aarch64/predicates.md (aarch64_simd_register): New.
11280 2013-08-27  Richard Biener  <rguenther@suse.de>
11282         PR tree-optimization/57521
11283         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
11284         one edge is non-critical.
11285         (find_phi_replacement_condition): Make sure to use a non-critical
11286         edge.  Cleanup and remove old bug workarounds.
11287         (bb_postdominates_preds): Remove.
11288         (if_convertible_loop_p_1): Do not compute post-dominators.
11289         (combine_blocks): Do not free post-dominators.
11290         (main_tree_if_conversion): Likewise.
11291         (pass_data_if_conversion): Add TODO_verify_ssa.
11293 2013-08-27  DJ Delorie  <dj@redhat.com>
11295         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
11297 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
11299         * function.c (assign_parm_find_data_types): Set passed_mode and
11300         nominal_mode to the TYPE_MODE of nominal_type for the built
11301         pointer type in case of the struct-pass-by-reference.
11303 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
11305         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
11306         (UINT16_TYPE): Change default to "unsigned int".
11308         * config/avr/avr.opt (mfract-convert-truncate): New option.
11309         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
11310         is set, round negative fractional integers according to n1169
11311         when converting to integer types.
11313 2013-08-26  Jan Hubicka  <jh@suse.cz>
11315         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
11316         methods can not be called indirectly when their address is not taken.
11318 2013-08-26  Jan Hubicka  <jh@suse.cz>
11320         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
11321         ctor_for_folding.
11323 2013-08-26  Jan Hubicka  <jh@suse.cz>
11325         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
11326         can be unshared.
11328 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
11330         * reload.c (find_valid_class): Allow classes that do not include
11331         FIRST_PSEUDO_REGISTER - 1.
11333 2013-08-26  Jan Hubicka  <jh@suse.cz>
11335         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
11336         fix edge count/frequency when speculation failed; fix type check
11337         for the direct call.
11339 2013-08-26  Jan Hubicka  <jh@suse.cz>
11341         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
11343 2013-08-26  Jan Hubicka  <jh@suse.cz>
11345         * ipa-inline-transform.c (inline_transform): Be ready for basic block
11346         to be changed by edge redirection.
11348 2013-08-26  Jan Hubicka  <jh@suse.cz>
11350         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
11351         formating; add sanity check.
11352         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
11353         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
11354         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
11355         (resolve_noninline_speculation): Update callee keys, too.
11357 2013-08-26  Jan Hubicka  <jh@suse.cz>
11359         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
11360         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
11362 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
11364         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
11365         into proper place.
11367 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
11369         * config/i386/i386.c (ix86_debug_options): Remove prototype.
11370         (x86_64_elf_select_section): Ditto.
11371         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
11372         arguments.
11373         (ix86_pass_by_reference): Ditto.
11374         (output_set_got): Ditto.
11375         (ix86_unary_operator_ok): Ditto.
11376         (ix86_expand_builtin): Ditto.
11378 2013-08-23  Jan Hubicka  <jh@suse.cz>
11380         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
11382 2013-08-23  Jan Hubicka  <jh@suse.cz>
11384         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
11385         (tree_decl_with_vis): Add FINAL field.
11387 2013-08-23  Jeff Law  <law@redhat.com>
11389         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
11390         the debugging dump when the expression is fully redundant.
11392 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11394         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
11395         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
11396         * pretty-print.c (pp_formatted_text_data): Likewise.
11397         (pp_write_text_to_stream): Likewise.
11398         (pp_write_text_as_dot_label_to_stream): Likewise.
11399         (pp_append_r): Likewise.
11400         (pp_format): Likewise.
11401         (pp_flush): Likewise.
11402         (pp_clear_output_area): Likewise.
11403         (pp_append_text): Likewise.
11404         (pp_formatted_text): Likewise.
11405         (pp_remaining_character_count_for_line): Likewise.
11406         (pp_newline): Likewise.
11407         (pp_character): Likewise.
11408         (output_buffer::~output_buffer): Define.
11409         (pretty_printer::~pretty_printer): Destruct output buffer.
11410         * pretty-print.h (output_buffer::~output_buffer): Declare.
11411         (pretty_printer::~pretty_printer): Declare virtual.
11413 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
11415         PR other/57324
11416         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
11417         HOST_WIDE_INT_M1U): New macros.
11418         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
11419         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
11420         unsigned -1 for lshift.
11421         * cse.c (cse_insn): Likewise.
11422         * double-int.c (rshift_double, lshift_double): Likewise.
11423         * builtins.c (fold_builtin_bitop): Likewise.
11424         * combine.c (force_to_mode): Likewise.
11425         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
11426         * simplify-rtx.c (simplify_const_unary_operation,
11427         simplify_const_binary_operation): Likewise.
11428         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
11429         check_va_list_escapes): Likewise.
11430         * rtlanal.c (nonzero_bits1): Likewise.
11431         * expmed.c (expand_smod_pow2): Likewise.
11432         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
11434 2013-08-23  Jan Hubicka  <jh@suse.cz>
11436         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
11437         as having address taken.
11439 2013-08-23  Jan Hubicka  <jh@suse.cz>
11441         * ipa-utils.h (method_class_type): Declare.
11442         * ipa-devirt.c (method_class_type): Export.
11444         * cgraphunit.c (analyze_functions): Do basic devirtualization;
11445         do not walk base classes of anonymous types.
11447 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
11449         PR rtl-optimization/58220
11450         PR regression/58221
11451         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
11452         handle SEQUENCE insns properly.
11454 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11456         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
11457         definition.
11458         (pp_newline_and_indent): Likewise.
11459         (pp_separate_with): Likewise.
11460         * pretty-print.c (pp_newline_and_flush): Define.
11461         (pp_newline_and_indent): Likewise.
11462         (pp_separate_with): Likewise.
11464 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
11466         PR target/58218
11467         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
11468         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
11470 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
11472         * config/i386/predicates.md (ext_sse_reg_operand): New.
11473         * config/i386/i386.md (*movti_internal): Use
11474         predicate to determine if EVEX is needed.
11475         (*movsi_internal): Ditto.
11476         (*movdf_internal): Ditto.
11477         (*movsf_internal): Ditto.
11478         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
11480 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
11482         PR tree-optimization/58209
11483         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
11484         (find_tail_calls): Give up for pointer result types if m is non-NULL.
11485         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
11486         emit POINTER_PLUS_EXPR.
11487         (create_tailcall_accumulator): For pointer result type accumulate in
11488         sizetype type.
11490 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
11492         * configure.ac: Add backslashes missing from the last change.
11493         * configure: Regenerate.
11495 2013-08-22  Jan Hubicka  <jh@susue.cz>
11497         * ipa.c (function_and_variable_visibility): First remember function
11498         was global and then make it local.
11500 2013-08-22  Julian Brown  <julian@codesourcery.com>
11502         * configure.ac: Add aarch64 to list of arches which use "nop" in
11503         debug_line test.
11504         * configure: Regenerate.
11506 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11508         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
11509         gnu_libc_has_function.
11510         * config/s390/tpf.h: Likewise.
11512 2013-08-22  Jan Hubicka  <jh@susue.cz>
11514         * timevar.c (validate_phases): Add cast.
11516 2013-08-22  Jan Hubicka  <jh@susue.cz>
11518         * timevar.c (validate_phases): Use size_t for memory.
11519         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
11521 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11523         * pretty-print.h (output_buffer::output_buffer): Declare.
11524         (pretty_printer::pretty_printer): Likewise.
11525         (pp_construct): Remove.
11526         * pretty-print.c (output_buffer::output_buffer): Define.
11527         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
11528         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
11529         (print_gimple_expr): Likewise.
11530         (print_gimple_seq): Likewise.
11531         (gimple_dump_bb): Likewise.
11532         * sched-vis.c (dump_value_slim): Likewise.
11533         (dump_insn_slim): Likewise.
11534         (dump_rtl_slim): Likewise.
11535         (str_pattern_slim): Likewise.
11536         * tree-mudflap.c (mf_varname_tree): Likewise.
11537         * graph.c (print_graph_cfg): Likewise.
11538         (start_graph_dump): Likewise.
11539         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
11540         placement-new.
11541         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
11542         pretty printer initialization.
11543         * coretypes.h (diagnostic_context): Remove superflous type alias
11544         declaration.
11545         (pretty_printer): Likewise.  Declare directly as a class.
11546         (pretty_print_info): Remove declaration as class.
11547         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
11548         and pp_clear_output_area.
11549         (asan_add_global): Likewise.
11551 2013-08-22  Jan Hubicka  <jh@suse.cz>
11553         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
11554         * ipa-utils.h (update_type_inheritance_graph): Declare.
11555         (possible_polymorphic_call_target_p): Declare.
11556         (possible_polymorphic_call_target_p): New.
11557         * ipa-devirt.c: Update toplevel comments.
11558         (cached_polymorphic_call_targets): Move up.
11559         (odr_type_d): Move ID down.
11560         (polymorphic_type_binfo_p): Update comment.
11561         (odr_hasher::remove): Likewise;
11562         (get_odr_type): Set anonymous_namespace.
11563         (dump_odr_type): Dump it.
11564         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
11565         (maybe_record_node): Record node in cached_polymorphic_call_targets.
11566         (record_binfo): Add comment.
11567         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
11568         built.
11569         (devirt_node_removal_hook): Do not iCE when cache is freed.
11570         (possible_polymorphic_call_target_p): New predicate.
11571         (update_type_inheritance_graph): New function.
11573 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
11574             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
11575             Sergey Lega  <sergey.s.lega@intel.com>
11576             Anna Tikhonova  <anna.tikhonova@intel.com>
11577             Ilya Tocar  <ilya.tocar@intel.com>
11578             Andrey Turetskiy  <andrey.turetskiy@intel.com>
11579             Ilya Verbin  <ilya.verbin@intel.com>
11580             Kirill Yukhin  <kirill.yukhin@intel.com>
11581             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
11583         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
11584         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
11585         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
11586         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
11587         (OPTION_MASK_ISA_AVX2_UNSET): Update.
11588         (OPTION_MASK_ISA_AVX512F_UNSET): New.
11589         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
11590         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
11591         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
11592         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
11593         OPT_mavx512pf, OPT_mavx512er cases.
11594         * config/i386/constraints.md (v): New constraint.
11595         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
11596         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
11597         (bit_AVX512CD): New.
11598         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
11599         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
11600         * config/i386/i386-c.c (ix86_target_macros_internal):
11601         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
11602         __AVX512PF__.
11603         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
11604         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
11605         * config/i386/i386.c (regclass_map, dbx_register_map)
11606         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
11607         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
11608         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
11609         -mavx512pf options.
11610         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
11611         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
11612         -mavx512pf options.  Fix formatting.
11613         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
11614         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
11615         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
11616         -mavx512cd, -mavx512pf options.
11617         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
11618         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
11619         (ix86_preferred_output_reload_class): Replace SSE_REGS with
11620         ALL_SSE_REGS.
11621         (ix86_hard_regno_mode_ok): Support 512-bit registers.
11622         (ix86_set_reg_reg_cost): Ditto.
11623         (x86_order_regs_for_local_alloc): Ditto.
11624         (MAX_VECT_LEN): Extend to 64-byte.
11625         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
11626         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
11627         (TARGET_AVX512ER, TARGET_AVX512CD): New.
11628         (BIGGEST_ALIGNMENT): Extend to 512-bits.
11629         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
11630         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
11631         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
11632         (SSE_REG_MODE_P): Support new modes.
11633         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
11634         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
11635         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
11636         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
11637         (REG_CLASS_CONTENTS): Add new registers.
11638         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
11639         (EXT_REX_SSE_REGNO_P): New.
11640         (HI_REGISTER_NAMES): Add new registers.
11641         * config/i386/i386.md: Define constants for new registers.
11642         (mode): Add new 512-bit modes.
11643         (prefix): Support evex prefix.
11644         (isa): Support avx512f, noavx512f, fma_avx512f.
11645         (ssemodesuffix): Add new 512-bit modes.
11646         (movxi): New.
11647         (*movxi_internal_avx512f): Ditto.
11648         (*movdi_internal): Replace constraint "x" with the new constraint "v".
11649         Support MODE_XI.
11650         (*movsi_internal): Likewise.
11651         (*movdf_internal): Likewise.
11652         (*movsf_internal): Likewise.
11653         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
11654         (<code><mode>3): Likewise.
11655         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
11656         New.
11657         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
11658         with the new constraint "v".
11659         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
11660         modes.
11661         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
11662         with the new constraint "v".
11663         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
11664         (<sse2>_storedqu<avxsizesuffix>): Likewise.
11665         (*<plusminus_insn><mode>3): Likewise.
11666         (<sse>_vm<plusminus_insn><mode>3): Likewise.
11667         (*mul<mode>3): Likewise.
11668         (<sse>_vmmul<mode>3): Likewise.
11669         (<sse>_div<mode>3): Likewise.
11670         (<sse>_vmdiv<mode>3): Likewise.
11671         (<sse>_sqrt<mode>2): Likewise.
11672         (<sse>_vmsqrt<mode>2): Likewise.
11673         (*<code><mode>3_finite): Likewise.
11674         (*<code><mode>3) <smaxmin>: Likewise.
11675         (<sse>_vm<code><mode>3): Likewise.
11676         (*<code><mode>3) <any_logic>: Likewise.
11677         (*fma_fmadd_<mode>): Likewise.
11678         (*fma_fmsub_<mode>): Likewise.
11679         (*fma_fnmadd_<mode>): Likewise.
11680         (*fma_fnmsub_<mode>): Likewise.
11681         (*fma_fmaddsub_<mode>): Likewise.
11682         (*fma_fmsubadd_<mode>): Likewise.
11683         (*fmai_fmadd_<mode>): Likewise.
11684         (*fmai_fmsub_<mode>): Likewise.
11685         (*fmai_fnmadd_<mode>): Likewise.
11686         (*fmai_fnmsub_<mode>): Likewise.
11687         (sse_cvtsi2ss): Likewise.
11688         (sse_cvtsi2ssq): Likewise.
11689         (sse_cvtss2si): Likewise.
11690         (sse_cvtss2si_2): Likewise.
11691         (sse_cvtss2siq): Likewise.
11692         (sse_cvtss2siq_2): Likewise.
11693         (sse_cvttss2si): Likewise.
11694         (sse_cvtss2siq_2): Likewise.
11695         (float<sseintvecmodelower><mode>2): Likewise.
11696         (sse2_cvtsd2si_2): Likewise.
11697         (sse2_cvtsd2siq_2): Likewise.
11698         (*<plusminus_insn><mode>3): Likewise.
11699         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
11700         (*<sse4_1_avx2>_mul<mode>3): Likewise.
11701         (ashr<mode>3): Likewise.
11702         (<shift_insn><mode>3): Likewise.
11703         (avx2_<code><mode>3): Likewise.
11704         (*avx2_<code><mode>3): Likewise.
11705         (*andnot<mode>3): Likewise.
11706         (*<code><mode>3) <any_logic>: Likewise.
11707         (abs<mode>2): Likewise.
11708         (avx2_permvar<mode>): Likewise.
11709         (avx2_perm<mode>_1): Likewise.
11710         (*avx_vpermilp<mode>): Likewise.
11711         (avx_vpermilvar<mode>3): Likewise.
11712         (avx2_ashrv<mode>): Likewise.
11713         (avx2_<shift_insn>v<mode>): Likewise.
11714         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
11715         -mavx512cd.
11716         * doc/rtl.texi: Document XImode.
11718 2013-08-21  Jeff Law  <law@redhat.com>
11720         * tree-flow.h (register_jump_thread): Pass vector of edges
11721         instead of each important edge.
11722         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
11723         thread path into a vector and pass that to register_jump_thread.
11724         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
11725         passed in edge vector to the current 3-edge form.
11727         Revert:
11728         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
11730         * dce.c (fini_dce): Call df_analyze again just in case
11731         delete_unmarked_insns removed anything.
11733 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
11735         * reload.h (struct reg_equivs): Rename to ..
11736         (struct reg_equivs_s): .. this.
11738 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
11740         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
11742 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11744         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
11746 2013-08-21  Jeff Law  <law@redhat.com>
11748         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
11749         simplify assignments too.  If the RHS collapses to a singleton
11750         range, then return the value for the range.
11752 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
11754         * config/i386/sse.md (V16): Rename to...
11755         (VMOVE): this.
11756         (mov<mode>): Update iterator name.
11757         (*mov<mode>_internal): Ditto.
11758         (push<mode>1): Ditto.
11759         (movmisalign<mode>): Ditto.
11761 2013-08-20  Jan Hubicka  <jh@suse.cz>
11763         PR bootstrap/58186
11764         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
11765         entry for direct edges.
11766         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
11768 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
11770         Revert my last two changes, r201865 and r201864:
11772         Revert r201865:
11773         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
11775         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
11776         instances can own GC refs.
11778         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
11779         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
11780         (gcc::context::gt_pch_nx): Likewise.
11781         (gcc::context::gt_pch_nx):  Likewise.
11782         * ggc.h (gt_ggc_mx <T>): New.
11783         (gt_pch_nx_with_op <T>): New.
11784         (gt_pch_nx <T>): New.
11785         * passes.c (opt_pass::gt_ggc_mx): New.
11786         (opt_pass::gt_pch_nx): New.
11787         (opt_pass::gt_pch_nx_with_op): New.
11788         (pass_manager::gt_ggc_mx): New.
11789         (pass_manager::gt_pch_nx): New.
11790         (pass_manager::gt_pch_nx_with_op): New.
11791         (pass_manager::operator new): Use
11792         ggc_internal_cleared_alloc_stat rather than xcalloc.
11793         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
11794         (pass_manager::gt_ggc_mx): New.
11795         (pass_manager::gt_pch_nx): New.
11796         (pass_manager::gt_pch_nx_with_op): New.
11797         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
11798         (opt_pass::operator new): New.
11799         (opt_pass::gt_ggc_mx): New.
11800         (opt_pass::gt_pch_nx): New.
11801         (opt_pass::gt_pch_nx_with_op): New.
11803         Revert r201864:
11804         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
11806         * Makefile.in (GTFILES): Add context.h.
11807         * context.c (gcc::context::operator new): New.
11808         (gcc::context::gt_ggc_mx): New.
11809         (gcc::context::gt_pch_nx): New.
11810         (gcc::context::gt_pch_nx): New.
11811         * context.h (gcc::context): Add GTY((user)) marking.
11812         (gcc::context::operator new): New.
11813         (gcc::context::gt_ggc_mx): New.
11814         (gcc::context::gt_pch_nx): New.
11815         (gcc::context::gt_pch_nx): New.
11816         (g): Add GTY marking.
11817         (gt_ggc_mx (gcc::context *)): New.
11818         (gt_pch_nx (gcc::context *)): New.
11819         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
11820         void *cookie)): New.
11821         * gengtype.c (open_base_files) <ifiles>: Add context.h.
11823 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
11825         * dce.c (fini_dce): Call df_analyze again just in case
11826         delete_unmarked_insns removed anything.
11828 2013-08-20  Teresa Johnson  <tejohnson@google.com>
11830         PR rtl-optimizations/57451
11831         * final.c (reemit_insn_block_notes): Prevent lexical blocks
11832         from crossing split section boundaries.
11834 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
11836         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
11837         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
11838         with MULTLIB_DEFAULTS.
11840 2013-08-20  Nick Clifton  <nickc@redhat.com>
11842         * target.def (narrow_volatile_bitfield): Note that the default
11843         value is false, not !TARGET_STRICT_ALIGN.
11844         * doc/tm.texi: Regenerate.
11846 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
11848         Fix LIB_SPEC for systems without libpthread.
11850         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
11851         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
11852         for Android.
11853         * config/i386/linux-common.h: Likewise.
11854         * config/mips/linux-common.h: Likewise.
11856 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
11858         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
11859         checks.
11861 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
11863         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
11864         instances can own GC refs.
11866         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
11867         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
11868         (gcc::context::gt_pch_nx): Likewise.
11869         (gcc::context::gt_pch_nx):  Likewise.
11870         * ggc.h (gt_ggc_mx <T>): New.
11871         (gt_pch_nx_with_op <T>): New.
11872         (gt_pch_nx <T>): New.
11873         * passes.c (opt_pass::gt_ggc_mx): New.
11874         (opt_pass::gt_pch_nx): New.
11875         (opt_pass::gt_pch_nx_with_op): New.
11876         (pass_manager::gt_ggc_mx): New.
11877         (pass_manager::gt_pch_nx): New.
11878         (pass_manager::gt_pch_nx_with_op): New.
11879         (pass_manager::operator new): Use
11880         ggc_internal_cleared_alloc_stat rather than xcalloc.
11881         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
11882         (pass_manager::gt_ggc_mx): New.
11883         (pass_manager::gt_pch_nx): New.
11884         (pass_manager::gt_pch_nx_with_op): New.
11885         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
11886         (opt_pass::operator new): New.
11887         (opt_pass::gt_ggc_mx): New.
11888         (opt_pass::gt_pch_nx): New.
11889         (opt_pass::gt_pch_nx_with_op): New.
11891 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
11893         * Makefile.in (GTFILES): Add context.h.
11894         * context.c (gcc::context::operator new): New.
11895         (gcc::context::gt_ggc_mx): New.
11896         (gcc::context::gt_pch_nx): New.
11897         (gcc::context::gt_pch_nx): New.
11898         * context.h (gcc::context): Add GTY((user)) marking.
11899         (gcc::context::operator new): New.
11900         (gcc::context::gt_ggc_mx): New.
11901         (gcc::context::gt_pch_nx): New.
11902         (gcc::context::gt_pch_nx): New.
11903         (g): Add GTY marking.
11904         (gt_ggc_mx (gcc::context *)): New.
11905         (gt_pch_nx (gcc::context *)): New.
11906         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
11907         void *cookie)): New.
11908         * gengtype.c (open_base_files) <ifiles>: Add context.h.
11910 2013-08-20  Alan Modra  <amodra@gmail.com>
11912         PR target/57865
11913         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
11914         (rs6000_emit_epilogue): Likewise.
11916 2013-08-19  Dehao Chen  <dehao@google.com>
11918         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
11920 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
11921             Jakub Jelinek  <jakub@redhat.com>
11923         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
11924         (BUILT_IN_FABSD64): Likewise.
11925         (BUILT_IN_FABSD128): Likewise.
11926         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
11927         (fold_builtin_1): Likewise.
11928         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
11929         destination and source operands.
11930         (*abstd2_fpr): Likewise.
11931         (*nabstd2_fpr): Likewise.
11933 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
11935         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
11936         JUMP_P and INSN_P.
11938 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
11940         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
11941         incomplete.
11943 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
11945         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
11946         * builtins.c (default_libc_has_function): New.
11947         (gnu_libc_has_function): Ditto.
11948         (no_c99_libc_has_function): Ditto.
11949         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
11950         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
11951         (fold_builtin_sincos): Likewise.
11952         (fold_builtin_cexp): Likewise.
11953         * builtins.def (DEF_C94_BUILTIN): Likewise.
11954         (DEF_C99_BUILTIN): Likewise.
11955         (DEF_C99_C90RES_BUILTIN): Likewise.
11956         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
11957         definitions to using this define.
11958         * config/darwin-protos.h (darwin_libc_has_function): New.
11959         * config/darwin.c (darwin_libc_has_function): Ditto.
11960         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
11961         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
11962         * config/darwin.h: Ditto.
11963         * config/elfos.h: Ditto.
11964         * config/freebsd.h: Ditto.
11965         * config/i386/cygming.h: Ditto.
11966         * config/i386/djgpp.h: Ditto.
11967         * config/i386/i386-interix.h: Ditto.
11968         * config/microblaze/microblaze.h: Ditto.
11969         * config/mmix/mmix.h: Ditto.
11970         * config/gnu-user.h: Ditto.
11971         * config/ia64/hpux.h: Ditto.
11972         * config/pa/pa-hpux.h: Ditto.
11973         * config/pdp11/pdp11.h: Ditto.
11974         * config/picochip/picochip.h: Ditto.
11975         * config/linux.h: Ditto.
11976         * config/netbsd.h: Ditto.
11977         * config/openbsd.h: Ditto.
11978         * config/rs6000/aix43.h: Ditto.
11979         * config/rs6000/aix51.h: Ditto.
11980         * config/rs6000/aix52.h: Ditto.
11981         * config/rs6000/aix53.h: Ditto.
11982         * config/rs6000/aix61.h: Ditto.
11983         * config/rs6000/darwin.h: Ditto.
11984         * config/rs6000/linux.h: Ditto.
11985         * config/rs6000/linux64.h: Ditto.
11986         * config/s390/tpf.h: Ditto.
11987         * config/sol2-10.h: Ditto.
11988         * config/sol2.h: Ditto.
11989         * config/vms/vms.h: Ditto.
11990         * config/vxworks.h: Ditto.
11991         * config/linux-android.c (linux_android_libc_has_function):
11992         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
11993         * config/linux-protos.h (linux_android_libc_has_function):
11994         New declaration.
11995         * config/i386/i386.c (ix86_libc_has_function): New.
11996         * config/i386/i386-protos.h
11997         (ix86_libc_has_function): New declaration.
11998         * config/i386/i386.md
11999         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
12000         ("isinf<mode>2): Likewise.
12001         * convert.c (convert_to_integer): Using new target hook
12002         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
12003         TARGET_C99_FUNCTIONS.
12004         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
12005         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
12006         * coretypes.h (function_class): New enum for different
12007         classes of functions.
12008         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
12009         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
12010         (TARGET_HAS_SINCOS): Likewise.
12011         (TARGET_LIBC_HAS_FUNCTION): New.
12012         * doc/tm.texi: Regenerated.
12013         * targhooks.h (default_libc_has_function): New declaration.
12014         (no_c99_libc_has_function): Ditto.
12015         (gnu_libc_has_function): Ditto.
12016         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
12017         and TARGET_HAS_SINCOS.
12019 2013-08-18  Jan Hubicka  <jh@suse.cz>
12021         * Makeifle-in (ipa-devirt.o): New.
12022         (GTFILES): Add ipa-utils.h and ipa-devirt.c
12023         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
12024         (analyze_functions): Look into possible targets of polymorphic call.
12025         * dumpfile.c (dump_files): Add type-inheritance dump.
12026         * dumpfile.h (TDI_inheritance): New.
12027         * ipa-devirt.c: New file.
12028         * ipa-utils.h (odr_type_d): Forward declare.
12029         (odr_type): New type.
12030         (build_type_inheritance_graph): Declare.
12031         (possible_polymorphic_call_targets): Declare and introduce inline
12032         variant when only edge is pased.
12033         (dump_possible_polymorphic_call_targets): Likewise.
12034         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
12035         * tree.c (type_in_anonymous_namespace_p): Break out from ...
12036         (types_same_for_odr): ... here.
12037         * tree.h (type_in_anonymous_namespace_p): Declare.
12039 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
12041         PR tree-optimization/58006
12042         * tree-parloops.c (take_address_of): Don't ICE if get_name
12043         returns NULL.
12044         (eliminate_local_variables_stmt): Remove clobber stmts.
12046 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
12048         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
12049         error message is issued for an alias to undefined symbol.
12051 2013-08-18  Jan Hubicka  <jh@suse.cz>
12053         * cgraph.c (cgraph_create_indirect_edge): Discover
12054         polymorphic calls and record basic info into indirect_info.
12055         * gimple-fold.c (gimple_fold_call): When doing BINFO based
12056         devirtualization, ignore objc function calls.
12057         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
12058         call with no parm index info.
12059         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
12060         * tree.c (virtual_method_call_p): New function.
12061         * tree.h (virtual_method_call_p): Declare.
12063 2013-08-16  Jan Hubicka  <jh@suse.cz>
12065         PR middle-end/58179
12066         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
12068 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
12070         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
12071         attribute.
12073 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
12075         * gengtype.c (type_for_name): Add special-case support for
12076         locating types within the "gcc::" namespace.
12077         (open_base_files): Emit a "using namespace gcc" directive.
12079 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
12081         PR target/58160
12082         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
12083         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
12085         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
12086         array instead of each individual operand as a separate argument.
12087         (emit_fusion_gpr_load): Likewise.
12088         (expand_fusion_gpr_load): Add new function declaration.
12090         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
12091         signature to have the operands passed as an array, instead of as
12092         separate arguments.  Allow ZERO_EXTEND to be in the memory
12093         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
12094         depend on the register live/dead flags when peepholes are run.
12095         (expand_fusion_gpr_load): New function to be called from the
12096         peephole2 pass, to change the register that addis sets to be the
12097         target register.
12098         (emit_fusion_gpr_load): Change the calling signature to have the
12099         operands passed as an array, instead of as separate arguments.
12100         Allow ZERO_EXTEND to be in the memory address, and also
12101         SIGN_EXTEND if -mpower8-fusion-sign.
12103         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
12104         unspec enumeration.
12105         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
12106         adjust the register addis loads up in the peephole2 pass.  Do not
12107         depend on the register live/dead state when the peephole pass is done.
12109 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
12111         * gengtype.c (create_user_defined_type): Ensure that the kind
12112         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
12113         declaration is seen before the GTY((user)) marking.
12115 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
12117         PR target/58105
12118         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
12120 2013-08-16  Jan Hubicka  <jh@suse.cz>
12122         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
12123         arugment expected_type.
12124         (gimple_fold_call): Use it.
12125         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
12126         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
12127         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
12128         (try_make_edge_direct_virtual_call): Likewise.
12129         * tree.c (obj_type_ref_class): New.
12130         * tree.h (obj_type_ref_class): Use it.
12132 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
12134         * sched-vis.c (rtl_slim_pp_initialized): Remove.
12135         (rtl_slim_pp): Likewise.
12136         (init_rtl_slim_pretty_print): Likewise.
12137         (dump_value_slim):  Don't call it.  Use local pretty printer.
12138         (dump_insn_slim): Likewise.
12139         (dump_rtl_slim): Likewise.
12140         (str_pattern_slim): Likewise.
12141         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
12142         Simplify.
12144 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
12146         PR tree-optimization/58164
12147         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
12148         walk gimple_goto_dest of GIMPLE_GOTO.
12150         PR tree-optimization/58165
12151         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
12152         bi_call must be the last stmt in a bb, don't split_block, instead
12153         use fallthru edge from it and give up if there is none.
12154         Release conds vector when returning early.
12156 2013-08-14  Xinliang David Li  <davidxl@google.com>
12158         * config/i386/i386.c (ix86_option_override_internal):
12159         Remove unused variable and field.
12161 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12163         PR target/57949
12164         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
12165         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
12166         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
12167         and Linux, correct BLKmode alignment when 128-bit alignment is
12168         required and compatibility flag is not set.
12169         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
12170         for zero-size arguments when compatibility flag is not set.
12172 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
12174         PR tree-optimization/58145
12175         * tree-sra.c (build_ref_for_offset): If prev_base has
12176         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
12178 2013-08-14  Xinliang David Li  <davidxl@google.com>
12180         * config/i386/i386.c (ix86_option_override_internal):
12181         Fix uninitialized variable error.
12183 2013-08-14  Xinliang David Li  <davidxl@google.com>
12185         * config/i386/i386.opt: Define two new options.
12186         * config/i386/x86-tune.def: Add arch selector field in macros.
12187         * config/i386/i386.h: Adjust macro definition.
12188         * config/i386/i386.c (ix86_option_override_internal):
12189         Refactor the code.
12190         (parse_mtune_ctrl_str): New function.
12191         (set_ix86_tune_features): New function.
12192         (ix86_function_specific_restore): Call the new helper function.
12194 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
12196         PR rtl-optimization/57662
12197         * sel-sched.c (code_motion_process_successors): When the current insn
12198         is removed after the recursive traversal, break from the loop.
12199         Add comments and debug printouts.
12201 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
12202             Alexandre Oliva  <aoliva@redhat.com>
12204         PR target/58067
12205         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
12206         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
12207         there also UNSPEC_PLTOFF.
12209 2013-08-14  Marek Polacek  <polacek@redhat.com>
12211         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
12212         NUM_CONDITIONS bit positions.
12214 2013-08-13  Cary Coutant  <ccoutant@google.com>
12216         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
12217         (attr_checksum): Hash vector contents instead of pointer.
12218         (attr_checksum_ordered): Likewise.
12220 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
12222         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
12223         when Pmode != word_mode.  Add length_address attribute.
12224         (sse3_monitor_<mode>): Merge from sse3_monitor and
12225         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
12226         Pmode != word_mode.  Update insn length attribute.
12227         * config/i386/i386.c (ix86_option_override_internal): Update
12228         ix86_gen_monitor selection for merged sse3_monitor insn.
12230 2013-08-13  Julian Brown  <julian@codesourcery.com>
12232         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
12233         perform invalid legitimization on greater-than-word-size modes for
12234         TARGET_E500_DOUBLE.
12236 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
12238         * ira.c (setup_class_translate_array): Use aclass instead of cl
12239         for classes not fully covered by allocno classes.
12241 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
12243         PR tree-optimization/57661
12244         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
12245         * tree-inline.c (tree_function_versioning): Initialize it.
12246         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
12247         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
12248         that is not being copied.
12250         PR sanitizer/56417
12251         * asan.c (instrument_strlen_call): Fix typo in comment.
12252         Use char * type even for the lhs of POINTER_PLUS_EXPR.
12254 2013-08-13  Steve Ellcey  <sellcey@mips.com>
12256         * config/mips/mips.md (prefetch): Use lw instead of ld on
12257         loongson in 32bit mode.
12259 2013-08-13  Nick Clifton  <nickc@redhat.com>
12261         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
12263 2013-08-13  Jan Hubicka  <jh@suse.cz>
12265         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
12266         introduced edge; fix typo in sanity check.
12267         (cgraph_resolve_speculation): Export; improve diagnostic.
12268         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
12269         speculation at type mismatch.
12270         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
12271         (cgraph_resolve_speculation): Declare.
12272         (symtab_can_be_discarded): New function.
12273         * value-prof.c (gimple_ic_transform): Remove actual transform code.
12274         * ipa-inline-transform.c (speculation_removed): New global var.
12275         (clone_inlined_nodes): See if speculation can be removed.
12276         (inline_call): If speculations was removed, we growths may not match.
12277         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
12278         (speculation_useful_p): New function.
12279         (resolve_noninline_speculation): New function.
12280         (inline_small_functions): Resolve useless speculations.
12281         * ipa-inline.h (speculation_useful_p): Declare
12282         * ipa.c (can_replace_by_local_alias): Simplify.
12283         (ipa_profile): Produce speculative calls in non-lto, too;
12284         add simple cost model; produce local aliases.
12286 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
12288         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
12289         PASS_MANAGER_H.
12291 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
12293         * config/i386/i386.c (ix86_function_versions): Use error + inform.
12295 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
12297         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
12298         iterator instead of X87MODEF.
12300 2013-08-12  Perez Read  <netfirewall@gmail.com>
12302         PR target/58132
12303         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
12304         operand 0 for intel asm alternative.
12305         (*movabs<mode>_2): Ditto for operand 1.
12307 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
12309         * config/aarch64/arm_none.h
12310         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
12312 2013-08-12  Nick Clifton  <nickc@redhat.com>
12314         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
12316 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
12318         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
12319         expand for QI/HImode operand to produce more effictive code for
12320         unsigned char(short) --> float(double) conversion.
12322 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
12324         * doc/invoke.texi: Mention that -ftls-model does not force the final
12325         model.
12327 2013-08-12  Marek Polacek  <polacek@redhat.com>
12328             Marc Glisse  <marc.glisse@inria.fr>
12330         PR tree-optimization/57980
12331         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
12332         when creating -1 constant.
12334 2013-08-10  Jan Hubicka  <jh@suse.cz>
12336         Workaround binutils PR14342.
12337         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
12338         (gimple_init_edge_profiler): Likewise.
12339         (gimple_gen_ic_func_profiler): Likewise.
12341 2013-08-09  Jan Hubicka  <jh@suse.cz>
12343         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
12345 2013-08-09  Xinliang David Li  <davidxl@google.com>
12347         * config/i386/stringop.def: New file.
12348         * config/i386/stringop.opt: New file.
12349         * config/i386/i386-opts.h: Include stringopt.def.
12350         * config/i386/i386.opt: Include stringopt.opt.
12351         * config/i386/i386.c (ix86_option_override_internal):
12352         Override default size based stringop inline strategies with options.
12353         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
12354         New function.
12356 2013-08-09  Jan Hubicka  <jh@suse.cz>
12358         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
12360 2013-08-09  Jan Hubicka  <jh@suse.cz>
12362         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
12363         CGRAPH_FREQ_MAX.
12364         (dump_cgraph_node): Dump profile-id.
12365         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
12366         and common_target_probability.
12367         * lto-cgraph.c (lto_output_edge): Stream common targets.
12368         (lto_output_node): Stream profile ids.
12369         (input_node): Stream profile ids.
12370         (input_edge): Stream common targets.
12371         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
12372         * ipa.c: Include value-prof.h
12373         (ipa_profile_generate_summary): Turn indirect call statement histograms
12374         into common targets.
12375         (ipa_profile): Turn common targets into speculative edges.
12377 2013-08-09  Jan Hubicka  <jh@suse.cz>
12379         * cgraph.h (cgraph_node): Add profile_id.
12380         * value-prof.c (cgraph_node_map): Turn into pointer_map.
12381         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
12382         (del_node_map): Update.
12383         (find_func_by_funcdef_no): Replace by ...
12384         (find_func_by_profile_id): ... this one.
12385         (gimple_ic_transform): Do not remove useful histograms when
12386         speculation is not done; dump info when indirect call removal
12387         can happen at LTO.
12388         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
12389         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
12390         (__gcov_indirect_call_profiler_v2): .. this one.
12391         * profile.h (init_node_map): Update.
12392         * coverage.c (coverage_compute_profile_id): New function.
12393         * coverage.h (coverage_compute_profile_id): Declare.
12394         * tree-profile.c (init_ic_make_global_vars): Make
12395         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
12396         (gimple_init_edge_profiler): Update prototype of
12397         __gcov_indirect_call_profiler.
12398         (gimple_gen_ic_func_profiler): Simplify.
12399         (tree_profiling): Use init_node_map
12401 2013-08-09  Jan Hubicka  <jh@suse.cz>
12403         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
12404         non-speculative refs.
12405         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
12406         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
12407         (cgraph_set_call_stmt): Likewise.
12408         (cgraph_create_edge_1): Fix release checking compilatoin;
12409         clear lto_stmt_uid.
12410         (cgraph_free_edge): Free indirect info.
12411         (cgraph_turn_edge_to_speculative): New function.
12412         (cgraph_speculative_call_info): New function.
12413         (cgraph_make_edge_direct): Return direct edge; handle speculation.
12414         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
12415         (dump_cgraph_node): Dump speculation.
12416         (verify_edge_count_and_frequency): Accept speculative edges.
12417         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
12418         (verify_cgraph_node): Handle speculation.
12419         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
12420         (cgraph_set_call_stmt): Update prototype.
12421         (cgraph_make_edge_direct): Update prototype.
12422         (cgraph_speculative_call_info): Declare.
12423         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
12424         to change; update call of ipa_find_references.
12425         * ipa-ref.c (ipa_record_reference): Fix return value; clear
12426         lto_stmt_uid and speculative flags.
12427         (ipa_dump_references): Dump speculation.
12428         (ipa_clone_references): Clone speculative flag.
12429         (ipa_clone_referring): Likewise.
12430         (ipa_clone_ref): New function.
12431         (ipa_find_reference): Look into lto_stmt_uids
12432         (ipa_clear_stmts_in_references): Do not clear speculative calls.
12433         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
12434         (ipa_find_reference): Update declaration.
12435         (ipa_clone_ref): Declare.
12436         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
12437         stream speculative flag.
12438         (lto_output_ref): Stream statements uids and speculation.
12439         (input_ref): Likewise.
12440         (input_edge): Stream speuclation.
12441         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
12442         (cgraph_set_call_stmt_including_clones): Handle speculation.
12443         * ipa-inline.c (heap_edge_removal_hook): New function.
12444         (inline_small_functions): Register it.
12445         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
12446         also initialize refs.
12447         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
12448         edge to change.
12449         (try_make_edge_direct_simple_call): Likewise.
12450         (try_make_edge_direct_simple_call): Likewise.
12451         (update_indirect_edges_after_inlining): Likewise.
12452         (remove_described_reference): Look proper lto_stmt_uid.
12453         (propagate_controlled_uses): Likewise.
12454         (propagate_controlled_uses): Liekwise.
12455         * tree-inline.c (copy_bb): Copy speculative edges.
12456         (redirect_all_calls): New function.
12457         (copy_cfg_body): Do redirection after loop info is updated.
12458         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
12460 2013-08-09  Jan Hubicka  <jh@suse.cz>
12462         * lto-streamer-out.c (output_function): Renumber PHIs.
12463         * lto-streamer-in.c (input_function): Likewise.
12465 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
12467         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
12468         (get_lane_unsigned): Likewise.
12469         (dup_lane_scalar): Likewise.
12470         (get_lane): enable for VALL.
12471         * config/aarch64/aarch64-simd.md
12472         (aarch64_dup_lane_scalar<mode>): Remove.
12473         (aarch64_get_lane_signed<mode>): Likewise.
12474         (aarch64_get_lane_unsigned<mode>): Likewise.
12475         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
12476         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
12477         (aarch64_get_lane<mode>): Enable for all vector modes.
12478         (aarch64_get_lanedi): Remove misleading constraints.
12479         * config/aarch64/arm_neon.h
12480         (__aarch64_vget_lane_any): Define.
12481         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
12482         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
12483         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
12484         * config/aarch64/iterators.md (VDQQH): New.
12485         (VDQQHS): Likewise.
12486         (vwcore): Likewise.
12488 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
12490         * configure.ac: Add GAS check for LEON instructions on SPARC.
12491         * configure: Regenerate.
12492         * config.in: Likewise.
12493         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
12494         sparc*-*-* block.
12495         * config/sparc/sparc.opt (LEON, LEON3): New masks.
12496         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
12497         for LEON or LEON3.
12498         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
12499         (AS_LEON_FLAG): New macro.
12500         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
12501         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
12502         Deal with LEON and LEON3 for the memory model.
12503         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
12504         (atomic_compare_and_swap<mode>_1): Likewise.
12505         (*atomic_compare_and_swap<mode>_1): Likewise.
12507 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
12509         * config/arm/neon.md (vcond): Fix floating-point vector
12510         comparisons against 0.
12512 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
12514         * lra-constraints.c (emit_spill_move): Remove assert.
12515         (process_alt_operands): Add more debugging
12516         output.  Increase reject for spilling into memory.  Decrease
12517         reject for reloading scratch.
12518         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
12520 2013-08-08  Steve Ellcey  <sellcey@mips.com>
12522         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
12523         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
12524         micromips incompatible.  Add nan2008.
12525         (MULTILIB_DIRNAMES): Add nan2008.
12526         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
12527         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
12528         and micromips incompatible.  Add nan2008.
12529         (MULTILIB_DIRNAMES): Add nan2008.
12530         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
12532 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
12534         PR rtl-optimization/58079
12535         * combine.c (combine_simplify_rtx): Avoid using SUBST if
12536         simplify_comparison has widened a comparison with an integer.
12538 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12540         * config/arm/neon.md (movmisalign<mode>): Disable when we
12541         don't allow unaligned accesses.
12542         (*movmisalign<mode>_neon_store): Likewise.
12543         (*movmisalign<mode>_neon_load): Likewise.
12544         (*movmisalign<mode>_neon_store): Likewise.
12545         (*movmisalign<mode>_neon_load): Likewise.
12547 2013-08-08  Jan Hubicka  <jh@suse.cz>
12549         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
12550         (make_pass_rebuild_cgraph_edges): Also clear references.
12551         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
12552         * ipa-inline-transform.c (inline_transform): Remove all references
12553         after inlining.
12554         * cgraphunit.c (expand_function): Remove all references after
12555         expansion.
12556         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
12557         (ipa_find_reference): Rewrite to iterator.
12558         (remove_stmt_references): Likewise.
12559         (ipa_clear_stmts_in_references): New function.
12560         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
12561         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
12562         clear references.
12563         * ipa-split.c (split_function): Remove references in split function.
12565 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
12567         PR target/57431
12568         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
12569         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
12571 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
12573         PR target/56979
12574         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
12575         suggested mode for the assignment isn't compatible with the
12576         registers required.
12578 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
12580         PR target/58065
12581         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
12583 2013-08-07  Xinliang David Li  <davidxl@google.com>
12585         * config/i386/i386.opt: New option -mtune-ctrl=.
12586         * config/i386/x86-tune.def: New file.
12587         * config/i386/i386.h: include x86-tune.def.
12588         * config/i386/i386.c (ix86_option_override_internal):
12589         Parsing -mtune-ctrl= option and set tune features.
12591 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
12593         PR other/12081
12594         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
12595         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
12596         to gen_2arg_fn_t.
12598 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
12600         * rtl.h (update_alignments): Declare.
12601         * final.c (grow_label_align): New function extracted from...
12602         (shorten_branches): ...here.  Call it.
12603         (update_alignments): New function.
12604         * reorg.c (sibling_labels): New variable.
12605         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
12606         the new label along with it onto the sibling_labels vector.
12607         (fill_simple_delay_slots): Adjust call to get_label_before.
12608         (fill_slots_from_thread): Likewise.
12609         (relax_delay_slots): Likewise.
12610         (make_return_insns): Likewise.
12611         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
12613 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
12615         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
12616         document its semantics.
12617         (diagnostic_report_diagnostic): Adjust accordingly.
12619 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
12621         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
12622         (sparc_option_override): ...and port to new C++ pass API.
12623         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
12625 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
12627         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
12629 2013-08-06  Caroline Tice  <cmtice@google.com>
12631         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
12632         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
12633         * tree-pass.h: Add pass_vtable_verify.
12634         * varasm.c (assemble_variable): Add code to properly set the comdat
12635         section and name for the .vtable_map_vars section.
12636         (assemble_vtyv_preinit_initializer): New function.
12637         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
12638         LINK_ONCE flag.
12639         * output.h: Add function decl for assemble_vtv_preinit_initializer.
12640         * vtable-verify.c: New file.
12641         * vtable-verify.h: New file.
12642         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
12643         initialiation levels.
12644         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
12645         * passes.def: Insert pass_vtable_verify.
12646         * aclocal.m4: Reorder includes.
12647         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
12648         -fvtv-counts options.
12649         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
12650         as appropriate, if -fvtable-verify=... is used.
12651         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
12652         -fvtable-verify=... is used.
12653         * Makefile.in (OBJS):  Add vtable-verify.o to list.
12654         (vtable-verify.o): Add new build rule.
12655         (GTFILES): Add vtable-verify.c to list.
12656         * common.opt (fvtable-verify=): New flag.
12657         (vtv_priority): Values for fvtable-verify= flag.
12658         (fvtv-counts): New flag.
12659         (fvtv-debug): New flag.
12660         * tree.h (save_vtable_map_decl): New extern function decl.
12662 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
12664         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
12665         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
12666         (pass_data_rl78_devirt): ...new pass_data instance and...
12667         (make_pass_rl78_devirt): ...new function.
12668         (rl78_asm_file_start): Port pass registration to new C++ API.
12670 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
12672         * coretypes.h (rtl_opt_pass): Add.
12673         (gcc::context): Add.
12674         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
12675         (epiphany_init): Port to new C++ pass API.
12676         (epiphany_optimize_mode_switching): Likewise.
12677         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
12678         (pass_manager::get_pass_mode_switching): New.
12679         (pass_manager::get_pass_peephole2): New.
12680         * mode-switching.c (pass_mode_switching): Add clone method.
12681         * recog.c (pass_peephole2): Add clone method.
12682         (pass_split_all_insns): Add clone method.
12684 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
12686         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
12687         (mips_option_override): ...here, porting to new C++ API for passes.
12689 2013-08-06  Jan Hubicka  <jh@suse.cz>
12691         * cgraph.c (cgraph_get_body): New function based on lto.c
12692         implementation.
12693         * cgraph.h (cgraph_get_body): Declare.
12694         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
12695         LTO paths.
12696         * cgraphunit.c (expand_function): Get body prior expanding.
12697         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
12698         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
12699         really need.
12700         * passes.c (do_per_function_toporder): Get body.
12701         * tree-inline.c (expand_call_inline): Get body prior inlining it.
12702         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
12704 2013-08-06  Martin Jambor  <mjambor@suse.cz>
12706         PR fortran/57987
12707         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
12708         is not re-finalized.  Rename second parameter to no_collect.
12710 2013-08-06  Martin Jambor  <mjambor@suse.cz>
12712         PR middle-end/58041
12713         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
12714         MEM_REF has proper alignment information.
12716 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
12718         PR other/12081
12719         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
12720         class insn_gen_fn.
12721         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
12722         rtx (*) (rtx, ...) with insn_gen_fn.
12723         * genoutput.c (output_insn_data): Cast gen_? function pointers to
12724         insn_gen_fn::stored_funcptr.  Add initializer braces.
12726 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
12728         Rewrite how instances of passes are cloned to remove assumptions
12729         about their sizes (thus allowing pass subclasses to have
12730         additional data fields, albeit non-GC-managed ones at this point).
12732         * passes.c (make_pass_instance): Now that passes have clone
12733         methods, rewrite this function to eliminate XNEW and memcpy
12734         calls that used hardcoded sizes.  Since this function no longer
12735         creates pass instances, rename it to...
12736         (add_pass_instance): ...this.  Document the old way that passes were
12737         numbered and flagged, and rework this function to continue using it.
12738         (next_pass_1): Add an initial_pass argument for use by
12739         add_pass_instance.
12740         (position_pass): When adding multiple instances of a pass, use
12741         the pass's clone method, rather than relying on the XNEW/memcpy
12742         within the former make_pass_instance (now add_pass_instance).
12743         (pass_manager::pass_manager): When invoking next_pass_1, also supply
12744         the initial instance of the current pass within the pass manager.
12746 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
12748         This is the automated part of the conversion of passes from C
12749         structs to C++ classes.
12751         Patch autogenerated by refactor_passes.py from
12752         https://github.com/davidmalcolm/gcc-refactoring-scripts
12753         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
12755         * asan.c (pass_asan): Convert from a global struct to a subclass of
12756         gimple_opt_pass along with...
12757         (pass_data_asan): ...new pass_data instance and...
12758         (make_pass_asan): ...new function.
12759         (pass_asan_O0): Convert from a global struct to a subclass of
12760         gimple_opt_pass along with...
12761         (pass_data_asan_O0): ...new pass_data instance and...
12762         (make_pass_asan_O0): ...new function.
12763         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
12764         subclass of rtl_opt_pass along with...
12765         (pass_data_inc_dec): ...new pass_data instance and...
12766         (make_pass_inc_dec): ...new function.
12767         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
12768         a subclass of rtl_opt_pass along with...
12769         (pass_data_reorder_blocks): ...new pass_data instance and...
12770         (make_pass_reorder_blocks): ...new function.
12771         (pass_duplicate_computed_gotos): Convert from a global struct to a
12772         subclass of rtl_opt_pass along with...
12773         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
12774         (make_pass_duplicate_computed_gotos): ...new function.
12775         (pass_partition_blocks): Convert from a global struct to a subclass of
12776         rtl_opt_pass along with...
12777         (pass_data_partition_blocks): ...new pass_data instance and...
12778         (make_pass_partition_blocks): ...new function.
12779         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
12780         struct to a subclass of rtl_opt_pass along with...
12781         (pass_data_branch_target_load_optimize1): ...new pass_data instance
12782         and...
12783         (make_pass_branch_target_load_optimize1): ...new function.
12784         (pass_branch_target_load_optimize2): Convert from a global struct to a
12785         subclass of rtl_opt_pass along with...
12786         (pass_data_branch_target_load_optimize2): ...new pass_data instance
12787         and...
12788         (make_pass_branch_target_load_optimize2): ...new function.
12789         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
12790         of rtl_opt_pass along with...
12791         (pass_data_jump): ...new pass_data instance and...
12792         (make_pass_jump): ...new function.
12793         (pass_jump2): Convert from a global struct to a subclass of
12794         rtl_opt_pass along with...
12795         (pass_data_jump2): ...new pass_data instance and...
12796         (make_pass_jump2): ...new function.
12797         * cfgexpand.c (pass_expand): Convert from a global struct to a
12798         subclass of rtl_opt_pass along with...
12799         (pass_data_expand): ...new pass_data instance and...
12800         (make_pass_expand): ...new function.
12801         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
12802         of rtl_opt_pass along with...
12803         (pass_data_free_cfg): ...new pass_data instance and...
12804         (make_pass_free_cfg): ...new function.
12805         (pass_into_cfg_layout_mode): Convert from a global struct to a
12806         subclass of rtl_opt_pass along with...
12807         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
12808         (make_pass_into_cfg_layout_mode): ...new function.
12809         (pass_outof_cfg_layout_mode): Convert from a global struct to a
12810         subclass of rtl_opt_pass along with...
12811         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
12812         (make_pass_outof_cfg_layout_mode): ...new function.
12813         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
12814         struct to a subclass of gimple_opt_pass along with...
12815         (pass_data_build_cgraph_edges): ...new pass_data instance and...
12816         (make_pass_build_cgraph_edges): ...new function.
12817         (pass_rebuild_cgraph_edges): Convert from a global struct to a
12818         subclass of gimple_opt_pass along with...
12819         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
12820         (make_pass_rebuild_cgraph_edges): ...new function.
12821         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
12822         subclass of gimple_opt_pass along with...
12823         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
12824         and...
12825         (make_pass_remove_cgraph_callee_edges): ...new function.
12826         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
12827         struct to a subclass of rtl_opt_pass along with...
12828         (pass_data_stack_adjustments): ...new pass_data instance and...
12829         (make_pass_stack_adjustments): ...new function.
12830         * combine.c (pass_combine): Convert from a global struct to a subclass
12831         of rtl_opt_pass along with...
12832         (pass_data_combine): ...new pass_data instance and...
12833         (make_pass_combine): ...new function.
12834         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
12835         global struct to a subclass of rtl_opt_pass along with...
12836         (pass_data_compare_elim_after_reload): ...new pass_data instance
12837         and...
12838         (make_pass_compare_elim_after_reload): ...new function.
12839         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
12840         of rtl_opt_pass along with...
12841         (pass_data_rtl_cprop): ...new pass_data instance and...
12842         (make_pass_rtl_cprop): ...new function.
12843         * cse.c (pass_cse): Convert from a global struct to a subclass of
12844         rtl_opt_pass along with...
12845         (pass_data_cse): ...new pass_data instance and...
12846         (make_pass_cse): ...new function.
12847         (pass_cse2): Convert from a global struct to a subclass of
12848         rtl_opt_pass along with...
12849         (pass_data_cse2): ...new pass_data instance and...
12850         (make_pass_cse2): ...new function.
12851         (pass_cse_after_global_opts): Convert from a global struct to a
12852         subclass of rtl_opt_pass along with...
12853         (pass_data_cse_after_global_opts): ...new pass_data instance and...
12854         (make_pass_cse_after_global_opts): ...new function.
12855         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
12856         of rtl_opt_pass along with...
12857         (pass_data_ud_rtl_dce): ...new pass_data instance and...
12858         (make_pass_ud_rtl_dce): ...new function.
12859         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
12860         rtl_opt_pass along with...
12861         (pass_data_fast_rtl_dce): ...new pass_data instance and...
12862         (make_pass_fast_rtl_dce): ...new function.
12863         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
12864         a subclass of rtl_opt_pass along with...
12865         (pass_data_df_initialize_opt): ...new pass_data instance and...
12866         (make_pass_df_initialize_opt): ...new function.
12867         (pass_df_initialize_no_opt): Convert from a global struct to a
12868         subclass of rtl_opt_pass along with...
12869         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
12870         (make_pass_df_initialize_no_opt): ...new function.
12871         (pass_df_finish): Convert from a global struct to a subclass of
12872         rtl_opt_pass along with...
12873         (pass_data_df_finish): ...new pass_data instance and...
12874         (make_pass_df_finish): ...new function.
12875         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
12876         rtl_opt_pass along with...
12877         (pass_data_rtl_dse1): ...new pass_data instance and...
12878         (make_pass_rtl_dse1): ...new function.
12879         (pass_rtl_dse2): Convert from a global struct to a subclass of
12880         rtl_opt_pass along with...
12881         (pass_data_rtl_dse2): ...new pass_data instance and...
12882         (make_pass_rtl_dse2): ...new function.
12883         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
12884         subclass of rtl_opt_pass along with...
12885         (pass_data_dwarf2_frame): ...new pass_data instance and...
12886         (make_pass_dwarf2_frame): ...new function.
12887         * except.c (pass_set_nothrow_function_flags): Convert from a global
12888         struct to a subclass of rtl_opt_pass along with...
12889         (pass_data_set_nothrow_function_flags): ...new pass_data instance
12890         and...
12891         (make_pass_set_nothrow_function_flags): ...new function.
12892         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
12893         subclass of rtl_opt_pass along with...
12894         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
12895         and...
12896         (make_pass_convert_to_eh_region_ranges): ...new function.
12897         * final.c (pass_compute_alignments): Convert from a global struct to a
12898         subclass of rtl_opt_pass along with...
12899         (pass_data_compute_alignments): ...new pass_data instance and...
12900         (make_pass_compute_alignments): ...new function.
12901         (pass_final): Convert from a global struct to a subclass of
12902         rtl_opt_pass along with...
12903         (pass_data_final): ...new pass_data instance and...
12904         (make_pass_final): ...new function.
12905         (pass_shorten_branches): Convert from a global struct to a subclass of
12906         rtl_opt_pass along with...
12907         (pass_data_shorten_branches): ...new pass_data instance and...
12908         (make_pass_shorten_branches): ...new function.
12909         (pass_clean_state): Convert from a global struct to a subclass of
12910         rtl_opt_pass along with...
12911         (pass_data_clean_state): ...new pass_data instance and...
12912         (make_pass_clean_state): ...new function.
12913         * function.c (pass_instantiate_virtual_regs): Convert from a global
12914         struct to a subclass of rtl_opt_pass along with...
12915         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
12916         (make_pass_instantiate_virtual_regs): ...new function.
12917         (pass_leaf_regs): Convert from a global struct to a subclass of
12918         rtl_opt_pass along with...
12919         (pass_data_leaf_regs): ...new pass_data instance and...
12920         (make_pass_leaf_regs): ...new function.
12921         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
12922         subclass of rtl_opt_pass along with...
12923         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
12924         and...
12925         (make_pass_thread_prologue_and_epilogue): ...new function.
12926         (pass_match_asm_constraints): Convert from a global struct to a
12927         subclass of rtl_opt_pass along with...
12928         (pass_data_match_asm_constraints): ...new pass_data instance and...
12929         (make_pass_match_asm_constraints): ...new function.
12930         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
12931         subclass of rtl_opt_pass along with...
12932         (pass_data_rtl_fwprop): ...new pass_data instance and...
12933         (make_pass_rtl_fwprop): ...new function.
12934         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
12935         rtl_opt_pass along with...
12936         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
12937         (make_pass_rtl_fwprop_addr): ...new function.
12938         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
12939         rtl_opt_pass along with...
12940         (pass_data_rtl_pre): ...new pass_data instance and...
12941         (make_pass_rtl_pre): ...new function.
12942         (pass_rtl_hoist): Convert from a global struct to a subclass of
12943         rtl_opt_pass along with...
12944         (pass_data_rtl_hoist): ...new pass_data instance and...
12945         (make_pass_rtl_hoist): ...new function.
12946         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
12947         subclass of gimple_opt_pass along with...
12948         (pass_data_lower_cf): ...new pass_data instance and...
12949         (make_pass_lower_cf): ...new function.
12950         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
12951         from a global struct to a subclass of gimple_opt_pass along with...
12952         (pass_data_strength_reduction): ...new pass_data instance and...
12953         (make_pass_strength_reduction): ...new function.
12954         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
12955         of rtl_opt_pass along with...
12956         (pass_data_rtl_ifcvt): ...new pass_data instance and...
12957         (make_pass_rtl_ifcvt): ...new function.
12958         (pass_if_after_combine): Convert from a global struct to a subclass of
12959         rtl_opt_pass along with...
12960         (pass_data_if_after_combine): ...new pass_data instance and...
12961         (make_pass_if_after_combine): ...new function.
12962         (pass_if_after_reload): Convert from a global struct to a subclass of
12963         rtl_opt_pass along with...
12964         (pass_data_if_after_reload): ...new pass_data instance and...
12965         (make_pass_if_after_reload): ...new function.
12966         * init-regs.c (pass_initialize_regs): Convert from a global struct to
12967         a subclass of rtl_opt_pass along with...
12968         (pass_data_initialize_regs): ...new pass_data instance and...
12969         (make_pass_initialize_regs): ...new function.
12970         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
12971         of ipa_opt_pass_d along with...
12972         (pass_data_ipa_cp): ...new pass_data instance and...
12973         (make_pass_ipa_cp): ...new function.
12974         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
12975         global struct to a subclass of gimple_opt_pass along with...
12976         (pass_data_inline_parameters): ...new pass_data instance and...
12977         (make_pass_inline_parameters): ...new function.
12978         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
12979         subclass of gimple_opt_pass along with...
12980         (pass_data_early_inline): ...new pass_data instance and...
12981         (make_pass_early_inline): ...new function.
12982         (pass_ipa_inline): Convert from a global struct to a subclass of
12983         ipa_opt_pass_d along with...
12984         (pass_data_ipa_inline): ...new pass_data instance and...
12985         (make_pass_ipa_inline): ...new function.
12986         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
12987         struct to a subclass of gimple_opt_pass along with...
12988         (pass_data_local_pure_const): ...new pass_data instance and...
12989         (make_pass_local_pure_const): ...new function.
12990         (pass_ipa_pure_const): Convert from a global struct to a subclass of
12991         ipa_opt_pass_d along with...
12992         (pass_data_ipa_pure_const): ...new pass_data instance and...
12993         (make_pass_ipa_pure_const): ...new function.
12994         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
12995         to a subclass of ipa_opt_pass_d along with...
12996         (pass_data_ipa_reference): ...new pass_data instance and...
12997         (make_pass_ipa_reference): ...new function.
12998         * ipa-split.c (pass_split_functions): Convert from a global struct to
12999         a subclass of gimple_opt_pass along with...
13000         (pass_data_split_functions): ...new pass_data instance and...
13001         (make_pass_split_functions): ...new function.
13002         (pass_feedback_split_functions): Convert from a global struct to a
13003         subclass of gimple_opt_pass along with...
13004         (pass_data_feedback_split_functions): ...new pass_data instance and...
13005         (make_pass_feedback_split_functions): ...new function.
13006         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
13007         global struct to a subclass of simple_ipa_opt_pass along with...
13008         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
13009         instance and...
13010         (make_pass_ipa_function_and_variable_visibility): ...new function.
13011         (pass_ipa_free_inline_summary): Convert from a global struct to a
13012         subclass of simple_ipa_opt_pass along with...
13013         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
13014         (make_pass_ipa_free_inline_summary): ...new function.
13015         (pass_ipa_whole_program_visibility): Convert from a global struct to a
13016         subclass of ipa_opt_pass_d along with...
13017         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
13018         and...
13019         (make_pass_ipa_whole_program_visibility): ...new function.
13020         (pass_ipa_profile): Convert from a global struct to a subclass of
13021         ipa_opt_pass_d along with...
13022         (pass_data_ipa_profile): ...new pass_data instance and...
13023         (make_pass_ipa_profile): ...new function.
13024         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
13025         ipa_opt_pass_d along with...
13026         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
13027         (make_pass_ipa_cdtor_merge): ...new function.
13028         * ira.c (pass_ira): Convert from a global struct to a subclass of
13029         rtl_opt_pass along with...
13030         (pass_data_ira): ...new pass_data instance and...
13031         (make_pass_ira): ...new function.
13032         (pass_reload): Convert from a global struct to a subclass of
13033         rtl_opt_pass along with...
13034         (pass_data_reload): ...new pass_data instance and...
13035         (make_pass_reload): ...new function.
13036         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
13037         subclass of rtl_opt_pass along with...
13038         (pass_data_cleanup_barriers): ...new pass_data instance and...
13039         (make_pass_cleanup_barriers): ...new function.
13040         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
13041         of rtl_opt_pass along with...
13042         (pass_data_loop2): ...new pass_data instance and...
13043         (make_pass_loop2): ...new function.
13044         (pass_rtl_loop_init): Convert from a global struct to a subclass of
13045         rtl_opt_pass along with...
13046         (pass_data_rtl_loop_init): ...new pass_data instance and...
13047         (make_pass_rtl_loop_init): ...new function.
13048         (pass_rtl_loop_done): Convert from a global struct to a subclass of
13049         rtl_opt_pass along with...
13050         (pass_data_rtl_loop_done): ...new pass_data instance and...
13051         (make_pass_rtl_loop_done): ...new function.
13052         (pass_rtl_move_loop_invariants): Convert from a global struct to a
13053         subclass of rtl_opt_pass along with...
13054         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
13055         (make_pass_rtl_move_loop_invariants): ...new function.
13056         (pass_rtl_unswitch): Convert from a global struct to a subclass of
13057         rtl_opt_pass along with...
13058         (pass_data_rtl_unswitch): ...new pass_data instance and...
13059         (make_pass_rtl_unswitch): ...new function.
13060         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
13061         subclass of rtl_opt_pass along with...
13062         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
13063         and...
13064         (make_pass_rtl_unroll_and_peel_loops): ...new function.
13065         (pass_rtl_doloop): Convert from a global struct to a subclass of
13066         rtl_opt_pass along with...
13067         (pass_data_rtl_doloop): ...new pass_data instance and...
13068         (make_pass_rtl_doloop): ...new function.
13069         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
13070         a subclass of rtl_opt_pass along with...
13071         (pass_data_lower_subreg): ...new pass_data instance and...
13072         (make_pass_lower_subreg): ...new function.
13073         (pass_lower_subreg2): Convert from a global struct to a subclass of
13074         rtl_opt_pass along with...
13075         (pass_data_lower_subreg2): ...new pass_data instance and...
13076         (make_pass_lower_subreg2): ...new function.
13077         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
13078         struct to a subclass of ipa_opt_pass_d along with...
13079         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
13080         (make_pass_ipa_lto_gimple_out): ...new function.
13081         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
13082         of ipa_opt_pass_d along with...
13083         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
13084         (make_pass_ipa_lto_finish_out): ...new function.
13085         * mode-switching.c (pass_mode_switching): Convert from a global struct
13086         to a subclass of rtl_opt_pass along with...
13087         (pass_data_mode_switching): ...new pass_data instance and...
13088         (make_pass_mode_switching): ...new function.
13089         * modulo-sched.c (pass_sms): Convert from a global struct to a
13090         subclass of rtl_opt_pass along with...
13091         (pass_data_sms): ...new pass_data instance and...
13092         (make_pass_sms): ...new function.
13093         * omp-low.c (pass_expand_omp): Convert from a global struct to a
13094         subclass of gimple_opt_pass along with...
13095         (pass_data_expand_omp): ...new pass_data instance and...
13096         (make_pass_expand_omp): ...new function.
13097         (pass_lower_omp): Convert from a global struct to a subclass of
13098         gimple_opt_pass along with...
13099         (pass_data_lower_omp): ...new pass_data instance and...
13100         (make_pass_lower_omp): ...new function.
13101         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
13102         of gimple_opt_pass along with...
13103         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
13104         (make_pass_diagnose_omp_blocks): ...new function.
13105         * passes.c (pass_early_local_passes): Convert from a global struct to
13106         a subclass of simple_ipa_opt_pass along with...
13107         (pass_data_early_local_passes): ...new pass_data instance and...
13108         (make_pass_early_local_passes): ...new function.
13109         (pass_all_early_optimizations): Convert from a global struct to a
13110         subclass of gimple_opt_pass along with...
13111         (pass_data_all_early_optimizations): ...new pass_data instance and...
13112         (make_pass_all_early_optimizations): ...new function.
13113         (pass_all_optimizations): Convert from a global struct to a subclass
13114         of gimple_opt_pass along with...
13115         (pass_data_all_optimizations): ...new pass_data instance and...
13116         (make_pass_all_optimizations): ...new function.
13117         (pass_all_optimizations_g): Convert from a global struct to a subclass
13118         of gimple_opt_pass along with...
13119         (pass_data_all_optimizations_g): ...new pass_data instance and...
13120         (make_pass_all_optimizations_g): ...new function.
13121         (pass_rest_of_compilation): Convert from a global struct to a subclass
13122         of rtl_opt_pass along with...
13123         (pass_data_rest_of_compilation): ...new pass_data instance and...
13124         (make_pass_rest_of_compilation): ...new function.
13125         (pass_postreload): Convert from a global struct to a subclass of
13126         rtl_opt_pass along with...
13127         (pass_data_postreload): ...new pass_data instance and...
13128         (make_pass_postreload): ...new function.
13129         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
13130         subclass of rtl_opt_pass along with...
13131         (pass_data_gcse2): ...new pass_data instance and...
13132         (make_pass_gcse2): ...new function.
13133         * postreload.c (pass_postreload_cse): Convert from a global struct to
13134         a subclass of rtl_opt_pass along with...
13135         (pass_data_postreload_cse): ...new pass_data instance and...
13136         (make_pass_postreload_cse): ...new function.
13137         * predict.c (pass_profile): Convert from a global struct to a subclass
13138         of gimple_opt_pass along with...
13139         (pass_data_profile): ...new pass_data instance and...
13140         (make_pass_profile): ...new function.
13141         (pass_strip_predict_hints): Convert from a global struct to a subclass
13142         of gimple_opt_pass along with...
13143         (pass_data_strip_predict_hints): ...new pass_data instance and...
13144         (make_pass_strip_predict_hints): ...new function.
13145         * recog.c (pass_peephole2): Convert from a global struct to a subclass
13146         of rtl_opt_pass along with...
13147         (pass_data_peephole2): ...new pass_data instance and...
13148         (make_pass_peephole2): ...new function.
13149         (pass_split_all_insns): Convert from a global struct to a subclass of
13150         rtl_opt_pass along with...
13151         (pass_data_split_all_insns): ...new pass_data instance and...
13152         (make_pass_split_all_insns): ...new function.
13153         (pass_split_after_reload): Convert from a global struct to a subclass
13154         of rtl_opt_pass along with...
13155         (pass_data_split_after_reload): ...new pass_data instance and...
13156         (make_pass_split_after_reload): ...new function.
13157         (pass_split_before_regstack): Convert from a global struct to a
13158         subclass of rtl_opt_pass along with...
13159         (pass_data_split_before_regstack): ...new pass_data instance and...
13160         (make_pass_split_before_regstack): ...new function.
13161         (pass_split_before_sched2): Convert from a global struct to a subclass
13162         of rtl_opt_pass along with...
13163         (pass_data_split_before_sched2): ...new pass_data instance and...
13164         (make_pass_split_before_sched2): ...new function.
13165         (pass_split_for_shorten_branches): Convert from a global struct to a
13166         subclass of rtl_opt_pass along with...
13167         (pass_data_split_for_shorten_branches): ...new pass_data instance
13168         and...
13169         (make_pass_split_for_shorten_branches): ...new function.
13170         * ree.c (pass_ree): Convert from a global struct to a subclass of
13171         rtl_opt_pass along with...
13172         (pass_data_ree): ...new pass_data instance and...
13173         (make_pass_ree): ...new function.
13174         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
13175         subclass of rtl_opt_pass along with...
13176         (pass_data_stack_regs): ...new pass_data instance and...
13177         (make_pass_stack_regs): ...new function.
13178         (pass_stack_regs_run): Convert from a global struct to a subclass of
13179         rtl_opt_pass along with...
13180         (pass_data_stack_regs_run): ...new pass_data instance and...
13181         (make_pass_stack_regs_run): ...new function.
13182         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
13183         subclass of rtl_opt_pass along with...
13184         (pass_data_cprop_hardreg): ...new pass_data instance and...
13185         (make_pass_cprop_hardreg): ...new function.
13186         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
13187         subclass of rtl_opt_pass along with...
13188         (pass_data_reginfo_init): ...new pass_data instance and...
13189         (make_pass_reginfo_init): ...new function.
13190         * regmove.c (pass_regmove): Convert from a global struct to a subclass
13191         of rtl_opt_pass along with...
13192         (pass_data_regmove): ...new pass_data instance and...
13193         (make_pass_regmove): ...new function.
13194         * regrename.c (pass_regrename): Convert from a global struct to a
13195         subclass of rtl_opt_pass along with...
13196         (pass_data_regrename): ...new pass_data instance and...
13197         (make_pass_regrename): ...new function.
13198         * reorg.c (pass_delay_slots): Convert from a global struct to a
13199         subclass of rtl_opt_pass along with...
13200         (pass_data_delay_slots): ...new pass_data instance and...
13201         (make_pass_delay_slots): ...new function.
13202         (pass_machine_reorg): Convert from a global struct to a subclass of
13203         rtl_opt_pass along with...
13204         (pass_data_machine_reorg): ...new pass_data instance and...
13205         (make_pass_machine_reorg): ...new function.
13206         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
13207         of rtl_opt_pass along with...
13208         (pass_data_sched): ...new pass_data instance and...
13209         (make_pass_sched): ...new function.
13210         (pass_sched2): Convert from a global struct to a subclass of
13211         rtl_opt_pass along with...
13212         (pass_data_sched2): ...new pass_data instance and...
13213         (make_pass_sched2): ...new function.
13214         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
13215         to a subclass of rtl_opt_pass along with...
13216         (pass_data_stack_ptr_mod): ...new pass_data instance and...
13217         (make_pass_stack_ptr_mod): ...new function.
13218         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
13219         to a subclass of rtl_opt_pass along with...
13220         (pass_data_rtl_store_motion): ...new pass_data instance and...
13221         (make_pass_rtl_store_motion): ...new function.
13222         * tracer.c (pass_tracer): Convert from a global struct to a subclass
13223         of gimple_opt_pass along with...
13224         (pass_data_tracer): ...new pass_data instance and...
13225         (make_pass_tracer): ...new function.
13226         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
13227         to a subclass of gimple_opt_pass along with...
13228         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
13229         (make_pass_diagnose_tm_blocks): ...new function.
13230         (pass_lower_tm): Convert from a global struct to a subclass of
13231         gimple_opt_pass along with...
13232         (pass_data_lower_tm): ...new pass_data instance and...
13233         (make_pass_lower_tm): ...new function.
13234         (pass_tm_init): Convert from a global struct to a subclass of
13235         gimple_opt_pass along with...
13236         (pass_data_tm_init): ...new pass_data instance and...
13237         (make_pass_tm_init): ...new function.
13238         (pass_tm_mark): Convert from a global struct to a subclass of
13239         gimple_opt_pass along with...
13240         (pass_data_tm_mark): ...new pass_data instance and...
13241         (make_pass_tm_mark): ...new function.
13242         (pass_tm_edges): Convert from a global struct to a subclass of
13243         gimple_opt_pass along with...
13244         (pass_data_tm_edges): ...new pass_data instance and...
13245         (make_pass_tm_edges): ...new function.
13246         (pass_tm_memopt): Convert from a global struct to a subclass of
13247         gimple_opt_pass along with...
13248         (pass_data_tm_memopt): ...new pass_data instance and...
13249         (make_pass_tm_memopt): ...new function.
13250         (pass_ipa_tm): Convert from a global struct to a subclass of
13251         simple_ipa_opt_pass along with...
13252         (pass_data_ipa_tm): ...new pass_data instance and...
13253         (make_pass_ipa_tm): ...new function.
13254         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
13255         subclass of gimple_opt_pass along with...
13256         (pass_data_call_cdce): ...new pass_data instance and...
13257         (make_pass_call_cdce): ...new function.
13258         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
13259         subclass of gimple_opt_pass along with...
13260         (pass_data_build_cfg): ...new pass_data instance and...
13261         (make_pass_build_cfg): ...new function.
13262         (pass_split_crit_edges): Convert from a global struct to a subclass of
13263         gimple_opt_pass along with...
13264         (pass_data_split_crit_edges): ...new pass_data instance and...
13265         (make_pass_split_crit_edges): ...new function.
13266         (pass_warn_function_return): Convert from a global struct to a
13267         subclass of gimple_opt_pass along with...
13268         (pass_data_warn_function_return): ...new pass_data instance and...
13269         (make_pass_warn_function_return): ...new function.
13270         (pass_warn_function_noreturn): Convert from a global struct to a
13271         subclass of gimple_opt_pass along with...
13272         (pass_data_warn_function_noreturn): ...new pass_data instance and...
13273         (make_pass_warn_function_noreturn): ...new function.
13274         (pass_warn_unused_result): Convert from a global struct to a subclass
13275         of gimple_opt_pass along with...
13276         (pass_data_warn_unused_result): ...new pass_data instance and...
13277         (make_pass_warn_unused_result): ...new function.
13278         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
13279         a subclass of gimple_opt_pass along with...
13280         (pass_data_merge_phi): ...new pass_data instance and...
13281         (make_pass_merge_phi): ...new function.
13282         * tree-complex.c (pass_lower_complex): Convert from a global struct to
13283         a subclass of gimple_opt_pass along with...
13284         (pass_data_lower_complex): ...new pass_data instance and...
13285         (make_pass_lower_complex): ...new function.
13286         (pass_lower_complex_O0): Convert from a global struct to a subclass of
13287         gimple_opt_pass along with...
13288         (pass_data_lower_complex_O0): ...new pass_data instance and...
13289         (make_pass_lower_complex_O0): ...new function.
13290         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
13291         subclass of gimple_opt_pass along with...
13292         (pass_data_lower_eh): ...new pass_data instance and...
13293         (make_pass_lower_eh): ...new function.
13294         (pass_refactor_eh): Convert from a global struct to a subclass of
13295         gimple_opt_pass along with...
13296         (pass_data_refactor_eh): ...new pass_data instance and...
13297         (make_pass_refactor_eh): ...new function.
13298         (pass_lower_resx): Convert from a global struct to a subclass of
13299         gimple_opt_pass along with...
13300         (pass_data_lower_resx): ...new pass_data instance and...
13301         (make_pass_lower_resx): ...new function.
13302         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
13303         of gimple_opt_pass along with...
13304         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
13305         (make_pass_lower_eh_dispatch): ...new function.
13306         (pass_cleanup_eh): Convert from a global struct to a subclass of
13307         gimple_opt_pass along with...
13308         (pass_data_cleanup_eh): ...new pass_data instance and...
13309         (make_pass_cleanup_eh): ...new function.
13310         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
13311         to a subclass of simple_ipa_opt_pass along with...
13312         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
13313         (make_pass_ipa_lower_emutls): ...new function.
13314         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
13315         a subclass of gimple_opt_pass along with...
13316         (pass_data_if_conversion): ...new pass_data instance and...
13317         (make_pass_if_conversion): ...new function.
13318         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
13319         subclass of gimple_opt_pass along with...
13320         (pass_data_build_ssa): ...new pass_data instance and...
13321         (make_pass_build_ssa): ...new function.
13322         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
13323         global struct to a subclass of gimple_opt_pass along with...
13324         (pass_data_loop_distribution): ...new pass_data instance and...
13325         (make_pass_loop_distribution): ...new function.
13326         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
13327         subclass of gimple_opt_pass along with...
13328         (pass_data_mudflap_1): ...new pass_data instance and...
13329         (make_pass_mudflap_1): ...new function.
13330         (pass_mudflap_2): Convert from a global struct to a subclass of
13331         gimple_opt_pass along with...
13332         (pass_data_mudflap_2): ...new pass_data instance and...
13333         (make_pass_mudflap_2): ...new function.
13334         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
13335         subclass of gimple_opt_pass along with...
13336         (pass_data_mudflap_1): ...new pass_data instance and...
13337         (make_pass_mudflap_1): ...new function.
13338         (pass_mudflap_2): Convert from a global struct to a subclass of
13339         gimple_opt_pass along with...
13340         (pass_data_mudflap_2): ...new pass_data instance and...
13341         (make_pass_mudflap_2): ...new function.
13342         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
13343         gimple_opt_pass along with...
13344         (pass_data_nrv): ...new pass_data instance and...
13345         (make_pass_nrv): ...new function.
13346         (pass_return_slot): Convert from a global struct to a subclass of
13347         gimple_opt_pass along with...
13348         (pass_data_return_slot): ...new pass_data instance and...
13349         (make_pass_return_slot): ...new function.
13350         * tree-object-size.c (pass_object_sizes): Convert from a global struct
13351         to a subclass of gimple_opt_pass along with...
13352         (pass_data_object_sizes): ...new pass_data instance and...
13353         (make_pass_object_sizes): ...new function.
13354         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
13355         global struct to a subclass of gimple_opt_pass along with...
13356         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
13357         and...
13358         (make_pass_cleanup_cfg_post_optimizing): ...new function.
13359         (pass_fixup_cfg): Convert from a global struct to a subclass of
13360         gimple_opt_pass along with...
13361         (pass_data_fixup_cfg): ...new pass_data instance and...
13362         (make_pass_fixup_cfg): ...new function.
13363         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
13364         (make_pass_mudflap_1): ...new function.
13365         (pass_mudflap_2): Replace declaration with that of...
13366         (make_pass_mudflap_2): ...new function.
13367         (pass_asan): Replace declaration with that of...
13368         (make_pass_asan): ...new function.
13369         (pass_asan_O0): Replace declaration with that of...
13370         (make_pass_asan_O0): ...new function.
13371         (pass_tsan): Replace declaration with that of...
13372         (make_pass_tsan): ...new function.
13373         (pass_tsan_O0): Replace declaration with that of...
13374         (make_pass_tsan_O0): ...new function.
13375         (pass_lower_cf): Replace declaration with that of...
13376         (make_pass_lower_cf): ...new function.
13377         (pass_refactor_eh): Replace declaration with that of...
13378         (make_pass_refactor_eh): ...new function.
13379         (pass_lower_eh): Replace declaration with that of...
13380         (make_pass_lower_eh): ...new function.
13381         (pass_lower_eh_dispatch): Replace declaration with that of...
13382         (make_pass_lower_eh_dispatch): ...new function.
13383         (pass_lower_resx): Replace declaration with that of...
13384         (make_pass_lower_resx): ...new function.
13385         (pass_build_cfg): Replace declaration with that of...
13386         (make_pass_build_cfg): ...new function.
13387         (pass_early_tree_profile): Replace declaration with that of...
13388         (make_pass_early_tree_profile): ...new function.
13389         (pass_cleanup_eh): Replace declaration with that of...
13390         (make_pass_cleanup_eh): ...new function.
13391         (pass_sra): Replace declaration with that of...
13392         (make_pass_sra): ...new function.
13393         (pass_sra_early): Replace declaration with that of...
13394         (make_pass_sra_early): ...new function.
13395         (pass_early_ipa_sra): Replace declaration with that of...
13396         (make_pass_early_ipa_sra): ...new function.
13397         (pass_tail_recursion): Replace declaration with that of...
13398         (make_pass_tail_recursion): ...new function.
13399         (pass_tail_calls): Replace declaration with that of...
13400         (make_pass_tail_calls): ...new function.
13401         (pass_tree_loop): Replace declaration with that of...
13402         (make_pass_tree_loop): ...new function.
13403         (pass_tree_loop_init): Replace declaration with that of...
13404         (make_pass_tree_loop_init): ...new function.
13405         (pass_lim): Replace declaration with that of...
13406         (make_pass_lim): ...new function.
13407         (pass_tree_unswitch): Replace declaration with that of...
13408         (make_pass_tree_unswitch): ...new function.
13409         (pass_predcom): Replace declaration with that of...
13410         (make_pass_predcom): ...new function.
13411         (pass_iv_canon): Replace declaration with that of...
13412         (make_pass_iv_canon): ...new function.
13413         (pass_scev_cprop): Replace declaration with that of...
13414         (make_pass_scev_cprop): ...new function.
13415         (pass_empty_loop): Replace declaration with that of...
13416         (make_pass_empty_loop): ...new function.
13417         (pass_record_bounds): Replace declaration with that of...
13418         (make_pass_record_bounds): ...new function.
13419         (pass_graphite): Replace declaration with that of...
13420         (make_pass_graphite): ...new function.
13421         (pass_graphite_transforms): Replace declaration with that of...
13422         (make_pass_graphite_transforms): ...new function.
13423         (pass_if_conversion): Replace declaration with that of...
13424         (make_pass_if_conversion): ...new function.
13425         (pass_loop_distribution): Replace declaration with that of...
13426         (make_pass_loop_distribution): ...new function.
13427         (pass_vectorize): Replace declaration with that of...
13428         (make_pass_vectorize): ...new function.
13429         (pass_slp_vectorize): Replace declaration with that of...
13430         (make_pass_slp_vectorize): ...new function.
13431         (pass_complete_unroll): Replace declaration with that of...
13432         (make_pass_complete_unroll): ...new function.
13433         (pass_complete_unrolli): Replace declaration with that of...
13434         (make_pass_complete_unrolli): ...new function.
13435         (pass_parallelize_loops): Replace declaration with that of...
13436         (make_pass_parallelize_loops): ...new function.
13437         (pass_loop_prefetch): Replace declaration with that of...
13438         (make_pass_loop_prefetch): ...new function.
13439         (pass_iv_optimize): Replace declaration with that of...
13440         (make_pass_iv_optimize): ...new function.
13441         (pass_tree_loop_done): Replace declaration with that of...
13442         (make_pass_tree_loop_done): ...new function.
13443         (pass_ch): Replace declaration with that of...
13444         (make_pass_ch): ...new function.
13445         (pass_ccp): Replace declaration with that of...
13446         (make_pass_ccp): ...new function.
13447         (pass_phi_only_cprop): Replace declaration with that of...
13448         (make_pass_phi_only_cprop): ...new function.
13449         (pass_build_ssa): Replace declaration with that of...
13450         (make_pass_build_ssa): ...new function.
13451         (pass_build_alias): Replace declaration with that of...
13452         (make_pass_build_alias): ...new function.
13453         (pass_build_ealias): Replace declaration with that of...
13454         (make_pass_build_ealias): ...new function.
13455         (pass_dominator): Replace declaration with that of...
13456         (make_pass_dominator): ...new function.
13457         (pass_dce): Replace declaration with that of...
13458         (make_pass_dce): ...new function.
13459         (pass_dce_loop): Replace declaration with that of...
13460         (make_pass_dce_loop): ...new function.
13461         (pass_cd_dce): Replace declaration with that of...
13462         (make_pass_cd_dce): ...new function.
13463         (pass_call_cdce): Replace declaration with that of...
13464         (make_pass_call_cdce): ...new function.
13465         (pass_merge_phi): Replace declaration with that of...
13466         (make_pass_merge_phi): ...new function.
13467         (pass_split_crit_edges): Replace declaration with that of...
13468         (make_pass_split_crit_edges): ...new function.
13469         (pass_pre): Replace declaration with that of...
13470         (make_pass_pre): ...new function.
13471         (pass_profile): Replace declaration with that of...
13472         (make_pass_profile): ...new function.
13473         (pass_strip_predict_hints): Replace declaration with that of...
13474         (make_pass_strip_predict_hints): ...new function.
13475         (pass_lower_complex_O0): Replace declaration with that of...
13476         (make_pass_lower_complex_O0): ...new function.
13477         (pass_lower_complex): Replace declaration with that of...
13478         (make_pass_lower_complex): ...new function.
13479         (pass_lower_vector): Replace declaration with that of...
13480         (make_pass_lower_vector): ...new function.
13481         (pass_lower_vector_ssa): Replace declaration with that of...
13482         (make_pass_lower_vector_ssa): ...new function.
13483         (pass_lower_omp): Replace declaration with that of...
13484         (make_pass_lower_omp): ...new function.
13485         (pass_diagnose_omp_blocks): Replace declaration with that of...
13486         (make_pass_diagnose_omp_blocks): ...new function.
13487         (pass_expand_omp): Replace declaration with that of...
13488         (make_pass_expand_omp): ...new function.
13489         (pass_expand_omp_ssa): Replace declaration with that of...
13490         (make_pass_expand_omp_ssa): ...new function.
13491         (pass_object_sizes): Replace declaration with that of...
13492         (make_pass_object_sizes): ...new function.
13493         (pass_strlen): Replace declaration with that of...
13494         (make_pass_strlen): ...new function.
13495         (pass_fold_builtins): Replace declaration with that of...
13496         (make_pass_fold_builtins): ...new function.
13497         (pass_stdarg): Replace declaration with that of...
13498         (make_pass_stdarg): ...new function.
13499         (pass_early_warn_uninitialized): Replace declaration with that of...
13500         (make_pass_early_warn_uninitialized): ...new function.
13501         (pass_late_warn_uninitialized): Replace declaration with that of...
13502         (make_pass_late_warn_uninitialized): ...new function.
13503         (pass_cse_reciprocals): Replace declaration with that of...
13504         (make_pass_cse_reciprocals): ...new function.
13505         (pass_cse_sincos): Replace declaration with that of...
13506         (make_pass_cse_sincos): ...new function.
13507         (pass_optimize_bswap): Replace declaration with that of...
13508         (make_pass_optimize_bswap): ...new function.
13509         (pass_optimize_widening_mul): Replace declaration with that of...
13510         (make_pass_optimize_widening_mul): ...new function.
13511         (pass_warn_function_return): Replace declaration with that of...
13512         (make_pass_warn_function_return): ...new function.
13513         (pass_warn_function_noreturn): Replace declaration with that of...
13514         (make_pass_warn_function_noreturn): ...new function.
13515         (pass_cselim): Replace declaration with that of...
13516         (make_pass_cselim): ...new function.
13517         (pass_phiopt): Replace declaration with that of...
13518         (make_pass_phiopt): ...new function.
13519         (pass_forwprop): Replace declaration with that of...
13520         (make_pass_forwprop): ...new function.
13521         (pass_phiprop): Replace declaration with that of...
13522         (make_pass_phiprop): ...new function.
13523         (pass_tree_ifcombine): Replace declaration with that of...
13524         (make_pass_tree_ifcombine): ...new function.
13525         (pass_dse): Replace declaration with that of...
13526         (make_pass_dse): ...new function.
13527         (pass_nrv): Replace declaration with that of...
13528         (make_pass_nrv): ...new function.
13529         (pass_rename_ssa_copies): Replace declaration with that of...
13530         (make_pass_rename_ssa_copies): ...new function.
13531         (pass_sink_code): Replace declaration with that of...
13532         (make_pass_sink_code): ...new function.
13533         (pass_fre): Replace declaration with that of...
13534         (make_pass_fre): ...new function.
13535         (pass_check_data_deps): Replace declaration with that of...
13536         (make_pass_check_data_deps): ...new function.
13537         (pass_copy_prop): Replace declaration with that of...
13538         (make_pass_copy_prop): ...new function.
13539         (pass_vrp): Replace declaration with that of...
13540         (make_pass_vrp): ...new function.
13541         (pass_uncprop): Replace declaration with that of...
13542         (make_pass_uncprop): ...new function.
13543         (pass_return_slot): Replace declaration with that of...
13544         (make_pass_return_slot): ...new function.
13545         (pass_reassoc): Replace declaration with that of...
13546         (make_pass_reassoc): ...new function.
13547         (pass_rebuild_cgraph_edges): Replace declaration with that of...
13548         (make_pass_rebuild_cgraph_edges): ...new function.
13549         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
13550         (make_pass_remove_cgraph_callee_edges): ...new function.
13551         (pass_build_cgraph_edges): Replace declaration with that of...
13552         (make_pass_build_cgraph_edges): ...new function.
13553         (pass_local_pure_const): Replace declaration with that of...
13554         (make_pass_local_pure_const): ...new function.
13555         (pass_tracer): Replace declaration with that of...
13556         (make_pass_tracer): ...new function.
13557         (pass_warn_unused_result): Replace declaration with that of...
13558         (make_pass_warn_unused_result): ...new function.
13559         (pass_diagnose_tm_blocks): Replace declaration with that of...
13560         (make_pass_diagnose_tm_blocks): ...new function.
13561         (pass_lower_tm): Replace declaration with that of...
13562         (make_pass_lower_tm): ...new function.
13563         (pass_tm_init): Replace declaration with that of...
13564         (make_pass_tm_init): ...new function.
13565         (pass_tm_mark): Replace declaration with that of...
13566         (make_pass_tm_mark): ...new function.
13567         (pass_tm_memopt): Replace declaration with that of...
13568         (make_pass_tm_memopt): ...new function.
13569         (pass_tm_edges): Replace declaration with that of...
13570         (make_pass_tm_edges): ...new function.
13571         (pass_split_functions): Replace declaration with that of...
13572         (make_pass_split_functions): ...new function.
13573         (pass_feedback_split_functions): Replace declaration with that of...
13574         (make_pass_feedback_split_functions): ...new function.
13575         (pass_strength_reduction): Replace declaration with that of...
13576         (make_pass_strength_reduction): ...new function.
13577         (pass_ipa_lower_emutls): Replace declaration with that of...
13578         (make_pass_ipa_lower_emutls): ...new function.
13579         (pass_ipa_function_and_variable_visibility): Replace declaration with
13580         that of...
13581         (make_pass_ipa_function_and_variable_visibility): ...new function.
13582         (pass_ipa_tree_profile): Replace declaration with that of...
13583         (make_pass_ipa_tree_profile): ...new function.
13584         (pass_early_local_passes): Replace declaration with that of...
13585         (make_pass_early_local_passes): ...new function.
13586         (pass_ipa_whole_program_visibility): Replace declaration with that
13587         of...
13588         (make_pass_ipa_whole_program_visibility): ...new function.
13589         (pass_ipa_lto_gimple_out): Replace declaration with that of...
13590         (make_pass_ipa_lto_gimple_out): ...new function.
13591         (pass_ipa_increase_alignment): Replace declaration with that of...
13592         (make_pass_ipa_increase_alignment): ...new function.
13593         (pass_ipa_inline): Replace declaration with that of...
13594         (make_pass_ipa_inline): ...new function.
13595         (pass_ipa_free_lang_data): Replace declaration with that of...
13596         (make_pass_ipa_free_lang_data): ...new function.
13597         (pass_ipa_free_inline_summary): Replace declaration with that of...
13598         (make_pass_ipa_free_inline_summary): ...new function.
13599         (pass_ipa_cp): Replace declaration with that of...
13600         (make_pass_ipa_cp): ...new function.
13601         (pass_ipa_reference): Replace declaration with that of...
13602         (make_pass_ipa_reference): ...new function.
13603         (pass_ipa_pure_const): Replace declaration with that of...
13604         (make_pass_ipa_pure_const): ...new function.
13605         (pass_ipa_pta): Replace declaration with that of...
13606         (make_pass_ipa_pta): ...new function.
13607         (pass_ipa_lto_finish_out): Replace declaration with that of...
13608         (make_pass_ipa_lto_finish_out): ...new function.
13609         (pass_ipa_tm): Replace declaration with that of...
13610         (make_pass_ipa_tm): ...new function.
13611         (pass_ipa_profile): Replace declaration with that of...
13612         (make_pass_ipa_profile): ...new function.
13613         (pass_ipa_cdtor_merge): Replace declaration with that of...
13614         (make_pass_ipa_cdtor_merge): ...new function.
13615         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
13616         of...
13617         (make_pass_cleanup_cfg_post_optimizing): ...new function.
13618         (pass_init_datastructures): Replace declaration with that of...
13619         (make_pass_init_datastructures): ...new function.
13620         (pass_fixup_cfg): Replace declaration with that of...
13621         (make_pass_fixup_cfg): ...new function.
13622         (pass_expand): Replace declaration with that of...
13623         (make_pass_expand): ...new function.
13624         (pass_instantiate_virtual_regs): Replace declaration with that of...
13625         (make_pass_instantiate_virtual_regs): ...new function.
13626         (pass_rtl_fwprop): Replace declaration with that of...
13627         (make_pass_rtl_fwprop): ...new function.
13628         (pass_rtl_fwprop_addr): Replace declaration with that of...
13629         (make_pass_rtl_fwprop_addr): ...new function.
13630         (pass_jump): Replace declaration with that of...
13631         (make_pass_jump): ...new function.
13632         (pass_jump2): Replace declaration with that of...
13633         (make_pass_jump2): ...new function.
13634         (pass_lower_subreg): Replace declaration with that of...
13635         (make_pass_lower_subreg): ...new function.
13636         (pass_cse): Replace declaration with that of...
13637         (make_pass_cse): ...new function.
13638         (pass_fast_rtl_dce): Replace declaration with that of...
13639         (make_pass_fast_rtl_dce): ...new function.
13640         (pass_ud_rtl_dce): Replace declaration with that of...
13641         (make_pass_ud_rtl_dce): ...new function.
13642         (pass_rtl_dce): Replace declaration with that of...
13643         (make_pass_rtl_dce): ...new function.
13644         (pass_rtl_dse1): Replace declaration with that of...
13645         (make_pass_rtl_dse1): ...new function.
13646         (pass_rtl_dse2): Replace declaration with that of...
13647         (make_pass_rtl_dse2): ...new function.
13648         (pass_rtl_dse3): Replace declaration with that of...
13649         (make_pass_rtl_dse3): ...new function.
13650         (pass_rtl_cprop): Replace declaration with that of...
13651         (make_pass_rtl_cprop): ...new function.
13652         (pass_rtl_pre): Replace declaration with that of...
13653         (make_pass_rtl_pre): ...new function.
13654         (pass_rtl_hoist): Replace declaration with that of...
13655         (make_pass_rtl_hoist): ...new function.
13656         (pass_rtl_store_motion): Replace declaration with that of...
13657         (make_pass_rtl_store_motion): ...new function.
13658         (pass_cse_after_global_opts): Replace declaration with that of...
13659         (make_pass_cse_after_global_opts): ...new function.
13660         (pass_rtl_ifcvt): Replace declaration with that of...
13661         (make_pass_rtl_ifcvt): ...new function.
13662         (pass_into_cfg_layout_mode): Replace declaration with that of...
13663         (make_pass_into_cfg_layout_mode): ...new function.
13664         (pass_outof_cfg_layout_mode): Replace declaration with that of...
13665         (make_pass_outof_cfg_layout_mode): ...new function.
13666         (pass_loop2): Replace declaration with that of...
13667         (make_pass_loop2): ...new function.
13668         (pass_rtl_loop_init): Replace declaration with that of...
13669         (make_pass_rtl_loop_init): ...new function.
13670         (pass_rtl_move_loop_invariants): Replace declaration with that of...
13671         (make_pass_rtl_move_loop_invariants): ...new function.
13672         (pass_rtl_unswitch): Replace declaration with that of...
13673         (make_pass_rtl_unswitch): ...new function.
13674         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
13675         (make_pass_rtl_unroll_and_peel_loops): ...new function.
13676         (pass_rtl_doloop): Replace declaration with that of...
13677         (make_pass_rtl_doloop): ...new function.
13678         (pass_rtl_loop_done): Replace declaration with that of...
13679         (make_pass_rtl_loop_done): ...new function.
13680         (pass_web): Replace declaration with that of...
13681         (make_pass_web): ...new function.
13682         (pass_cse2): Replace declaration with that of...
13683         (make_pass_cse2): ...new function.
13684         (pass_df_initialize_opt): Replace declaration with that of...
13685         (make_pass_df_initialize_opt): ...new function.
13686         (pass_df_initialize_no_opt): Replace declaration with that of...
13687         (make_pass_df_initialize_no_opt): ...new function.
13688         (pass_reginfo_init): Replace declaration with that of...
13689         (make_pass_reginfo_init): ...new function.
13690         (pass_inc_dec): Replace declaration with that of...
13691         (make_pass_inc_dec): ...new function.
13692         (pass_stack_ptr_mod): Replace declaration with that of...
13693         (make_pass_stack_ptr_mod): ...new function.
13694         (pass_initialize_regs): Replace declaration with that of...
13695         (make_pass_initialize_regs): ...new function.
13696         (pass_combine): Replace declaration with that of...
13697         (make_pass_combine): ...new function.
13698         (pass_if_after_combine): Replace declaration with that of...
13699         (make_pass_if_after_combine): ...new function.
13700         (pass_ree): Replace declaration with that of...
13701         (make_pass_ree): ...new function.
13702         (pass_partition_blocks): Replace declaration with that of...
13703         (make_pass_partition_blocks): ...new function.
13704         (pass_match_asm_constraints): Replace declaration with that of...
13705         (make_pass_match_asm_constraints): ...new function.
13706         (pass_regmove): Replace declaration with that of...
13707         (make_pass_regmove): ...new function.
13708         (pass_split_all_insns): Replace declaration with that of...
13709         (make_pass_split_all_insns): ...new function.
13710         (pass_fast_rtl_byte_dce): Replace declaration with that of...
13711         (make_pass_fast_rtl_byte_dce): ...new function.
13712         (pass_lower_subreg2): Replace declaration with that of...
13713         (make_pass_lower_subreg2): ...new function.
13714         (pass_mode_switching): Replace declaration with that of...
13715         (make_pass_mode_switching): ...new function.
13716         (pass_sms): Replace declaration with that of...
13717         (make_pass_sms): ...new function.
13718         (pass_sched): Replace declaration with that of...
13719         (make_pass_sched): ...new function.
13720         (pass_ira): Replace declaration with that of...
13721         (make_pass_ira): ...new function.
13722         (pass_reload): Replace declaration with that of...
13723         (make_pass_reload): ...new function.
13724         (pass_clean_state): Replace declaration with that of...
13725         (make_pass_clean_state): ...new function.
13726         (pass_branch_prob): Replace declaration with that of...
13727         (make_pass_branch_prob): ...new function.
13728         (pass_value_profile_transformations): Replace declaration with that
13729         of...
13730         (make_pass_value_profile_transformations): ...new function.
13731         (pass_postreload_cse): Replace declaration with that of...
13732         (make_pass_postreload_cse): ...new function.
13733         (pass_gcse2): Replace declaration with that of...
13734         (make_pass_gcse2): ...new function.
13735         (pass_split_after_reload): Replace declaration with that of...
13736         (make_pass_split_after_reload): ...new function.
13737         (pass_branch_target_load_optimize1): Replace declaration with that
13738         of...
13739         (make_pass_branch_target_load_optimize1): ...new function.
13740         (pass_thread_prologue_and_epilogue): Replace declaration with that
13741         of...
13742         (make_pass_thread_prologue_and_epilogue): ...new function.
13743         (pass_stack_adjustments): Replace declaration with that of...
13744         (make_pass_stack_adjustments): ...new function.
13745         (pass_peephole2): Replace declaration with that of...
13746         (make_pass_peephole2): ...new function.
13747         (pass_if_after_reload): Replace declaration with that of...
13748         (make_pass_if_after_reload): ...new function.
13749         (pass_regrename): Replace declaration with that of...
13750         (make_pass_regrename): ...new function.
13751         (pass_cprop_hardreg): Replace declaration with that of...
13752         (make_pass_cprop_hardreg): ...new function.
13753         (pass_reorder_blocks): Replace declaration with that of...
13754         (make_pass_reorder_blocks): ...new function.
13755         (pass_branch_target_load_optimize2): Replace declaration with that
13756         of...
13757         (make_pass_branch_target_load_optimize2): ...new function.
13758         (pass_leaf_regs): Replace declaration with that of...
13759         (make_pass_leaf_regs): ...new function.
13760         (pass_split_before_sched2): Replace declaration with that of...
13761         (make_pass_split_before_sched2): ...new function.
13762         (pass_compare_elim_after_reload): Replace declaration with that of...
13763         (make_pass_compare_elim_after_reload): ...new function.
13764         (pass_sched2): Replace declaration with that of...
13765         (make_pass_sched2): ...new function.
13766         (pass_stack_regs): Replace declaration with that of...
13767         (make_pass_stack_regs): ...new function.
13768         (pass_stack_regs_run): Replace declaration with that of...
13769         (make_pass_stack_regs_run): ...new function.
13770         (pass_df_finish): Replace declaration with that of...
13771         (make_pass_df_finish): ...new function.
13772         (pass_compute_alignments): Replace declaration with that of...
13773         (make_pass_compute_alignments): ...new function.
13774         (pass_duplicate_computed_gotos): Replace declaration with that of...
13775         (make_pass_duplicate_computed_gotos): ...new function.
13776         (pass_variable_tracking): Replace declaration with that of...
13777         (make_pass_variable_tracking): ...new function.
13778         (pass_free_cfg): Replace declaration with that of...
13779         (make_pass_free_cfg): ...new function.
13780         (pass_machine_reorg): Replace declaration with that of...
13781         (make_pass_machine_reorg): ...new function.
13782         (pass_cleanup_barriers): Replace declaration with that of...
13783         (make_pass_cleanup_barriers): ...new function.
13784         (pass_delay_slots): Replace declaration with that of...
13785         (make_pass_delay_slots): ...new function.
13786         (pass_split_for_shorten_branches): Replace declaration with that of...
13787         (make_pass_split_for_shorten_branches): ...new function.
13788         (pass_split_before_regstack): Replace declaration with that of...
13789         (make_pass_split_before_regstack): ...new function.
13790         (pass_convert_to_eh_region_ranges): Replace declaration with that
13791         of...
13792         (make_pass_convert_to_eh_region_ranges): ...new function.
13793         (pass_shorten_branches): Replace declaration with that of...
13794         (make_pass_shorten_branches): ...new function.
13795         (pass_set_nothrow_function_flags): Replace declaration with that of...
13796         (make_pass_set_nothrow_function_flags): ...new function.
13797         (pass_dwarf2_frame): Replace declaration with that of...
13798         (make_pass_dwarf2_frame): ...new function.
13799         (pass_final): Replace declaration with that of...
13800         (make_pass_final): ...new function.
13801         (pass_rtl_seqabstr): Replace declaration with that of...
13802         (make_pass_rtl_seqabstr): ...new function.
13803         (pass_release_ssa_names): Replace declaration with that of...
13804         (make_pass_release_ssa_names): ...new function.
13805         (pass_early_inline): Replace declaration with that of...
13806         (make_pass_early_inline): ...new function.
13807         (pass_inline_parameters): Replace declaration with that of...
13808         (make_pass_inline_parameters): ...new function.
13809         (pass_update_address_taken): Replace declaration with that of...
13810         (make_pass_update_address_taken): ...new function.
13811         (pass_convert_switch): Replace declaration with that of...
13812         (make_pass_convert_switch): ...new function.
13813         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
13814         to a subclass of simple_ipa_opt_pass along with...
13815         (pass_data_ipa_tree_profile): ...new pass_data instance and...
13816         (make_pass_ipa_tree_profile): ...new function.
13817         * tree-sra.c (pass_sra_early): Convert from a global struct to a
13818         subclass of gimple_opt_pass along with...
13819         (pass_data_sra_early): ...new pass_data instance and...
13820         (make_pass_sra_early): ...new function.
13821         (pass_sra): Convert from a global struct to a subclass of
13822         gimple_opt_pass along with...
13823         (pass_data_sra): ...new pass_data instance and...
13824         (make_pass_sra): ...new function.
13825         (pass_early_ipa_sra): Convert from a global struct to a subclass of
13826         gimple_opt_pass along with...
13827         (pass_data_early_ipa_sra): ...new pass_data instance and...
13828         (make_pass_early_ipa_sra): ...new function.
13829         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
13830         subclass of gimple_opt_pass along with...
13831         (pass_data_ccp): ...new pass_data instance and...
13832         (make_pass_ccp): ...new function.
13833         (pass_fold_builtins): Convert from a global struct to a subclass of
13834         gimple_opt_pass along with...
13835         (pass_data_fold_builtins): ...new pass_data instance and...
13836         (make_pass_fold_builtins): ...new function.
13837         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
13838         subclass of gimple_opt_pass along with...
13839         (pass_data_copy_prop): ...new pass_data instance and...
13840         (make_pass_copy_prop): ...new function.
13841         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
13842         global struct to a subclass of gimple_opt_pass along with...
13843         (pass_data_rename_ssa_copies): ...new pass_data instance and...
13844         (make_pass_rename_ssa_copies): ...new function.
13845         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
13846         subclass of gimple_opt_pass along with...
13847         (pass_data_dce): ...new pass_data instance and...
13848         (make_pass_dce): ...new function.
13849         (pass_dce_loop): Convert from a global struct to a subclass of
13850         gimple_opt_pass along with...
13851         (pass_data_dce_loop): ...new pass_data instance and...
13852         (make_pass_dce_loop): ...new function.
13853         (pass_cd_dce): Convert from a global struct to a subclass of
13854         gimple_opt_pass along with...
13855         (pass_data_cd_dce): ...new pass_data instance and...
13856         (make_pass_cd_dce): ...new function.
13857         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
13858         subclass of gimple_opt_pass along with...
13859         (pass_data_dominator): ...new pass_data instance and...
13860         (make_pass_dominator): ...new function.
13861         (pass_phi_only_cprop): Convert from a global struct to a subclass of
13862         gimple_opt_pass along with...
13863         (pass_data_phi_only_cprop): ...new pass_data instance and...
13864         (make_pass_phi_only_cprop): ...new function.
13865         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
13866         subclass of gimple_opt_pass along with...
13867         (pass_data_dse): ...new pass_data instance and...
13868         (make_pass_dse): ...new function.
13869         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
13870         a subclass of gimple_opt_pass along with...
13871         (pass_data_forwprop): ...new pass_data instance and...
13872         (make_pass_forwprop): ...new function.
13873         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
13874         struct to a subclass of gimple_opt_pass along with...
13875         (pass_data_tree_ifcombine): ...new pass_data instance and...
13876         (make_pass_tree_ifcombine): ...new function.
13877         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
13878         subclass of gimple_opt_pass along with...
13879         (pass_data_ch): ...new pass_data instance and...
13880         (make_pass_ch): ...new function.
13881         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
13882         subclass of gimple_opt_pass along with...
13883         (pass_data_tree_loop): ...new pass_data instance and...
13884         (make_pass_tree_loop): ...new function.
13885         (pass_tree_loop_init): Convert from a global struct to a subclass of
13886         gimple_opt_pass along with...
13887         (pass_data_tree_loop_init): ...new pass_data instance and...
13888         (make_pass_tree_loop_init): ...new function.
13889         (pass_lim): Convert from a global struct to a subclass of
13890         gimple_opt_pass along with...
13891         (pass_data_lim): ...new pass_data instance and...
13892         (make_pass_lim): ...new function.
13893         (pass_tree_unswitch): Convert from a global struct to a subclass of
13894         gimple_opt_pass along with...
13895         (pass_data_tree_unswitch): ...new pass_data instance and...
13896         (make_pass_tree_unswitch): ...new function.
13897         (pass_predcom): Convert from a global struct to a subclass of
13898         gimple_opt_pass along with...
13899         (pass_data_predcom): ...new pass_data instance and...
13900         (make_pass_predcom): ...new function.
13901         (pass_vectorize): Convert from a global struct to a subclass of
13902         gimple_opt_pass along with...
13903         (pass_data_vectorize): ...new pass_data instance and...
13904         (make_pass_vectorize): ...new function.
13905         (pass_graphite): Convert from a global struct to a subclass of
13906         gimple_opt_pass along with...
13907         (pass_data_graphite): ...new pass_data instance and...
13908         (make_pass_graphite): ...new function.
13909         (pass_graphite_transforms): Convert from a global struct to a subclass
13910         of gimple_opt_pass along with...
13911         (pass_data_graphite_transforms): ...new pass_data instance and...
13912         (make_pass_graphite_transforms): ...new function.
13913         (pass_check_data_deps): Convert from a global struct to a subclass of
13914         gimple_opt_pass along with...
13915         (pass_data_check_data_deps): ...new pass_data instance and...
13916         (make_pass_check_data_deps): ...new function.
13917         (pass_iv_canon): Convert from a global struct to a subclass of
13918         gimple_opt_pass along with...
13919         (pass_data_iv_canon): ...new pass_data instance and...
13920         (make_pass_iv_canon): ...new function.
13921         (pass_scev_cprop): Convert from a global struct to a subclass of
13922         gimple_opt_pass along with...
13923         (pass_data_scev_cprop): ...new pass_data instance and...
13924         (make_pass_scev_cprop): ...new function.
13925         (pass_record_bounds): Convert from a global struct to a subclass of
13926         gimple_opt_pass along with...
13927         (pass_data_record_bounds): ...new pass_data instance and...
13928         (make_pass_record_bounds): ...new function.
13929         (pass_complete_unroll): Convert from a global struct to a subclass of
13930         gimple_opt_pass along with...
13931         (pass_data_complete_unroll): ...new pass_data instance and...
13932         (make_pass_complete_unroll): ...new function.
13933         (pass_complete_unrolli): Convert from a global struct to a subclass of
13934         gimple_opt_pass along with...
13935         (pass_data_complete_unrolli): ...new pass_data instance and...
13936         (make_pass_complete_unrolli): ...new function.
13937         (pass_parallelize_loops): Convert from a global struct to a subclass
13938         of gimple_opt_pass along with...
13939         (pass_data_parallelize_loops): ...new pass_data instance and...
13940         (make_pass_parallelize_loops): ...new function.
13941         (pass_loop_prefetch): Convert from a global struct to a subclass of
13942         gimple_opt_pass along with...
13943         (pass_data_loop_prefetch): ...new pass_data instance and...
13944         (make_pass_loop_prefetch): ...new function.
13945         (pass_iv_optimize): Convert from a global struct to a subclass of
13946         gimple_opt_pass along with...
13947         (pass_data_iv_optimize): ...new pass_data instance and...
13948         (make_pass_iv_optimize): ...new function.
13949         (pass_tree_loop_done): Convert from a global struct to a subclass of
13950         gimple_opt_pass along with...
13951         (pass_data_tree_loop_done): ...new pass_data instance and...
13952         (make_pass_tree_loop_done): ...new function.
13953         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
13954         struct to a subclass of gimple_opt_pass along with...
13955         (pass_data_cse_reciprocals): ...new pass_data instance and...
13956         (make_pass_cse_reciprocals): ...new function.
13957         (pass_cse_sincos): Convert from a global struct to a subclass of
13958         gimple_opt_pass along with...
13959         (pass_data_cse_sincos): ...new pass_data instance and...
13960         (make_pass_cse_sincos): ...new function.
13961         (pass_optimize_bswap): Convert from a global struct to a subclass of
13962         gimple_opt_pass along with...
13963         (pass_data_optimize_bswap): ...new pass_data instance and...
13964         (make_pass_optimize_bswap): ...new function.
13965         (pass_optimize_widening_mul): Convert from a global struct to a
13966         subclass of gimple_opt_pass along with...
13967         (pass_data_optimize_widening_mul): ...new pass_data instance and...
13968         (make_pass_optimize_widening_mul): ...new function.
13969         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
13970         subclass of gimple_opt_pass along with...
13971         (pass_data_phiopt): ...new pass_data instance and...
13972         (make_pass_phiopt): ...new function.
13973         (pass_cselim): Convert from a global struct to a subclass of
13974         gimple_opt_pass along with...
13975         (pass_data_cselim): ...new pass_data instance and...
13976         (make_pass_cselim): ...new function.
13977         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
13978         subclass of gimple_opt_pass along with...
13979         (pass_data_phiprop): ...new pass_data instance and...
13980         (make_pass_phiprop): ...new function.
13981         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
13982         subclass of gimple_opt_pass along with...
13983         (pass_data_pre): ...new pass_data instance and...
13984         (make_pass_pre): ...new function.
13985         (pass_fre): Convert from a global struct to a subclass of
13986         gimple_opt_pass along with...
13987         (pass_data_fre): ...new pass_data instance and...
13988         (make_pass_fre): ...new function.
13989         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
13990         subclass of gimple_opt_pass along with...
13991         (pass_data_reassoc): ...new pass_data instance and...
13992         (make_pass_reassoc): ...new function.
13993         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
13994         subclass of gimple_opt_pass along with...
13995         (pass_data_sink_code): ...new pass_data instance and...
13996         (make_pass_sink_code): ...new function.
13997         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
13998         subclass of gimple_opt_pass along with...
13999         (pass_data_strlen): ...new pass_data instance and...
14000         (make_pass_strlen): ...new function.
14001         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
14002         struct to a subclass of gimple_opt_pass along with...
14003         (pass_data_build_alias): ...new pass_data instance and...
14004         (make_pass_build_alias): ...new function.
14005         (pass_build_ealias): Convert from a global struct to a subclass of
14006         gimple_opt_pass along with...
14007         (pass_data_build_ealias): ...new pass_data instance and...
14008         (make_pass_build_ealias): ...new function.
14009         (pass_ipa_pta): Convert from a global struct to a subclass of
14010         simple_ipa_opt_pass along with...
14011         (pass_data_ipa_pta): ...new pass_data instance and...
14012         (make_pass_ipa_pta): ...new function.
14013         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
14014         subclass of gimple_opt_pass along with...
14015         (pass_data_uncprop): ...new pass_data instance and...
14016         (make_pass_uncprop): ...new function.
14017         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
14018         global struct to a subclass of gimple_opt_pass along with...
14019         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
14020         (make_pass_late_warn_uninitialized): ...new function.
14021         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
14022         to a subclass of gimple_opt_pass along with...
14023         (pass_data_init_datastructures): ...new pass_data instance and...
14024         (make_pass_init_datastructures): ...new function.
14025         (pass_early_warn_uninitialized): Convert from a global struct to a
14026         subclass of gimple_opt_pass along with...
14027         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
14028         (make_pass_early_warn_uninitialized): ...new function.
14029         (pass_update_address_taken): Convert from a global struct to a
14030         subclass of gimple_opt_pass along with...
14031         (pass_data_update_address_taken): ...new pass_data instance and...
14032         (make_pass_update_address_taken): ...new function.
14033         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
14034         struct to a subclass of gimple_opt_pass along with...
14035         (pass_data_release_ssa_names): ...new pass_data instance and...
14036         (make_pass_release_ssa_names): ...new function.
14037         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
14038         subclass of gimple_opt_pass along with...
14039         (pass_data_stdarg): ...new pass_data instance and...
14040         (make_pass_stdarg): ...new function.
14041         * tree-switch-conversion.c (pass_convert_switch): Convert from a
14042         global struct to a subclass of gimple_opt_pass along with...
14043         (pass_data_convert_switch): ...new pass_data instance and...
14044         (make_pass_convert_switch): ...new function.
14045         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
14046         to a subclass of gimple_opt_pass along with...
14047         (pass_data_tail_recursion): ...new pass_data instance and...
14048         (make_pass_tail_recursion): ...new function.
14049         (pass_tail_calls): Convert from a global struct to a subclass of
14050         gimple_opt_pass along with...
14051         (pass_data_tail_calls): ...new pass_data instance and...
14052         (make_pass_tail_calls): ...new function.
14053         * tree-vect-generic.c (pass_lower_vector): Convert from a global
14054         struct to a subclass of gimple_opt_pass along with...
14055         (pass_data_lower_vector): ...new pass_data instance and...
14056         (make_pass_lower_vector): ...new function.
14057         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
14058         gimple_opt_pass along with...
14059         (pass_data_lower_vector_ssa): ...new pass_data instance and...
14060         (make_pass_lower_vector_ssa): ...new function.
14061         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
14062         to a subclass of gimple_opt_pass along with...
14063         (pass_data_slp_vectorize): ...new pass_data instance and...
14064         (make_pass_slp_vectorize): ...new function.
14065         (pass_ipa_increase_alignment): Convert from a global struct to a
14066         subclass of simple_ipa_opt_pass along with...
14067         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
14068         (make_pass_ipa_increase_alignment): ...new function.
14069         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
14070         gimple_opt_pass along with...
14071         (pass_data_vrp): ...new pass_data instance and...
14072         (make_pass_vrp): ...new function.
14073         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
14074         subclass of simple_ipa_opt_pass along with...
14075         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
14076         (make_pass_ipa_free_lang_data): ...new function.
14077         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
14078         gimple_opt_pass along with...
14079         (pass_data_tsan): ...new pass_data instance and...
14080         (make_pass_tsan): ...new function.
14081         (pass_tsan_O0): Convert from a global struct to a subclass of
14082         gimple_opt_pass along with...
14083         (pass_data_tsan_O0): ...new pass_data instance and...
14084         (make_pass_tsan_O0): ...new function.
14085         * var-tracking.c (pass_variable_tracking): Convert from a global
14086         struct to a subclass of rtl_opt_pass along with...
14087         (pass_data_variable_tracking): ...new pass_data instance and...
14088         (make_pass_variable_tracking): ...new function.
14089         * web.c (pass_web): Convert from a global struct to a subclass of
14090         rtl_opt_pass along with...
14091         (pass_data_web): ...new pass_data instance and...
14092         (make_pass_web): ...new function.
14093         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
14094         declaration with that of...
14095         (make_pass_mode_switch_use): ...new function.
14096         (pass_resolve_sw_modes): Replace declaration with that of...
14097         (make_pass_resolve_sw_modes): ...new function.
14098         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
14099         from a global struct to a subclass of rtl_opt_pass along with...
14100         (pass_data_mode_switch_use): ...new pass_data instance and...
14101         (make_pass_mode_switch_use): ...new function.
14102         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
14103         from a global struct to a subclass of rtl_opt_pass along with...
14104         (pass_data_resolve_sw_modes): ...new pass_data instance and...
14105         (make_pass_resolve_sw_modes): ...new function.
14106         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
14107         struct to a subclass of rtl_opt_pass along with...
14108         (pass_data_insert_vzeroupper): ...new pass_data instance and...
14109         (make_pass_insert_vzeroupper): ...new function.
14110         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
14111         global struct to a subclass of rtl_opt_pass along with...
14112         (pass_data_work_around_errata): ...new pass_data instance and...
14113         (make_pass_work_around_errata): ...new function.
14114         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
14115         struct to a subclass of rtl_opt_pass along with...
14116         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
14117         (make_pass_mips_machine_reorg2): ...new function.
14119 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
14121         * passes.c (pass_manager::operator new): New.
14123 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
14125         Handwritten part of conversion of passes to C++ classes.
14127         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
14128         (toplev.o): Add dep on PASS_MANAGER_H.
14129         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
14130         of early local pases to reflect this moving from a global to a
14131         member of gcc::pass_manager.
14132         (cgraph_add_new_function): Likewise.
14133         * lto-cgraph.c (lto_output_node): Update for conversion of
14134         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
14135         * passes.c (opt_pass::clone): New.
14136         (opt_pass::gate): New.
14137         (opt_pass::execute): New.
14138         (opt_pass::opt_pass): New.
14139         (pass_manager::execute_early_local_passes): New.
14140         (pass_manager::execute_pass_mode_switching): new.
14141         (finish_optimization_passes): Convert to...
14142         (pass_manager::finish_optimization_passes): ...this.
14143         (finish_optimization_passes): Update for conversion of passes to
14144         C++ classes.
14145         (register_dump_files_1): Use has_gate since we cannot portably
14146         check a vtable entry against NULL.
14147         (dump_one_pass): Likewise.
14148         (ipa_write_summaries_2): Likewise.
14149         (ipa_write_optimization_summaries_1): Likewise.
14150         (ipa_read_summaries_1): Likewise.
14151         (ipa_read_optimization_summaries_1): Likewise.
14152         (execute_ipa_stmt_fixups): Likewise.
14153         (pass_manager::pass_manager): Rewrite pass-creation, invoking
14154         pass-creation functions rather than wiring up globals, and
14155         storing the results in fields of pass_manager generated using
14156         pass-instances.def.
14157         (pass_manager::dump_profile_report): Update for conversion of
14158         passes to C++ classes.
14159         (pass_manager::execute_ipa_summary_passes): Likewise.
14160         (execute_one_ipa_transform_pass): Likewise.
14161         (execute_one_pass): Use has_gate and has_execute since we cannot
14162         portably check a vtable entry against NULL.
14163         * pass_manager.h (pass_manager::finish_optimization_passes): New.
14164         (pass_manager): Use pass-instances.def to add fields for the
14165         various pass instances.
14166         * toplev.c (finalize): Update for move of
14167         finish_optimization_passes to a method of gcc::pass_manager.
14168         * toplev.h (finish_optimization_passes): Move to method of class
14169         pass_manager.
14170         * tree-pass.h (struct pass_data): New.
14171         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
14172         (opt_pass::gate): Convert to virtual function.
14173         (opt_pass::~opt_pass): New.
14174         (opt_pass::clone): New.
14175         (opt_pass::execute): Convert to virtual function.
14176         (opt_pass::opt_pass): New.
14177         (opt_pass::ctxt_): new.
14178         (gimple_opt_pass): Convert to subclass of opt_pass.
14179         (gimple_opt_pass::gimple_opt_pass): New.
14180         (rtl_opt_pass): Convert to subclass of opt_pass.
14181         (rtl_opt_pass::rtl_opt_pass): New.
14182         (ipa_opt_pass_d): Convert to subclass of opt_pass.
14183         (ipa_opt_pass_d::ipa_opt_pass_d): New.
14184         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
14185         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
14186         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
14187         invocation of pass_mode_switching to reflect this moving from a
14188         global to a member of gcc::pass_manager.
14189         (ix86_option_override): Rework how pass_insert_vzeroupper is
14190         added to the pass_manager to reflect autogenerated changes.
14191         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
14193 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
14195         PR rtl-optimization/57708
14196         * recog.c (peep2_find_free_register): Validate all regs in a
14197         multi-reg mode.
14199 2013-08-05  Jan Hubicka  <jh@suse.cz>
14201         PR lto/57602
14202         * cgraph.c (verify_cgraph_node): Accept local flags from other
14203         partitions.
14204         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
14205         (function_and_variable_visibility): Likewise.
14206         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
14208 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
14210         * graph.c (init_graph_slim_pretty_print): Remove.
14211         (print_graph_cfg): Do not call it.  Use local pretty printer.
14212         (start_graph_dump): Likewise.
14214 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
14216         * gimple-pretty-print.c (buffer): Remove.
14217         (initialized): Likewise.
14218         (maybe_init_pretty_print): Likewise.
14219         (print_gimple_stmt): Do not call it.  Use non-static local
14220         pretty_printer variable.
14221         (print_gimple_expr): Likewise.
14222         (print_gimple_seq): Likewise.
14223         (gimple_dump_bb): Likewise.
14225 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
14227         * asan.c (asan_pp): Remove.
14228         (asan_pp_initialized): Likewise.
14229         (asan_pp_initialize): Likewise.
14230         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
14231         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
14232         (asan_add_global): Likewise.
14234 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
14236         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
14237         * diagnostic.h (diagnostic_flush_buffer): Adjust.
14238         * pretty-print.c (pp_formatted_text_data): Likewise.
14239         (pp_indent): Rename from pp_base_indent.
14240         (pp_format): Rename from pp_base_format.
14241         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
14242         (pp_format_verbatim): Rename from pp_base_format_verbatim.
14243         (pp_flush): Rename from pp_base_flush.
14244         (pp_set_line_maximum_length): Rename from
14245         pp_base_set_line_maximum_length.
14246         (pp_clear_output_area): Rename from pp_base_clear_output_area.
14247         (pp_set_prefix): Rename from pp_base_set_prefix.
14248         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
14249         (pp_emit_prefix): Rename from pp_base_emit_prefix.
14250         (pp_append_text): Rename from pp_base_append_text.
14251         (pp_formatted_text): Rename from pp_base_formatted_text.
14252         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
14253         (pp_remaining_character_count_for_line): Rename from
14254         pp_base_remaining_character_count_for_line.
14255         (pp_newline): Rename from pp_base_newline.
14256         (pp_character): Rename from pp_base_character.
14257         (pp_string): Rename from pp_base_string.
14258         (pp_maybe_space): Rename from pp_base_maybe_space.
14259         * asan.c (asan_pp_string): Adjust.
14260         (asan_emit_stack_protection): Likewise.
14261         (asan_add_global): Likewise.
14262         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
14263         * tree-mudflap.c (mf_varname_tree): Likewise.
14264         * tree-pretty-print.c (pp_tree_identifier): Rename from
14265         pp_base_tree_identifier.
14266         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
14267         Declare as function.
14269 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
14271         * pretty-print.h (pp_bar_bar): New.
14272         (pp_ampersand_ampersand): Likewise.
14273         (pp_less_equal): Likewise.
14274         (pp_greater_equal): Likewise.
14275         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
14276         printer functions instead of pp_string or operators and punctuators.
14277         (dump_gimple_call): Likewise.
14278         (dump_gimple_omp_for): Likewise.
14279         (dump_gimple_transaction): Likewise.
14280         (dump_gimple_phi): Likewise.
14281         (pp_gimple_stmt_1): Likewise.
14282         * sched-vis.c (print_insn): Likewise.
14283         * tree-mudflap.c (mf_varname_tree): Likewise.
14284         * tree-pretty-print.c (dump_block_node): Likewise.
14285         (dump_generic_node): Likewise.
14287 2013-08-02  Jan Hubicka  <jh@suse.cz>
14289         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
14290         boundaries.
14291         * lto-streamer-out.c (tree_is_indexable): Results decls and
14292         parm decls are not indexable.
14293         (DFS_write_tree_body): Do not follow args and results.
14294         (hash_tree): Likewise.
14295         (output_functions): Rearrange so struct function is needed
14296         only when real body is output; be able to also ouptut abstract
14297         functions; output DECL_ARGUMENTS and DECL_RESULT.
14298         (lto_output): When not in WPA, ale store abstract functions.
14299         (write_symbol): Do not care about RESULT_DECL.
14300         (output_symbol_p): Handle correctly sbtract decls.
14301         * lto-streamer-in.c (input_function): Rearrange so struct
14302         function can be NULL at entry; allow streaming of
14303         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
14304         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
14305         sanity check during LTO.
14306         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
14307         RESULT_DECl and DECL_ARGUMENTS.
14308         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
14309         Likewise.
14311 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
14313         * pretty-print.h (pp_underscore): New.
14314         (pp_comma): Tidy.
14315         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
14316         printer functions instead of pp_character.
14317         (dump_binary_rhs): Likewise.
14318         (dump_ternary_rhs): Likewise.
14319         (dump_gimple_call_args): Likewise.
14320         (pp_points_to_solution): Likewise.
14321         (dump_gimple_call): Likewise.
14322         (dump_gimple_switch): Likewise.
14323         (dump_gimple_cond): Likewise.
14324         (dump_gimple_bind): Likewise.
14325         (dump_gimple_try): Likewise.
14326         (dump_gimple_omp_for): Likewise.
14327         (dump_gimple_omp_continue): Likewise.
14328         (dump_gimple_omp_single): Likewise.
14329         (dump_gimple_omp_sections): Likewise.
14330         (dump_gimple_omp_block): Likewise.
14331         (dump_gimple_omp_critical): Likewise.
14332         (dump_gimple_transaction): Likewise.
14333         (dump_gimple_asm): Likewise.
14334         (dump_gimple_phi): Likewise.
14335         (dump_gimple_omp_parallel): Likewise.
14336         (dump_gimple_omp_task): Likewise.
14337         (dump_gimple_omp_atomic_load): Likewise.
14338         (dump_gimple_omp_atomic_store): Likewise.
14339         (dump_gimple_mem_ops): Likewise.
14340         (pp_gimple_stmt_1): Likewise.
14341         (pp_cfg_jump): Likewise.
14342         (dump_implicit_edges): Likewise.
14343         (gimple_dump_bb_for_graph): Likewise.
14344         * graph.c (draw_cfg_node): Likewise.
14345         * langhooks.c (lhd_print_error_function): Likewise.
14346         * sched-vis.c (print_exp): Likewise.
14347         (print_value): Likewise.
14348         (print_pattern): Likewise.
14349         (print_insn): Likewise.
14350         (rtl_dump_bb_for_graph): Likewise.
14351         * tree-pretty-print.c (dump_function_declaration): Likewise.
14352         (dump_array_domain): Likewise.
14353         (dump_omp_clause): Likewise.
14354         (dump_location): Likewise.
14355         (dump_generic_node): Likewise.
14356         (print_struct_decl): Likewise.
14357         * diagnostic.c (diagnostic_show_locus): Use pp_space.
14359 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
14361         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
14362         candidate table when replacing a candidate statement.
14363         (replace_rhs_if_not_dup): Likewise.
14364         (replace_one_candidate): Likewise.
14366 2013-08-02  Jan Hubicka  <jh@suse.cz>
14367             Martin Liska  <marxin.liska@gmail.com>
14369         * cgraphunit.c (add_new_function): Fix logic when adding from
14370         late IPA pass.
14371         (assemble_thunk): Rename to ...
14372         (expand_thunk); .. this one; export; get it working with
14373         general functions; make produced gimple valid.
14374         * cgraph.h (expand_thunk): Declare.
14376 2013-08-02  Jan Hubicka  <jh@suse.cz>
14378         * ipa-cp.c (gather_context_independent_values): Use
14379         ipa_get_param_move_cost.
14380         (get_replacement_map): Remove PARAM; move parameter folding
14381         into tree-inline.c
14382         (create_specialized_node): Update.
14383         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
14384         assert that we have gimple body; update move_cost.
14385         (count_formal_params): Assert that we have gimple body.
14386         (ipa_dump_param): New function.
14387         (ipa_alloc_node_params): Break out from ...
14388         (ipa_initialize_node_params): ... here.
14389         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
14390         (ipa_write_node_info): Stream move costs.
14391         (ipa_read_node_info): Read move costs.
14392         (ipa_update_after_lto_read): Do not recompute node params.
14393         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
14394         (ipa_get_param): Check we are not in WPA.
14395         (ipa_get_param_move_cost): New.
14396         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
14397         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
14398         parm numbers to be present.
14400 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
14402         PR rtl-optimization/58048
14403         * lra-constraints.c (process_alt_operands): Don't check asm
14404         operand on register.
14406 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
14408         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
14409         the implied StoreLoad barrier for atomic operations if before.
14411 2013-08-02  Jan Hubicka  <jh@suse.cz>
14412             Martin Liska  <marxin.liska@gmail.com>
14414         * cgraph.c (cgraph_function_body_availability): Do not check
14415         cgraph flags.
14416         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
14417         symtab_node_availability): Declare.
14418         * ipa.c (can_replace_by_local_alias): New.
14419         (function_and_variable_visibility): Use it.
14420         * symtab.c (symtab_for_node_and_aliases,
14421         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
14423 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
14425         PR rtl-optimization/57963
14426         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
14427         (lra_constraints): Use them.
14429 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
14431         * config/arm/types.md (define_attr "type"): Add "load_acq"
14432         and "store_rel".
14433         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
14434         changes.
14435         (cortex_a53_store1): Likewise.
14437 2013-08-01  Jan Hubicka  <jh@suse.cz>
14439         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
14440         partitions are not needed.
14442 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
14444         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
14445         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
14446         MAYBE_NON_Q_CLASS_P where appropriate.
14448 2013-08-01  Jan Hubicka  <jh@suse.cz>
14450         * cgraph.h (release_function_body): Declare.
14451         * tree.c (free_lang_data_in_decl): Free, parameters and return values
14452         of unused delcarations.
14454 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14456         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
14457         RTL form when subtracting a constant.
14459 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14461         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
14462         Generate canonical plus rtx with negated immediate instead of minus
14463         where appropriate.
14464         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
14466 2013-08-01  Jan Hubicka  <jh@suse.cz>
14468         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
14469         (cgraph_release_function_body): Likewise.
14470         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
14471         * cgraph.h (cgrpah_node): Rename abstract_and_needed
14472         to used_as_abstract_origin.
14473         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
14474         symbols used as abstract origins.
14475         * cgraphunit.c (analyze_functions): Update.
14476         * ipa.c (symtab_remove_unreachable_nodes): Recompute
14477         used_as_abstract_origin.
14478         * tree-inline.c (tree_function_versioning): Update
14479         used_as_abstract_origin; be ready for DECL_RESULT and
14480         DECL_ARGUMENTS to be NULL.
14482         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
14483         for abstract functions.
14484         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
14485         real symbols.
14487 2013-08-01  Jan Hubicka  <jh@suse.cz>
14489         * profile.c (compute_value_histograms): Fix thinko.
14491 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
14493         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
14494         aarch-common-protos.h to extra_headers.
14495         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
14496         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
14497         * config/aarch64/t-aarch64 (aarch-common.o): Define.
14499 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
14501         * config/aarch64/aarch64.md (define_attr "type"): Delete.
14502         Include "../arm/types.md".  Define "type" attribute for all patterns.
14503         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
14504         attribute changes.
14506 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
14508         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
14509         to support power8 load fusion.
14510         (fusion_gpr_mem_load): Likewise.
14512         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
14514         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
14515         declarations for power8 load fusion.
14516         (emit_fusion_gpr_load): Likewise.
14518         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
14519         tuning for power8, turn on fusion mode by default.  Turn on sign
14520         extending fusion mode if normal fusion mode is on, and we are at
14521         -O2 or -O3.
14522         (fusion_gpr_load_p): New function, return true if we can fuse an
14523         addis instruction with a dependent load to a GPR.
14524         (emit_fusion_gpr_load): Emit the instructions for power8 load
14525         fusion to GPRs.
14527         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
14528         (VSX load fusion peepholes): New peepholes to fuse together an
14529         addi instruction with a VSX load instruction.
14531         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
14532         peepholes to fuse an addis instruction with a load to a GPR base
14533         register.  If we are supporting sign extending fusions, convert
14534         sign extending loads to zero extending loads and add an explicit
14535         sign extension.
14537 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
14539         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
14540         aarch-common-protos.h to extra_headers.
14541         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
14542         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
14543         (arm_early_store_addr_dep): Likewise.
14544         (arm_no_early_alu_shift_dep): Likewise.
14545         (arm_no_early_alu_shift_value_dep): Likewise.
14546         (arm_no_early_mul_dep): Likewise.
14547         (arm_no_early_store_addr_dep): Likewise.
14548         (arm_mac_accumulator_is_mul_result): Likewise.
14549         (arm_mac_accumulator_is_result): Likewise.
14550         * config/arm/aarch-common.c: ... here.  New file.
14551         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
14552         here to ...
14553         (arm_early_store_addr_dep): Likewise.
14554         (arm_no_early_alu_shift_dep): Likewise.
14555         (arm_no_early_alu_shift_value_dep): Likewise.
14556         (arm_no_early_mul_dep): Likewise.
14557         (arm_no_early_store_addr_dep): Likewise.
14558         (arm_mac_accumulator_is_mul_result): Likewise.
14559         (arm_mac_accumulator_is_result): Likewise.
14560         * config/arm/aarch-common-protos.h: ... here.  New file.
14561         * config/arm/t-arm (aarch-common.o): Define.
14563 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
14565         * config/arm/arm.md: Include new file "types.md".
14566         (define_attr "type"): Move from here to ...
14567         (define_attr "mul32"): Likewise.
14568         (define_attr "mul64"): Likewise.
14569         * config/arm/types.md: ... here.  New file.
14571 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
14573         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
14574         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
14576 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
14578         * gen-pass-instances.awk: Fix offset of substr().
14580 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
14582         * Makefile.in (pass-instances.def): New.
14583         (passes.o): Replace dependency on passes.def with one on
14584         pass-instances.def
14586         * gen-pass-instances.awk: New.
14588         * passes.c (pass_manager::pass_manager): Use pass-instances.def
14589         rather than passes.def, updating local definition of NEXT_PASS
14590         macro to add an extra NUM parameter (currently unused).
14592 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
14594         * Makefile.in (PASS_MANAGER_H): New.
14595         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
14596         (passes.o): Likewise.
14597         (statistics.o): Likewise.
14598         (cgraphunit.o): Likewise.
14599         (context.o): Depend on PASS_MANAGER_H.
14601         * pass_manager.h: New.
14603         * cgraphunit.c (cgraph_add_new_function): Update for moves
14604         of globals to fields of pass_manager.
14605         (analyze_function): Likewise.
14606         (expand_function): Likewise.
14607         (ipa_passes): Likewise.
14608         (compile): Likewise.
14610         * context.c (context::context): New.
14611         * context.h  (context::context): New.
14612         (context::get_passes): New.
14613         (context::passes_): New.
14615         * lto-cgraph.c (input_node): Update for moves of globals to
14616         fields of pass_manager.
14618         * passes.c (all_passes): Remove, in favor of a field of the
14619         same name within the new class pass_manager.
14620         (all_small_ipa_passes): Likewise.
14621         (all_lowering_passes): Likewise.
14622         (all_regular_ipa_passes): Likewise.
14623         (all_late_ipa_passes): Likewise.
14624         (all_lto_gen_passes): Likewise.
14625         (passes_by_id): Likewise.
14626         (passes_by_id_size): Likewise.
14627         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
14628         the new class pass_manager.
14629         (set_pass_for_id): Convert to...
14630         (pass_manager::set_pass_for_id): ...method.
14631         (get_pass_for_id): Convert to...
14632         (pass_manager::get_pass_for_id): ...method.
14633         (register_one_dump_file): Move body of implementation into...
14634         (pass_manager::register_one_dump_file): ...here.
14635         (register_dump_files_1): Convert to...
14636         (pass_manager::register_dump_files_1): ...method.
14637         (register_dump_files): Convert to...
14638         (pass_manager::register_dump_files): ...method.
14639         (create_pass_tab): Update for moves of globals to fields of
14640         pass_manager.
14641         (dump_passes): Move body of implementation into...
14642         (pass_manager::dump_passes): ...here.
14643         (register_pass): Move body of implementation into...
14644         (pass_manager::register_pass): ...here.
14645         (init_optimization_passes): Convert into...
14646         (pass_manager::pass_manager): ...constructor for new
14647         pass_manager class, and initialize the pass_lists array.
14648         (check_profile_consistency): Update for moves of globals to
14649         fields of pass_manager.
14650         (dump_profile_report): Move body of implementation into...
14651         (pass_manager::dump_profile_report): ...here.
14652         (ipa_write_summaries_1): Update for moves of pass lists from
14653         being globals to fields of pass_manager.
14654         (ipa_write_optimization_summaries): Likewise.
14655         (ipa_read_summaries):  Likewise.
14656         (ipa_read_optimization_summaries): Likewise.
14657         (execute_all_ipa_stmt_fixups): Likewise.
14659         * statistics.c (statistics_fini): Update for moves of globals to
14660         fields of pass_manager.
14662         * toplev.c (general_init): Replace call to
14663         init_optimization_passes with construction of the pass_manager
14664         instance.
14666         * tree-pass.h (all_passes): Remove, in favor of a field of the
14667         same name within the new class pass_manager.
14668         (all_small_ipa_passes): Likewise.
14669         (all_lowering_passes): Likewise.
14670         (all_regular_ipa_passes): Likewise.
14671         (all_lto_gen_passes): Likewise.
14672         (all_late_ipa_passes): Likewise.
14673         (passes_by_id): Likewise.
14674         (passes_by_id_size): Likewise.
14675         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
14676         the new class pass_manager.
14677         (get_pass_for_id): Remove.
14679 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
14681         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
14682         configs.
14684 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
14686         * arm.md (mulhi3): New expand pattern.
14688 2013-07-30  Jan Hubicka  <jh@suse.cz>
14689             Martin Liska  <marxin.liska@gmail.com>
14691         * profile.c (compute_value_histograms): Do not ICE when
14692         there is mismatch only on some counters.
14694 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
14696         PR rtl-optimization/57637
14697         * function.c (move_insn_for_shrink_wrap): Also check the
14698         GEN set of the LIVE problem for the liveness analysis
14699         if it exists, otherwise give up.
14701 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
14703         PR tree-optimization/57993
14704         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
14705         replaced statement in the candidate table.
14706         (phi_add_costs): Return infinite cost when the hidden basis does
14707         not dominate all phis on which the candidate is dependent.
14708         (replace_one_candidate): Record replaced statement in the
14709         candidate table.
14711 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
14713         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
14714         (ashlv2si3): New expander.
14715         (*ashlv2si3_i): New define_insn_and_split.
14716         * predicates.md (float_operation): Allow patterns with three
14717         basic sub-patterns.
14719         PR rtl-optimization/58021
14720         * mode-switching.c (create_pre_exit): Always split off preceding
14721         insns if we are not at the basic block head.
14723 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
14725         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
14726         (UCLIBC_DYNAMIC_LINKER): New macro.
14727         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
14728         `-mnan=2008'.
14729         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
14730         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
14731         `-mnan=2008'.
14732         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
14733         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
14734         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
14735         for SF and DF modes.  Use ieee_quad_format for TF mode.
14736         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
14737         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
14738         (mips_option_override): Handle `-mnan=legacy'.
14739         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
14740         `-mabs=2008' and `-mnan=2008'.
14741         (OPTION_DEFAULT_SPECS): Add "nan" default.
14742         (ASM_SPEC): Handle `-mnan='.
14743         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
14744         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
14745         comment accordingly.
14746         (neg<mode>2): Likewise.
14747         * config/mips/mips.opt (mabs, mnan): New options.
14748         * doc/install.texi (Configuration): Document `--with-nan=' option.
14749         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
14750         `-mnan=' options.
14751         (MIPS Options): Document them.
14752         * config.gcc <mips*-*-*>: Handle `--with-nan='.
14753         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
14754         * configure: Regenerate.
14755         * config.in: Regenerate.
14757 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
14759         * config/i386/i386.md (float post-reload splitters): Do not check
14760         for subregs of SSE registers.
14762 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
14763             H.J. Lu  <hongjiu.lu@intel.com>
14765         PR target/57954
14766         PR target/57988
14767         * config/i386/i386.md (post-reload splitter
14768         to avoid partial SSE reg dependency stalls): New pattern.
14770 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
14772         * config/s390/s390.md ("movcc"): Swap load and store instructions.
14774 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
14776         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
14777         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
14779 2013-07-26  Cary Coutant  <ccoutant@google.com>
14781         * dwarf2out.c (die_checksum_ordered): Don't include template
14782         instantiations in signature.
14783         (is_template_parameter): New function.
14784         (is_template_instantiation): New function.
14785         (generate_skeleton_bottom_up): Don't include template instantiations
14786         in type unit DIE.
14787         (generate_skeleton): Likewise.
14788         (break_out_comdat_types): Move recursive call to break out nested
14789         types earlier.
14790         (prune_unused_types_mark_generic_parms_dies): Call
14791         is_template_parameter.
14793 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
14795         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
14796         uses vector registers.
14797         * config/aarch64/iterators.md: Add attributes rtn and vas.
14799 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14800             Richard Earnshaw  <richard.earnshaw@arm.com>
14802         * combine.c (simplify_comparison): Re-canonicalize operands
14803         where appropriate.
14804         * config/arm/arm.md (movcond_addsi): New splitter.
14806 2013-07-25  Sterling Augustine  <saugustine@google.com>
14808         * dwarf2out.c (size_of_pubnames): Move code to...
14809         (include_pubname_in_output): ...here.  New.
14810         (want_pubnames): Rearrange.
14811         (output_pubnames): Call include_pubname_in_output.  Move assertion.
14813 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
14815         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
14817 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
14819         PR target/38836
14820         * doc/extend.texi: Remove obsolete builtins. Fix
14821         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
14823 2013-07-25  Jan Hubicka  <jh@suse.cz>
14825         * cgraph.c (release_function_body): Break out from ...
14826         (cgraph_release_function_body): ... this one; also release DECL_RESULT
14827         and DECL_ARGUMENTS.
14828         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
14829         old_tree in the map.
14830         (create_specialized_node): Update.
14831         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
14832         into index.
14833         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
14834         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
14835         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
14836         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
14837         DECL_RESULT.
14839 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14841         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
14842         addsi3_carryin_alt2_<optab>): Correct output template.
14844 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14846         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
14847         Adjust for arm_restrict_it.
14848         Remove trailing whitespace.
14850 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
14852         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
14853         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
14855         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
14857 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
14859         PR rtl-optimization/57960
14860         * lra-constraints.c (process_alt_operands): Use the right mode
14861         when checking strict_low.
14863 2013-07-25  Jan Hubicka  <jh@suse.cz>
14865         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
14866         * cgraph.c (cgraph_remove_node): Do not release function body
14867         when in cgraph streaming.
14868         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
14869         in other partitions are not considered reachable; fix handling of
14870         clones.
14872 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
14874         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
14876 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
14878         PR target/19599
14879         PR target/57731
14880         PR target/57837
14881         * config/arm/arm.md ("*sibcall_insn): Replace use of
14882         Ss with US. Adjust output for v5 and v4t.
14883         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
14885         * config/arm/constraints.md ("Ss"): Rename to US.
14887 2013-07-25  Terry Guo  <terry.guo@arm.com>
14889         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
14890         shift_add/shift_sub0/shift_sub1 RTXs.
14892 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
14893             Anton Blanchard  <anton@au1.ibm.com>
14895         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
14896         (altivec_vpks<VI_char>ss): Likewise.
14897         (altivec_vpks<VI_char>us): Likewise.
14898         (altivec_vpku<VI_char>us): Likewise.
14899         (altivec_vpku<VI_char>um): Likewise.
14901 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
14903         Introduce context class.
14905         * Makefile.in (CONTEXT_H): New.
14906         (OBJS): Add context.o.
14907         (toplev.o): Add CONTEXT_H to dependencies.
14908         (context.o): New.
14910         * toplev.c (general_init): Create the singleton gcc::context instance.
14912         * context.c: New.
14914         * context.h: New.
14916 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
14918         PR rtl-optimization/57968
14919         * mode-switching.c (create_pre_exit): Allow instructions that
14920         don't set a return register to need a non-exit mode.
14922 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
14923             Anton Blanchard  <anton@au1.ibm.com>
14925         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
14926         operands to vperm for little endian.
14927         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
14928         of lvsl to create the control mask for a vperm for little endian.
14930 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
14931             Anton Blanchard  <anton@au1.ibm.com>
14933         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
14934         two operands for little-endian.
14936 2013-07-23  Steve Ellcey  <sellcey@mips.com>
14938         * config/mips/mips.c (mips_case_values_threshold): New.
14939         (TARGET_CASE_VALUES_THRESHOLD): Define.
14941 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
14942             Anton Blanchard  <anton@au1.ibm.com>
14944         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
14945         selection of field for vector splat in little endian mode.
14947 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
14949         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
14950         expanders to rs6000.md.
14951         (ior<mode>3): Likewise.
14952         (and<mode>3): Likewise.
14953         (one_cmpl<mode>2): Likewise.
14954         (nor<mode>3): Likewise.
14955         (andc<mode>3): Likewise.
14956         (eqv<mode>3): Likewise.
14957         (nand<mode>3): Likewise.
14958         (orc<mode>3): Likewise.
14960         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
14961         declaration.
14963         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
14964         to split multi-word logical operations.
14965         (rs6000_split_logical_di): Likewise.
14966         (rs6000_split_logical): Likewise.
14968         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
14969         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
14970         and allow TImode operations in 32-bit.
14971         (vsx_and<mode>3_64bit): Likewise.
14972         (vsx_ior<mode>3_32bit): Likewise.
14973         (vsx_ior<mode>3_64bit): Likewise.
14974         (vsx_xor<mode>3_32bit): Likewise.
14975         (vsx_xor<mode>3_64bit): Likewise.
14976         (vsx_one_cmpl<mode>2_32bit): Likewise.
14977         (vsx_one_cmpl<mode>2_64bit): Likewise.
14978         (vsx_nor<mode>3_32bit): Likewise.
14979         (vsx_nor<mode>3_64bit): Likewise.
14980         (vsx_andc<mode>3_32bit): Likewise.
14981         (vsx_andc<mode>3_64bit): Likewise.
14982         (vsx_eqv<mode>3_32bit): Likewise.
14983         (vsx_eqv<mode>3_64bit): Likewise.
14984         (vsx_nand<mode>3_32bit): Likewise.
14985         (vsx_nand<mode>3_64bit): Likewise.
14986         (vsx_orc<mode>3_32bit): Likewise.
14987         (vsx_orc<mode>3_64bit): Likewise.
14989         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
14990         logical types in GPRs.
14992         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
14993         logical insns to rs6000.md, and allow TImode operations in
14994         32-bit.
14995         (altivec_ior<mode>3): Likewise.
14996         (altivec_xor<mode>3): Likewise.
14997         (altivec_one_cmpl<mode>2): Likewise.
14998         (altivec_nor<mode>3): Likewise.
14999         (altivec_andc<mode>3): Likewise.
15001         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
15002         attributes for moving the 128-bit logical operations into
15003         rs6000.md.
15004         (BOOL_REGS_OUTPUT): Likewise.
15005         (BOOL_REGS_OP1): Likewise.
15006         (BOOL_REGS_OP2): Likewise.
15007         (BOOL_REGS_UNARY): Likewise.
15008         (BOOL_REGS_AND_CR0): Likewise.
15009         (one_cmpl<mode>2): Add support for DI logical operations on
15010         32-bit, splitting the operations to 32-bit.
15011         (anddi3): Likewise.
15012         (iordi3): Likewise.
15013         (xordi3): Likewise.
15014         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
15015         changes to combine the 32/64-bit code, allow logical operations on
15016         TI mode in 32-bit, and to use similar match_operator patterns like
15017         scalar mode uses.  Combine the Altivec and VSX code for logical
15018         operations, and move it here.
15019         (ior<mode>3, 128-bit types): Likewise.
15020         (xor<mode>3, 128-bit types): Likewise.
15021         (one_cmpl<mode>3, 128-bit types): Likewise.
15022         (nor<mode>3, 128-bit types): Likewise.
15023         (andc<mode>3, 128-bit types): Likewise.
15024         (eqv<mode>3, 128-bit types): Likewise.
15025         (nand<mode>3, 128-bit types): Likewise.
15026         (orc<mode>3, 128-bit types): Likewise.
15027         (and<mode>3_internal): Likewise.
15028         (bool<mode>3_internal): Likewise.
15029         (boolc<mode>3_internal1): Likewise.
15030         (boolc<mode>3_internal2): Likewise.
15031         (boolcc<mode>3_internal1): Likewise.
15032         (boolcc<mode>3_internal2): Likewise.
15033         (eqv<mode>3_internal1): Likewise.
15034         (eqv<mode>3_internal2): Likewise.
15035         (one_cmpl1<mode>3_internal): Likewise.
15037 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
15039         * config/microblaze/microblaze.c (microblaze_expand_prologue):
15040         Rename flag_stack_usage to flag_stack_usage_info.
15042 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
15044         * config/microblaze/sync.md: New file.
15045         * config/microblaze/microblaze.md: Include sync.md
15046         * config/microblaze/microblaze.c: Add print_operand 'y'.
15047         * config/microblaze/constraints.md: Add memory_contraint
15048         'Q' which is a single register.
15050 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
15052         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
15054 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
15056         * reload.c (find_reloads): Exit loop once we find this operand
15057         cannot be reloaded somehow for this alternative.
15059         * reload.c (find_reloads): Exit loop once we find a hard register.
15061         * rtlanal.c (computed_jump_p): Exit loop once we find label
15062         reference is used.
15064         * i386.c (ix86_pad_returns): Exit loop after setting replace.
15066         * cfgloopmanip.c (remove_path): Exit loop after setting
15067         irred_invalidated.
15069         * gensupport.c (subst_dup): Avoid loop if code is not
15070         MATCH_DUP nor MATCH_OP_DUP.
15072 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
15074         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
15076 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
15078         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
15079         true for SP_REGNUM if mode == ptr_mode.
15080         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
15081         with value R0_REGNUM + 31.
15083 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
15085         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
15086         pad pointer-typed argument downward.
15088 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
15090         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
15091         and __ILP32__ when the ILP32 model is in use.
15093 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
15095         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
15096         (aarch64_load_symref_appropriately): In the case of
15097         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
15098         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
15099         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
15100         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
15101         if 'mode' doesn't equal to 'ptr_mode'.
15102         (aarch64_output_mi_thunk): Add an assertion on the alignment of
15103         'vcall_offset'; change to call aarch64_emit_move differently depending
15104         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
15105         to calculate the upper bound of 'vcall_offset'.
15106         (aarch64_cannot_force_const_mem): Change to also return true if
15107         mode != ptr_mode.
15108         (aarch64_legitimize_reload_address): In the case of large
15109         displacements, add new local variable 'xmode' and an assertion
15110         based on it; change to use 'xmode' to generate the new rtx and
15111         reload.
15112         (aarch64_asm_trampoline_template): Change to generate the template
15113         differently depending on TARGET_ILP32 or not; change to use
15114         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
15115         (aarch64_trampoline_size): Removed.
15116         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
15117         and replace immediate literals with it.  Change to use 'ptr_mode'
15118         instead of 'DImode' and call convert_memory_address if the mode
15119         of 'fnaddr' doesn't equal to 'ptr_mode'.
15120         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
15121         to output symbol.
15122         (aarch64_elf_asm_destructor): Likewise.
15123         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
15124         on TARGET_ILP32 instead of aarch64_trampoline_size.
15125         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
15126         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
15127         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
15128         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
15129         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
15130         (storewb_pair<GPI:mode>_<P:mode>): ... this.
15131         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
15132         depending on the value of 'mode'.
15133         (add_losym_<mode>): New.
15134         (ldr_got_small_<mode>): New, based on ldr_got_small.
15135         (ldr_got_small): Remove.
15136         (ldr_got_small_sidi): New.
15137         * config/aarch64/iterators.md (P): New.
15138         (PTR): Change to 'ptr_mode' in the condition.
15140 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
15142         * config.gcc (aarch64*-*-*): Support --with-abi.
15143         (aarch64*-*-elf): Support --with-multilib-list.
15144         (aarch64*-*-linux*): Likewise.
15145         (supported_defaults): Add abi to aarch64*-*-*.
15146         * configure.ac: Mention AArch64 for --with-multilib-list.
15147         * configure: Re-generated.
15148         * config/aarch64/biarchilp32.h: New file.
15149         * config/aarch64/biarchlp64.h: New file.
15150         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
15151         (ABI_SPEC): Ditto.
15152         (MULTILIB_DEFAULTS): Ditto.
15153         (DRIVER_SELF_SPECS): Ditto.
15154         (ASM_SPEC): Update to also substitute -mabi.
15155         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
15156         file whose name depends on -mabi= and -mbig-endian.
15157         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
15158         TARGET_ILP32.
15159         (POINTER_SIZE): New define.
15160         (POINTERS_EXTEND_UNSIGNED): Ditto.
15161         (enum aarch64_abi_type): New enumeration tag.
15162         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
15163         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
15164         (TARGET_ILP32): New define.
15165         * config/aarch64/aarch64.opt (mabi): New.
15166         (aarch64_abi): New.
15167         (ilp32, lp64): New values for -mabi.
15168         * config/aarch64/t-aarch64 (comma): New define.
15169         (MULTILIB_OPTIONS): Ditto.
15170         (MULTILIB_DIRNAMES): Ditto.
15171         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
15172         * doc/invoke.texi: Document -mabi for AArch64.
15174 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
15176         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
15178 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
15179             Anton Blanchard  <anton@au1.ibm.com>
15181         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
15182         endianness when selecting field to splat.
15184 2013-07-22  Eric Christopher  <echristo@gmail.com>
15186         * dwarf2out.c (die_odr_checksum): New function to use
15187         CHECKSUM_ macros and ULEB128 for DIE tag.
15188         (generate_type_signature): Use.
15190 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
15192         * config.gcc (sparc*-*-*): Accept leon3 processor.
15193         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
15194         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
15195         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
15196         * config/sparc/sparc.opt (enum processor_type): Add leon3.
15197         (mfix-ut699): Adjust comment.
15198         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
15199         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
15200         (CPP_CPU_SPEC): Likewise.
15201         (ASM_CPU_SPEC): Likewise.
15202         * config/sparc/sparc.c (leon3_cost): New constant.
15203         (sparc_option_override): Add leon3 support.
15204         (mem_ref): New function.
15205         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
15206         (sparc_do_work_around_errata): Look into the instruction in the delay
15207         slot and adjust accordingly.  Add fix for the data cache nullify issues
15208         of the UT699.  Change insertion position for the NOP.
15209         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
15210         (leon3_load): New reservation.
15211         (leon_store): Bump latency to 2.
15212         (grfpu): New automaton.
15213         (grfpu_alu): New unit.
15214         (grfpu_ds): Likewise.
15215         (leon_fp_alu): Adjust.
15216         (leon_fp_mult): Delete.
15217         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
15218         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
15219         * config/sparc/sparc.md (cpu): Add leon3.
15220         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
15221         (swapsi): Likewise.
15222         (atomic_test_and_set): Likewise.
15223         (ldstub): Likewise.
15225 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
15227         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
15228         default for R5900 targets.
15229         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
15230         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
15231         * config/mips/mips.c (mips_option_override): Report an error for
15232         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
15233         for -march=r5900 -mhard-float.
15235 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
15237         * df-problems.c (can_move_insns_across): Exit loop once we
15238         find a non-fixed, non-global register.
15240         * ipa-pure-const.c (propagate_nothrow): Exit loop after
15241         setting can_throw.
15243         * omega.c (omega_eliminate_red): Break after setting red_found.
15244         (omega_problem_has_red_equations): Similarly after setting found.
15245         (omega_query_variable): Similarly after setting coupled.
15247 2013-07-22  Marek Polacek  <polacek@redhat.com>
15249         * gimplify.c: Don't include gimple.h twice.
15251 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15253         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
15254         instead of TARGET_THUMB1.
15255         (Pz): New constraint.
15256         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
15257         encodings.
15258         (compare_negsi_si): Likewise.
15259         (compare_addsi2_op0): Likewise.
15260         (compare_addsi2_op1): Likewise.
15261         (addsi3_carryin_<optab>): Likewise.
15262         (addsi3_carryin_alt2_<optab>): Likewise.
15263         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
15264         for arm_restrict_it.
15265         (subsi3_carryin): Likewise.
15266         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
15267         (minmax_arithsi): Disable for arm_restrict_it.
15268         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
15269         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
15270         (satsi_<SAT:code>_shift): Likewise.
15271         (arm_shiftsi3): Add alternative for 16-bit encoding.
15272         (arm32_movhf): Disable for arm_restrict_it.
15273         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
15274         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
15276 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
15278         * config/arm/arm.md (attribute "insn"): Delete.
15279         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
15280         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
15281         (not_shiftsi): Update for attribute change.
15282         (not_shiftsi_compare0): Likewise.
15283         (not_shiftsi_compare0_scratch): Likewise.
15284         (arm_one_cmplsi2): Likewise.
15285         (thumb1_one_cmplsi2): Likewise.
15286         (notsi_compare0): Likewise.
15287         (notsi_compare0_scratch): Likewise.
15288         (thumb1_movdi_insn): Likewise.
15289         (arm_movsi_insn): Likewise.
15290         (movhi_insn_arch4): Likewise.
15291         (movhi_bytes): Likewise.
15292         (arm_movqi_insn): Likewise.
15293         (thumb1_movqi_insn): Likewise.
15294         (arm32_movhf): Likewise.
15295         (thumb1_movhf): Likewise.
15296         (arm_movsf_soft_insn): Likewise.
15297         (thumb1_movsf_insn): Likewise.
15298         (thumb_movdf_insn): Likewise.
15299         (movsicc_insn): Likewise.
15300         (movsfcc_soft_insn): Likewise.
15301         (and_scc): Likewise.
15302         (cond_move): Likewise.
15303         (if_move_not): Likewise.
15304         (if_not_move): Likewise.
15305         (if_shift_move): Likewise.
15306         (if_move_shift): Likewise.
15307         (if_shift_shift): Likewise.
15308         (if_not_arith): Likewise.
15309         (if_arith_not): Likewise.
15310         (cond_move_not): Likewise.
15311         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
15312         (neon_mov<mode>): Likewise.
15313         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
15314         (thumb2_movsi_vfp): Likewise.
15315         (movsf_vfp): Likewise.
15316         (thumb2_movsf_vfp): Likewise.
15317         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
15318         change.
15319         (cortexa7_older_only): Likewise.
15320         (cortexa7_younger): Likewise.
15321         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
15322         (1020alu_shift_op): Likewise.
15323         (1020alu_shift_reg_op): Likewise.
15324         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
15325         (alu_shift_op): Likewise.
15326         (alu_shift_reg_op): Likewise.
15327         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
15328         (11_alu_shift_op): Likewise.
15329         (11_alu_shift_reg_op): Likewise.
15330         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
15331         (9_alu_shift_reg_op): Likewise.
15332         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
15333         change.
15334         (cortex_a15_alu_shift): Likewise.
15335         (cortex_a15_alu_shift_reg): Likewise.
15336         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
15337         (cortex_a5_alu_shift): Likewise.
15338         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
15339         change.
15340         (cortex_a53_alu_shift): Likewise.
15341         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
15342         change.
15343         (cortex_a7_alu_reg): Likewise.
15344         (cortex_a7_alu_shift): Likewise.
15345         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
15346         (cortex_a8_alu_shift): Likewise.
15347         (cortex_a8_alu_shift_reg): Likewise.
15348         (cortex_a8_mov): Likewise.
15349         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
15350         (cortex_a9_dp_shift): Likewise.
15351         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
15352         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
15353         (cortex_r4_mov): Likewise.
15354         (cortex_r4_alu_shift): Likewise.
15355         (cortex_r4_alu_shift_reg): Likewise.
15356         * config/arm/fa526.md (526_alu_op): Update for attribute change.
15357         (526_alu_shift_op): Likewise.
15358         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
15359         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
15360         (626te_alu_shift_op): Likewise.
15361         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
15362         (726te_alu_op): Likewise.
15363         (726te_alu_shift_op): Likewise.
15364         (726te_alu_shift_reg_op): Likewise.
15365         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
15366         (mp626_alu_shift_op): Likewise.
15367         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
15368         (pj4_alu_e1_conds): Likewise.
15369         (pj4_alu): Likewise.
15370         (pj4_alu_conds): Likewise.
15371         (pj4_shift): Likewise.
15372         (pj4_shift_conds): Likewise.
15373         (pj4_alu_shift): Likewise.
15374         (pj4_alu_shift_conds): Likewise.
15376 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15378         * config/arm/predicates.md (shiftable_operator_strict_it):
15379         New predicate.
15380         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
15381         Disable cond_exec version for arm_restrict_it.
15382         (thumb2_smaxsi3): Convert to generate cond_exec.
15383         (thumb2_sminsi3): Likewise.
15384         (thumb32_umaxsi3): Likewise.
15385         (thumb2_uminsi3): Likewise.
15386         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
15387         (thumb2_neg_abssi2): Likewise.
15388         (thumb2_mov_scc): Add alternative for 16-bit encoding.
15389         (thumb2_movsicc_insn): Adjust alternatives.
15390         (thumb2_mov_negscc): Disable for arm_restrict_it.
15391         (thumb2_mov_negscc_strict_it): New pattern.
15392         (thumb2_mov_notscc_strict_it): New pattern.
15393         (thumb2_mov_notscc): Disable for arm_restrict_it.
15394         (thumb2_ior_scc): Likewise.
15395         (thumb2_ior_scc_strict_it): New pattern.
15396         (thumb2_cond_move): Adjust for arm_restrict_it.
15397         (thumb2_cond_arith): Disable for arm_restrict_it.
15398         (thumb2_cond_arith_strict_it): New pattern.
15399         (thumb2_cond_sub): Adjust for arm_restrict_it.
15400         (thumb2_movcond): Likewise.
15401         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
15402         (thumb2_zero_extendhisi2_v6): Likewise.
15403         (thumb2_zero_extendqisi2_v6): Likewise.
15404         (orsi_notsi_si): Likewise.
15405         (orsi_not_shiftsi_si): Likewise.
15407 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
15409         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
15410         instruction sequence is 1 byte shorter.
15412 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
15414         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
15415         it is not needed after split.
15417 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
15419         PR target/51784
15420         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
15421         second label for nonlocal goto receivers. Don't output pic base labels
15422         unless we're producing PIC; mark that action unreachable().
15423         (ix86_save_reg): If the function contains a nonlocal label, save the
15424         PIC base reg.
15425         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
15426         * config/darwin.c (emitted_pic_label_num): New GTY.
15427         (update_pic_label_number_if_needed): New.
15428         (machopic_output_function_base_name): Adjust for nonlocal receiver
15429         case.
15430         (machopic_should_output_picbase_label): New.
15431         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
15432         (nonlocal_goto_receiver): New insn and split.
15434 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
15436         * config/aarch64/aarch64-builtins.c
15437         (aarch64_fold_builtin): Fold abs in all modes.
15438         * config/aarch64/aarch64-simd-builtins.def
15439         (abs): Enable for all modes.
15440         * config/aarch64/arm_neon.h
15441         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
15442         (vabs_f64): Add missing intrinsic.
15444 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
15446         * config/aarch64/arm_neon.h (vabs_s64): New function
15448 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
15450         PR target/57516
15451         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
15452         * config/avr/avr.md (adjust_len): Add `round'.
15453         * config/avr/avr-protos.h (avr_out_round): New prototype.
15454         (avr_out_plus): Add `out_label' argument.
15455         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
15456         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
15457         Handle the case where `insn' is just a pattern.
15458         (avr_out_bitop): Handle the case where `insn' is just a pattern.
15459         (avr_out_round): New function.
15460         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
15462 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
15464         * config/microblaze/microblaze.c (microblaze_expand_prologue):
15465         Add check for flag_stack_usage to handle -fstack-usage support
15467 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
15469         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
15470         interaction for new Power8 flags and VSX.
15472 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
15474         PR middle-end/57698
15475         * tree-inline.c (expand_call_inline): Emit errors during
15476         early_inlining only if optimization is not turned on.
15478 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
15480         * passes.def: New.
15482         * passes.c (init_optimization_passes): Move the construction of
15483         the pass hierarchy into a new passes.def file.
15485         * Makefile.in (passes.o): Add dependency on passes.def.
15487 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
15489         * passes.c (init_optimization_passes): Introduce macros for
15490         constructing the tree of passes (INSERT_PASSES_AFTER,
15491         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
15492         TERMINATE_PASS_LIST).
15494 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
15495             Wei Mi  <wmi@google.com>
15497         PR rtl-optimization/57878
15498         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
15499         top.
15500         (reload_pseudo_compare_func): Check nregs first for reload
15501         pseudos.
15503 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
15505         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
15507 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
15509         * read-rtl.c (validate_const_int): Once an invalid character is
15510         seen, quit the loop.
15512         * gengtype.c (write_roots): Similarly once we find the "deletable"
15513         or "if_marked" option.
15515 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
15517         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
15518         "xtab" and "sat".  Move value "clz" from here to ...
15519         (attriubte "type"): ... here.
15520         (satsi_<SAT:code>): Delete "insn" attribute.
15521         (satsi_<SAT:code>_shift): Likewise.
15522         (arm_zero_extendqisi2addsi): Likewise.
15523         (arm_extendqisi2addsi): Likewise.
15524         (clzsi2): Update for attribute changes.
15525         (rbitsi2): Likewise.
15526         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
15527         attribute.
15528         (arm_usatsihi): Likewise.
15529         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
15531 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
15533         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
15534         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
15535         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
15536         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
15537         in alphabetical order.
15538         (attribute "core_cycles"): Update for attribute changes.
15539         (arm_addsi3): Likewise.
15540         (addsi3_compare0): Likewise.
15541         (addsi3_compare0_scratch): Likewise.
15542         (addsi3_compare_op1): Likewise.
15543         (addsi3_compare_op2): Likewise.
15544         (compare_addsi2_op0): Likewise.
15545         (compare_addsi2_op1): Likewise.
15546         (addsi3_carryin_shift_<optab>): Likewise.
15547         (subsi3_carryin_shift): Likewise.
15548         (rsbsi3_carryin_shift): Likewise.
15549         (arm_subsi3_insn): Likewise.
15550         (subsi3_compare0): Likewise.
15551         (subsi3_compare): Likewise.
15552         (arm_andsi3_insn): Likewise.
15553         (thumb1_andsi3_insn): Likewise.
15554         (andsi3_compare0): Likewise.
15555         (andsi3_compare0_scratch): Likewise.
15556         (zeroextractsi_compare0_scratch
15557         (andsi_not_shiftsi_si): Likewise.
15558         (iorsi3_insn): Likewise.
15559         (iorsi3_compare0): Likewise.
15560         (iorsi3_compare0_scratch): Likewise.
15561         (arm_xorsi3): Likewise.
15562         (thumb1_xorsi3_insn): Likewise.
15563         (xorsi3_compare0): Likewise.
15564         (xorsi3_compare0_scratch): Likewise.
15565         (satsi_<SAT:code>_shift): Likewise.
15566         (rrx): Likewise.
15567         (arm_shiftsi3): Likewise.
15568         (shiftsi3_compare0): Likewise.
15569         (not_shiftsi): Likewise.
15570         (not_shiftsi_compare0): Likewise.
15571         (not_shiftsi_compare0_scratch): Likewise.
15572         (arm_one_cmplsi2): Likewise.
15573         (thumb_one_complsi2): Likewise.
15574         (notsi_compare0): Likewise.
15575         (notsi_compare0_scratch): Likewise.
15576         (thumb1_zero_extendhisi2): Likewise.
15577         (arm_zero_extendhisi2): Likewise.
15578         (arm_zero_extendhisi2_v6): Likewise.
15579         (arm_zero_extendhisi2addsi): Likewise.
15580         (thumb1_zero_extendqisi2): Likewise.
15581         (thumb1_zero_extendqisi2_v6): Likewise.
15582         (arm_zero_extendqisi2): Likewise.
15583         (arm_zero_extendqisi2_v6): Likewise.
15584         (arm_zero_extendqisi2addsi): Likewise.
15585         (thumb1_extendhisi2): Likewise.
15586         (arm_extendhisi2): Likewise.
15587         (arm_extendhisi2_v6): Likewise.
15588         (arm_extendqisi): Likewise.
15589         (arm_extendqisi_v6): Likewise.
15590         (arm_extendqisi2addsi): Likewise.
15591         (thumb1_extendqisi2): Likewise.
15592         (thumb1_movdi_insn): Likewise.
15593         (arm_movsi_insn): Likewise.
15594         (movsi_compare0): Likewise.
15595         (movhi_insn_arch4): Likewise.
15596         (movhi_bytes): Likewise.
15597         (arm_movqi_insn): Likewise.
15598         (thumb1_movqi_insn): Likewise.
15599         (arm32_movhf): Likewise.
15600         (thumb1_movhf): Likewise.
15601         (arm_movsf_soft_insn): Likewise.
15602         (thumb1_movsf_insn): Likewise.
15603         (movdf_soft_insn): Likewise.
15604         (thumb_movdf_insn): Likewise.
15605         (arm_cmpsi_insn): Likewise.
15606         (cmpsi_shiftsi): Likewise.
15607         (cmpsi_shiftsi_swp): Likewise.
15608         (arm_cmpsi_negshiftsi_si): Likewise.
15609         (movsicc_insn): Likewise.
15610         (movsfcc_soft_insn): Likewise.
15611         (arith_shiftsi): Likewise.
15612         (arith_shiftsi_compare0
15613         (arith_shiftsi_compare0_scratch
15614         (sub_shiftsi): Likewise.
15615         (sub_shiftsi_compare0
15616         (sub_shiftsi_compare0_scratch
15617         (and_scc): Likewise.
15618         (cond_move): Likewise.
15619         (if_plus_move): Likewise.
15620         (if_move_plus): Likewise.
15621         (if_move_not): Likewise.
15622         (if_not_move): Likewise.
15623         (if_shift_move): Likewise.
15624         (if_move_shift): Likewise.
15625         (if_shift_shift): Likewise.
15626         (if_not_arith): Likewise.
15627         (if_arith_not): Likewise.
15628         (cond_move_not): Likewise.
15629         (thumb1_ashlsi3): Set type attribute.
15630         (thumb1_ashrsi3): Likewise.
15631         (thumb1_lshrsi3): Likewise.
15632         (thumb1_rotrsi3): Likewise.
15633         (shiftsi3_compare0_scratch): Likewise.
15634         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
15635         (neon_mov<mode>): Likewise.
15636         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
15637         attribute changes.
15638         (thumb2_movsi_insn): Likewise.
15639         (thumb2_cmpsi_neg_shiftsi): Likewise.
15640         (thumb2_extendqisi_v6): Likewise.
15641         (thumb2_zero_extendhisi2_v6): Likewise.
15642         (thumb2_zero_extendqisi2_v6): Likewise.
15643         (thumb2_shiftsi3_short): Likewise.
15644         (thumb2_addsi3_compare0_scratch): Likewise.
15645         (orsi_not_shiftsi_si): Likewise.
15646         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
15647         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
15648         changes.
15649         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
15650         (1020alu_shift_op): Likewise.
15651         (1020alu_shift_reg_op): Likewise.
15652         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
15653         (alu_shift_op): Likewise.
15654         (alu_shift_reg_op): Likewise.
15655         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
15656         (11_alu_shift_op): Likewise.
15657         (11_alu_shift_reg_op): Likewise.
15658         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
15659         (9_alu_shift_reg_op): Likewise.
15660         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
15661         changes.
15662         (cortex_a15_alu_shift): Likewise.
15663         (cortex_a15_alu_shift_reg): Likewise.
15664         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
15665         changes.
15666         (cortex_a5_alu_shift): Likewise.
15667         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
15668         changes.
15669         (cortex_a53_alu_shift): Likewise.
15670         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
15671         changes.
15672         (cortex_a7_alu_reg): Likewise.
15673         (cortex_a7_alu_shift): Likewise.
15674         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
15675         changes.
15676         (cortex_a8_alu_shift): Likewise.
15677         (cortex_a8_alu_shift_reg): Likewise.
15678         (cortex_a8_mov): Likewise.
15679         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
15680         (cortex_a9_dp_shift): Likewise.
15681         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
15682         changes.
15683         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
15684         changes.
15685         (cortex_r4_mov): Likewise.
15686         (cortex_r4_alu_shift): Likewise.
15687         (cortex_r4_alu_shift_reg): Likewise.
15688         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
15689         (526_alu_shift_op): Likewise.
15690         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
15691         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
15692         (626te_alu_shift_op): Likewise.
15693         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
15694         (726te_alu_op): Likewise.
15695         (726te_alu_shift_op): Likewise.
15696         (726te_alu_shift_reg_op): Likewise.
15697         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
15698         (mp626_alu_shift_op): Likewise.
15699         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
15700         (pj4_alu_e1_conds): Likewise.
15701         (pj4_alu): Likewise.
15702         (pj4_alu_conds): Likewise.
15703         (pj4_shift): Likewise.
15704         (pj4_shift_conds): Likewise.
15705         (pj4_alu_shift): Likewise.
15706         (pj4_alu_shift_conds): Likewise.
15707         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
15708         changes.
15709         (cortexa7_older_only): Likewise.
15710         (cortexa7_younger): Likewise.
15712 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
15714         * ipa-pure-const.c (generate_summary): Rename to...
15715         (pure_const_generate_summary): ... this.
15717 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
15719         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
15721 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
15723         PR target/57909
15724         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
15725         usage in HI mode.
15727 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15729         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
15730         enabled without -march=zEC12.
15731         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
15732         flags to be set.
15734 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
15736         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
15737         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
15738         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
15739         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
15740         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
15741         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
15742         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
15743         ISA_HAS_FP4.
15744         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
15745         and ISA_HAS_NMADD3_NMSUB3.
15746         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
15747         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
15748         (nmsub4<mode>, nmsub3<mode>): Likewise.
15749         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
15751 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
15753         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
15754         TARGET_MIPS5400 checking.
15756 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
15757             Peter Bergner  <bergner@vnet.ibm.com>
15759         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
15760         registers in the comment.
15761         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
15762         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
15763         rather than FIRST_PSEUDO_REGISTERS.
15765 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
15767         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
15768         enable extra ISA flags with TARGET_HTM.
15770 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
15772         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
15773         Fix comment typos.
15775 2013-07-15  Cong Hou  <congh@google.com>
15777         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
15778         in compare function for sorting.
15780 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
15782         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
15783         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
15784         * config/rs6000/rs6000.opt: Add -mhtm option.
15785         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
15786         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
15787         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
15788         __HTM__ if the HTM instructions are available.
15789         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
15790         htm_spr_reg_operand): New define_predicates.
15791         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
15792         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
15793         Include htm.md.
15794         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
15795         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
15796         HTM builtin functions.
15797         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
15798         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
15799         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
15800         (rs6000_builtin_mask_calculate): Likewise.
15801         (rs6000_option_override_internal): Likewise.
15802         (bdesc_htm): Add new HTM builtin support.
15803         (htm_spr_num): New function.
15804         (htm_spr_regno): Likewise.
15805         (rs6000_htm_spr_icode): Likewise.
15806         (htm_expand_builtin): Likewise.
15807         (htm_init_builtins): Likewise.
15808         (rs6000_expand_builtin): Add support for HTM builtin functions.
15809         (rs6000_init_builtins): Likewise.
15810         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
15811         option.
15812         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
15813         (TARGET_HTM, MASK_HTM): Define macros.
15814         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
15815         (FIXED_REGISTERS): Likewise.
15816         (CALL_USED_REGISTERS): Likewise.
15817         (CALL_REALLY_USED_REGISTERS): Likewise.
15818         (REG_ALLOC_ORDER): Likewise.
15819         (enum reg_class): Likewise.
15820         (REG_CLASS_NAMES): Likewise.
15821         (REG_CLASS_CONTENTS): Likewise.
15822         (REGISTER_NAMES): Likewise.
15823         (ADDITIONAL_REGISTER_NAMES): Likewise.
15824         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
15825         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
15826         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
15827         * config/rs6000/htm.md: New file.
15828         * config/rs6000/htmintrin.h: New file.
15829         * config/rs6000/htmxlintrin.h: New file.
15831 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15833         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
15834         Define SYMBOL_TINY_GOT, update comment.
15835         * config/aarch64/aarch64.c
15836         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
15837         (aarch64_expand_mov_immediate): Likewise.
15838         (aarch64_print_operand): Likewise.
15839         (aarch64_classify_symbol): Likewise.
15840         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
15841         (ldr_got_tiny): Define.
15843 2013-07-13  Tobias Grosser  <tobias@grosser.es>
15845         PR tree-optimization/54094
15846         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
15847           scheduling dimension for the parallelism check from the polyhedral
15848           information in the AST.
15849         * graphite-dependences.c (carries_deps): Do not assume the schedule is
15850           in 2D + 1 form.
15852 2013-07-13  Jason Merrill  <jason@redhat.com>
15854         * print-tree.c (debug_vec_tree): Use debug_raw.
15855         (debug_raw (vec<tree, va_gc> &)): New.
15856         (debug_raw (vec<tree, va_gc> *)): New.
15857         * tree.h: Declare them.
15859 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
15861         * ifcvt.c (ifcvt_after_combine): New static variable.
15862         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
15863         for size.
15864         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
15865         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
15866         rest_of_handle_if_after_reload): Pass new argument for if_convert.
15868 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
15870         * config/mips/mips.c (mips_expand_call): Remove empty statement.
15872 2013-07-12  Michael Matz  <matz@suse.de>
15874         PR middle-end/55771
15875         * convert.c (convert_to_real): Reject non-float inner types.
15877 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
15879         * config/aarch64/aarch64-protos.h
15880         (aarch64_simd_immediate_valid_for_move): Remove.
15881         * config/aarch64/aarch64.c (simd_immediate_info): New member.
15882         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
15883         cases.
15884         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
15886 2013-07-11  Steve Ellcey  <sellcey@mips.com>
15888         * config/mips/mips.c (mips_conditional_register_usage): Do not
15889         use t[0-7] registers in MIPS16 mode when optimizing for size.
15891 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
15893         * config/i386/i386.c (dispatch_function_versions): Fix array
15894         indexing of function_version_info to match actual_versions.
15896 2013-07-11  Teresa Johnson  <tejohnson@google.com>
15898         * vec.h (struct va_gc): Move release out-of-line.
15899         (va_gc::release): Call ggc_free on released vec.
15901 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
15903         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
15904         Require GOT register as additional operand in UNSPEC.
15905         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
15906         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
15907         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
15908         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
15909         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
15910         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
15911         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
15913 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
15915         PR target/57631
15916         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
15917         name seen by assembler/linker rather if available.
15919 2013-07-11  Andreas Schwab  <schwab@suse.de>
15921         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
15923 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
15925         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
15927 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
15929         * doc/tm.texi.in: Move hook documentation to ....
15930         * target.def: ... here.
15932         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
15933         text on @hook line.
15934         * doc/tm.texi: Regenerate.
15936 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
15938         PR c++/57869
15939         * doc/invoke.texi: Document -Wconditionally-supported.
15941 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
15943         PR target/57844
15944         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
15945         of my_fp.
15947 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
15949         PR target/57506
15950         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
15951         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
15952         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
15953         Remove duplicate devices.
15954         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
15955         * config/avr/t-multilib: Regenerate.
15956         * config/avr/avr-tables.opt: Regenerate.
15957         * doc/avr-mmcu.texi: Regenerate.
15959 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
15961         PR target/56987
15962         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
15964 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
15966         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
15967         the cost of MULT when optimizing for size.
15969 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
15971         * config/cr16/cr16-protos.h: Don't include target.h.
15973 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
15975         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
15976         adjust register size for TDmode and TFmode for VSX registers.
15978 2013-07-08  Kai Tietz  <ktietz@redhat.com>
15980         PR target/56892
15981         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
15982         hook_bool_const_tree_true.
15984 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15986         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
15987         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
15988         * config/s390/s390.md: Define FPR*_REGNUM constants.
15989         Fix FPR2_REGNUM constant (18 -> 17).
15990         ("*trunc<BFP:mode><DFP_ALL:mode>2")
15991         ("*trunc<DFP_ALL:mode><BFP:mode>2")
15992         ("trunc<BFP:mode><DFP_ALL:mode>2")
15993         ("trunc<DFP_ALL:mode><BFP:mode>2")
15994         ("*extend<BFP:mode><DFP_ALL:mode>2")
15995         ("*extend<DFP_ALL:mode><BFP:mode>2")
15996         ("extend<BFP:mode><DFP_ALL:mode>2")
15997         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
15998         FPR4_REGNUM.
16000 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
16002         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
16004 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16006         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
16007         and cfun_fpr_bit_p to cfun_fpr_save_p.
16008         (s390_frame_area, s390_register_info, s390_frame_info)
16009         (s390_emit_prologue, s390_emit_epilogue)
16010         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
16011         register numbers.
16012         * config/s390/s390.h: Define *_REGNUM macros for floating point
16013         register numbers.
16015 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
16017         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
16019 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
16021         PR rtl-optimization/57786
16022         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
16023         and break out of the loop when it is set to false.
16025 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
16027         PR target/57819
16028         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
16029         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
16030         (const_int 63)) 0)).
16031         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
16032         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
16033         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
16035         PR rtl-optimization/57829
16036         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
16037         mask bits outside of mode are just sign-extension from mode to HWI.
16039 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
16041         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
16042         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
16043         adjust_address instead of change_address to keep info about alignment.
16044         (emit_strmov): Remove.
16045         (emit_memmov): New function.
16046         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
16047         (expand_movmem_epilogue): Likewise and return updated rtx for
16048         destination.
16049         (expand_constant_movmem_prologue): Likewise and return updated rtx for
16050         destination and source.
16051         (decide_alignment): Refactor, handle vector_loop.
16052         (ix86_expand_movmem): Likewise.
16053         (ix86_expand_setmem): Likewise.
16054         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
16056 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
16058         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
16059         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
16061 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
16063         * config/i386/sse.md (sse_movlhps): Change alternative 3
16064         of operand 2 to "m".
16066 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
16068         PR target/57807
16069         * config/i386/sse.md (iptr): New mode attribute.
16070         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
16071         (<sse>_vm<plusminus_insn><mode>3): Ditto.
16072         (<sse>_vmmul<mode>3): Ditto.
16073         (<sse>_vmdiv<mode>3): Ditto.
16074         (sse_vmrcpv4sf2): Ditto.
16075         (<sse>_vmsqrt<mode>2): Ditto.
16076         (sse_vmrsqrtv4sf2): Ditto.
16077         (<sse>_vm<code><mode>3): Ditto.
16078         (avx_vmcmp<mode>3): Ditto.
16079         (<sse>_vmmaskcmp<mode>3): Ditto.
16080         (<sse>_comi): Ditto.
16081         (<sse>_ucomi): Ditto.
16082         (*xop_vmfrcz_<mode>): Ditto.
16083         (*fmai_fmadd_<mode>): Ditto.
16084         (*fmai_fmsub_<mode>): Ditto.
16085         (*fmai_fnmadd_<mode>): Ditto.
16086         (*fmai_fnmsub_<mode>): Ditto.
16087         (*fma4i_vmfmadd_<mode>): Ditto.
16088         (*fma4i_vmfmsub_<mode>): Ditto.
16089         (*fma4i_vmfnmadd_<mode>): Ditto.
16090         (*fma4i_vmfnmsub_<mode>): Ditto.
16091         (*xop_vmfrcz_<mode>): Ditto.
16092         (sse_cvtps2pi): Ditto.
16093         (sse_cvttps2pi): Ditto.
16094         (sse_cvtss2si): Ditto.
16095         (sse_cvtss2si_2): Ditto.
16096         (sse_cvtss2siq_2): Ditto.
16097         (sse_cvttss2si): Ditto.
16098         (sse_cvttss2siq): Ditto.
16099         (sse_cvtsd2si): Ditto.
16100         (sse_cvtsd2si_2): Ditto.
16101         (sse_cvtsd2siq_2): Ditto.
16102         (sse_cvttsd2si): Ditto.
16103         (sse_cvttsd2siq): Ditto.
16104         (sse_cvtsd2ss): Ditto.
16105         (sse_cvtss2sd): Ditto.
16106         (avx2_pbroadcast<mode>): Ditto.
16107         (avx2_pbroadcast<mode>_1): Ditto.
16108         (*avx_vperm_broadcast_v4sf): Ditto.
16110         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
16111         (sse_movlhps): Ditto.
16112         (sse_storehps): Ditto.
16113         (sse_loadhps): Ditto.
16114         (sse_storelps): Ditto.
16115         (sse_loadlps): Ditto.
16116         (*vec_concatv4sf): Ditto.
16117         (*vec_interleave_highv2df): Ditto.
16118         (*vec_interleave_lowv2df): Ditto.
16119         (*vec_extractv2df_1_sse): Ditto.
16120         (*vec_extractv2df_0_sse): Ditto.
16121         (sse2_storelpd): Ditto.
16122         (sse2_loadlpd): Ditto.
16123         (sse2_movsd): Ditto.
16124         (*vec_concatv4si): Ditto.
16125         (vec_concatv2di): Ditto.
16127         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
16128         for Intel asm dialect.
16129         (mmx_punpcklwd): Ditto.
16130         (mmx_punpckldq): Ditto.
16132         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
16133         for intel assembler dialect.
16135 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
16137         PR target/29776
16138         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
16139         for BUILT_IN_C{LZ,LRSB}*.
16140         * tree.h (CASE_INT_FN): Add FN##IMAX case.
16141         * tree-vrp.c (extract_range_basic): Handle
16142         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
16143         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
16144         fall thru to code calling set_value*.
16145         * builtins.c (expand_builtin): Remove *IMAX cases.
16146         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
16147         if width is bigger than 2*HWI.
16149 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
16151         PR rtl-optimization/55342
16152         * lra-int.h (lra_subreg_reload_pseudos): New.
16153         * lra.c: Add undoing optional reloads to the block diagram.
16154         (lra_subreg_reload_pseudos): New.
16155         (lra_optional_reload_pseudos): Change comments.
16156         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
16157         lra_optional_reload_pseudos after undo transformations.
16158         * lra-assigns.c (pseudo_prefix_title): New.
16159         (lra_setup_reg_renumber): Use it.
16160         (spill_for): Ditto.  Check subreg reload pseudos too.
16161         (assign_by_spills): Consider subreg reload pseudos too.
16162         * lra-constraints.c (simplify_operand_subreg): Use
16163         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
16164         (curr_insn_transform): Recognize and do optional reloads.
16165         (undo_optional_reloads): New.
16166         (lra_undo_inheritance): Call undo_optional_reloads.
16168 2013-07-05  Thomas Quinot  <quinot@adacore.com>
16170         * tree-complex.c (expand_complex_operations_1): Fix typo.
16172 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
16174         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
16175         (tune_params): New member 'const vec_costs'.
16176         * config/aarch64/aarch64.c (generic_vector_cost): New.
16177         (generic_tunings): New member 'generic_vector_cost'.
16178         (aarch64_builtin_vectorization_cost): New.
16179         (aarch64_add_stmt_cost): New.
16180         (TARGET_VECTORIZE_ADD_STMT_COST): New.
16181         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
16183 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
16185         PR target/57777
16186         * config/i386/predicates.md (vsib_address_operand): Disallow
16187         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
16189 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
16191         PR middle-end/55030
16192         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
16193         expand_builtin_setjmp_receiver.
16194         (expand_label): Adjust, call expand_builtin_setjmp_receiver
16195         with NULL for the label parameter.
16196         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
16197         the frame-pointer.  Adjust comments.
16198         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
16199         only if LABEL is non-NULL.
16201 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
16203         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
16204         (ARM_ABI_AAPCS64): Ditto.
16205         (arm_abi): Ditto.
16206         (ARM_DEFAULT_ABI): Ditto.
16208 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
16210         * config/aarch64/aarch64-builtins.c
16211         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
16212         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
16213         (st1): Likewise.
16214         * config/aarch64/aarch64-simd.md
16215         (aarch64_ld1<VALL:mode>): New.
16216         (aarch64_st1<VALL:mode>): Likewise.
16217         * config/aarch64/arm_neon.h
16218         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
16220 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
16222         * config/i386/i386.c (gate_insert_vzeroupper): Check if
16223         target ISA is AVX.
16224         (ix86_option_override_internal):Turn on all -mavx target flags by
16225         default as they are dependent on AVX anyway.
16227 2013-07-02  Cary Coutant  <ccoutant@google.com>
16229         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
16230         deterministic hash.
16231         (loc_checksum_ordered): Likewise.
16232         (hash_loc_operands): Remove inline keyword.
16234 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
16236         PR tree-optimization/57741
16237         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
16238         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
16239         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
16240         Allow REAL_CST step_exprs if flag_associative_math.
16241         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
16243 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
16245         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
16247 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
16249         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
16251 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16253         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
16254         encoding.
16255         (iorsi3_insn): Likewise.
16256         (arm_xorsi3): Likewise.
16258 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
16260         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
16261         to ...
16262         (attribute "type"): ... here, and prefix with "wmmx_".
16263         (attribute "core_cycles"): Update for attribute changes.
16264         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
16265         (tbcstv4hi): Likewise.
16266         (tbcstv2si): Likewise.
16267         (iwmmxt_iordi3): Likewise.
16268         (iwmmxt_xordi3): Likewise.
16269         (iwmmxt_anddi3): Likewise.
16270         (iwmmxt_nanddi3): Likewise.
16271         (iwmmxt_arm_movdi): Likewise.
16272         (iwmmxt_movsi_insn): Likewise.
16273         (mov<mode>_internal): Likewise.
16274         (and<mode>3_iwmmxt): Likewise.
16275         (ior<mode>3_iwmmxt): Likewise.
16276         (xor<mode>3_iwmmxt): Likewise.
16277         (add<mode>3_iwmmxt): Likewise.
16278         (ssaddv8qi3): Likewise.
16279         (ssaddv4hi3): Likewise.
16280         (ssaddv2si3): Likewise.
16281         (usaddv8qi3): Likewise.
16282         (usaddv4hi3): Likewise.
16283         (usaddv2si3): Likewise.
16284         (sub<mode>3_iwmmxt): Likewise.
16285         (sssubv8qi3): Likewise.
16286         (sssubv4hi3): Likewise.
16287         (sssubv2si3): Likewise.
16288         (ussubv8qi3): Likewise.
16289         (ussubv4hi3): Likewise.
16290         (ussubv2si3): Likewise.
16291         (mulv4hi3_iwmmxt): Likewise.
16292         (smulv4hi3_highpart): Likewise.
16293         (umulv4hi3_highpart): Likewise.
16294         (iwmmxt_wmacs): Likewise.
16295         (iwmmxt_wmacsz): Likewise.
16296         (iwmmxt_wmacu): Likewise.
16297         (iwmmxt_wmacuz): Likewise.
16298         (iwmmxt_clrdi): Likewise.
16299         (iwmmxt_clrv8qi): Likewise.
16300         (iwmmxt_clr4hi): Likewise.
16301         (iwmmxt_clr2si): Likewise.
16302         (iwmmxt_uavgrndv8qi3): Likewise.
16303         (iwmmxt_uavgrndv4hi3): Likewise.
16304         (iwmmxt_uavgv8qi3): Likewise.
16305         (iwmmxt_uavgv4hi3): Likewise.
16306         (iwmmxt_tinsrb): Likewise.
16307         (iwmmxt_tinsrh): Likewise.
16308         (iwmmxt_tinsrw): Likewise.
16309         (iwmmxt_textrmub): Likewise.
16310         (iwmmxt_textrmsb): Likewise.
16311         (iwmmxt_textrmuh): Likewise.
16312         (iwmmxt_textrmsh): Likewise.
16313         (iwmmxt_textrmw): Likewise.
16314         (iwmxxt_wshufh): Likewise.
16315         (eqv8qi3): Likewise.
16316         (eqv4hi3): Likewise.
16317         (eqv2si3): Likewise.
16318         (gtuv8qi3): Likewise.
16319         (gtuv4hi3): Likewise.
16320         (gtuv2si3): Likewise.
16321         (gtv8qi3): Likewise.
16322         (gtv4hi3): Likewise.
16323         (gtv2si3): Likewise.
16324         (smax<mode>3_iwmmxt): Likewise.
16325         (umax<mode>3_iwmmxt): Likewise.
16326         (smin<mode>3_iwmmxt): Likewise.
16327         (umin<mode>3_iwmmxt): Likewise.
16328         (iwmmxt_wpackhss): Likewise.
16329         (iwmmxt_wpackwss): Likewise.
16330         (iwmmxt_wpackdss): Likewise.
16331         (iwmmxt_wpackhus): Likewise.
16332         (iwmmxt_wpackwus): Likewise.
16333         (iwmmxt_wpackdus): Likewise.
16334         (iwmmxt_wunpckihb): Likewise.
16335         (iwmmxt_wunpckihh): Likewise.
16336         (iwmmxt_wunpckihw): Likewise.
16337         (iwmmxt_wunpckilb): Likewise.
16338         (iwmmxt_wunpckilh): Likewise.
16339         (iwmmxt_wunpckilw): Likewise.
16340         (iwmmxt_wunpckehub): Likewise.
16341         (iwmmxt_wunpckehuh): Likewise.
16342         (iwmmxt_wunpckehuw): Likewise.
16343         (iwmmxt_wunpckehsb): Likewise.
16344         (iwmmxt_wunpckehsh): Likewise.
16345         (iwmmxt_wunpckehsw): Likewise.
16346         (iwmmxt_wunpckelub): Likewise.
16347         (iwmmxt_wunpckeluh): Likewise.
16348         (iwmmxt_wunpckeluw): Likewise.
16349         (iwmmxt_wunpckelsb): Likewise.
16350         (iwmmxt_wunpckelsh): Likewise.
16351         (iwmmxt_wunpckelsw): Likewise.
16352         (ror<mode>3): Likewise.
16353         (ashr<mode>3_iwmmxt): Likewise.
16354         (lshr<mode>3_iwmmxt): Likewise.
16355         (ashl<mode>3_iwmmxt): Likewise.
16356         (ror<mode>3_di): Likewise.
16357         (ashr<mode>3_di): Likewise.
16358         (lshr<mode>3_di): Likewise.
16359         (ashl<mode>3_di): Likewise.
16360         (iwmmxt_wmadds): Likewise.
16361         (iwmmxt_wmaddu): Likewise.
16362         (iwmmxt_tmia): Likewise.
16363         (iwmmxt_tmiaph): Likewise.
16364         (iwmmxt_tmiabb): Likewise.
16365         (iwmmxt_tmiatb): Likewise.
16366         (iwmmxt_tmiabt): Likewise.
16367         (iwmmxt_tmiatt): Likewise.
16368         (iwmmxt_tmovmskb): Likewise.
16369         (iwmmxt_tmovmskh): Likewise.
16370         (iwmmxt_tmovmskw): Likewise.
16371         (iwmmxt_waccb): Likewise.
16372         (iwmmxt_wacch): Likewise.
16373         (iwmmxt_waccw): Likewise.
16374         (iwmmxt_waligni): Likewise.
16375         (iwmmxt_walignr): Likewise.
16376         (iwmmxt_walignr0): Likewise.
16377         (iwmmxt_walignr1): Likewise.
16378         (iwmmxt_walignr2): Likewise.
16379         (iwmmxt_walignr3): Likewise.
16380         (iwmmxt_wsadb): Likewise.
16381         (iwmmxt_wsadh): Likewise.
16382         (iwmmxt_wsadbz): Likewise.
16383         (iwmmxt_wsadhz): Likewise.
16384         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
16385         (iwmmxt_wabsdiffb): Likewise.
16386         (iwmmxt_wabsdiffh): Likewise.
16387         (iwmmxt_wabsdiffw): Likewise.
16388         (iwmmxt_waddsubhx): Likewise
16389         (iwmmxt_wsubaddhx): Likewise.
16390         (addc<mode>3): Likewise.
16391         (iwmmxt_avg4): Likewise.
16392         (iwmmxt_avg4r): Likewise.
16393         (iwmmxt_wmaddsx): Likewise.
16394         (iwmmxt_wmaddux): Likewise.
16395         (iwmmxt_wmaddsn): Likewise.
16396         (iwmmxt_wmaddun): Likewise.
16397         (iwmmxt_wmulwsm): Likewise.
16398         (iwmmxt_wmulwum): Likewise.
16399         (iwmmxt_wmulsmr): Likewise.
16400         (iwmmxt_wmulumr): Likewise.
16401         (iwmmxt_wmulwsmr): Likewise.
16402         (iwmmxt_wmulwumr): Likewise.
16403         (iwmmxt_wmulwl): Likewise.
16404         (iwmmxt_wqmulm): Likewise.
16405         (iwmmxt_wqmulwm): Likewise.
16406         (iwmmxt_wqmulmr): Likewise.
16407         (iwmmxt_wqmulwmr): Likewise.
16408         (iwmmxt_waddbhusm): Likewise.
16409         (iwmmxt_waddbhusl): Likewise.
16410         (iwmmxt_wqmiabb): Likewise.
16411         (iwmmxt_wqmiabt): Likewise.
16412         (iwmmxt_wqmiatb): Likewise.
16413         (iwmmxt_wqmiatt): Likewise.
16414         (iwmmxt_wqmiabbn): Likewise.
16415         (iwmmxt_wqmiabtn): Likewise.
16416         (iwmmxt_wqmiatbn): Likewise.
16417         (iwmmxt_wqmiattn): Likewise.
16418         (iwmmxt_wmiabb): Likewise.
16419         (iwmmxt_wmiabt): Likewise.
16420         (iwmmxt_wmiatb): Likewise.
16421         (iwmmxt_wmiatt): Likewise.
16422         (iwmmxt_wmiabbn): Likewise.
16423         (iwmmxt_wmiabtn): Likewise.
16424         (iwmmxt_wmiatbn): Likewise.
16425         (iwmmxt_wmiattn): Likewise.
16426         (iwmmxt_wmiawbb): Likewise.
16427         (iwmmxt_wmiawbt): Likewise.
16428         (iwmmxt_wmiawtb): Likewise.
16429         (iwmmxt_wmiawtt): Likewise.
16430         (iwmmxt_wmiawbbn): Likewise.
16431         (iwmmxt_wmiawbtn): Likewise.
16432         (iwmmxt_wmiawtbn): Likewise.
16433         (iwmmxt_wmiawttn): Likewise.
16434         (iwmmxt_wmerge): Likewise.
16435         (iwmmxt_tandc<mode>3): Likewise.
16436         (iwmmxt_torc<mode>3): Likewise.
16437         (iwmmxt_torvsc<mode>3): Likewise.
16438         (iwmmxt_textrc<mode>3): Likewise.
16439         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
16440         (wmmxt_pack): Likewise.
16441         (wmmxt_mult_c1): Likewise.
16442         (wmmxt_mult_c2): Likewise.
16443         (wmmxt_alu_c1): Likewise.
16444         (wmmxt_alu_c2): Likewise.
16445         (wmmxt_alu_c3): Likewise.
16446         (wmmxt_transfer_c1): Likewise.
16447         (wmmxt_transfer_c2): Likewise.
16448         (wmmxt_transfer_c3): Likewise.
16449         (marvell_f_iwmmxt_wstr): Likewise.
16450         (marvell_f_iwmmxt_wldr): Likewise.
16452 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
16454         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
16456 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
16458         Revert:
16459         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
16460         * lra-constraints.c (need_for_split_p): Check call used hard regs
16461         living through calls.
16463         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
16464         call used regs for call insn.
16466 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
16468         PR target/57736
16469         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
16470         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
16472 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
16474         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
16476 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
16478         * lra-constraints.c (need_for_split_p): Check call used hard regs
16479         living through calls.
16481 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
16483         PR target/57744
16484         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
16485         to tie with any other modes.  Eliminate Altivec vector mode tests,
16486         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
16487         code, to return 0 if testing MODE2 for a condition, if we've
16488         already tested MODE1 for the same condition.
16490 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
16492         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
16493         layout.
16495 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
16497         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
16498         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
16500 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
16502         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
16503         Define.
16504         (aarch64_symbolic_constant_p): Remove.
16505         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
16506         static.  Fix line length and white space.
16507         (aarch64_symbolic_constant_p): Remove.
16508         * config/aarch64/predicates.md (aarch64_valid_symref):
16509         Use aarch64_classify_symbol_expression.
16511 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16513         * config/arm/constraints.md (Ts): New constraint.
16514         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
16515         16-bit encodings.
16516         (compare_scc): Use "Ts" constraint for operand 0.
16517         (ior_scc_scc): Likewise.
16518         (and_scc_scc): Likewise.
16519         (and_scc_scc_nodom): Likewise.
16520         (ior_scc_scc_cmp): Likewise for operand 7.
16521         (and_scc_scc_cmp): Likewise.
16522         * config/arm/thumb2.md (thumb2_movsi_insn):
16523         Add alternatives for 16-bit encodings.
16524         (thumb2_movhi_insn): Likewise.
16525         (thumb2_movsicc_insn): Likewise.
16526         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
16527         (thumb2_negscc): Use "Ts" constraint.
16528         Move mvn instruction outside cond_exec block.
16529         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
16530         for 16-bit encodings.
16532 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16534         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
16535         encoding.
16536         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
16537         (mulsi3subsi): Likewise.
16538         (mulsidi3adddi): Likewise.
16539         (mulsidi3_v6): Likewise.
16540         (umulsidi3_v6): Likewise.
16541         (umulsidi3adddi_v6): Likewise.
16542         (smulsi3_highpart_v6): Likewise.
16543         (umulsi3_highpart_v6): Likewise.
16544         (mulhisi3tb): Likewise.
16545         (mulhisi3bt): Likewise.
16546         (mulhisi3tt): Likewise.
16547         (maddhisi4): Likewise.
16548         (maddhisi4tb): Likewise.
16549         (maddhisi4tt): Likewise.
16550         (maddhidi4): Likewise.
16551         (maddhidi4tb): Likewise.
16552         (maddhidi4tt): Likewise.
16553         (zeroextractsi_compare0_scratch): Likewise.
16554         (insv_zero): Likewise.
16555         (insv_t2): Likewise.
16556         (anddi_notzesidi_di): Likewise.
16557         (anddi_notsesidi_di): Likewise.
16558         (andsi_notsi_si): Likewise.
16559         (iordi_zesidi_di): Likewise.
16560         (xordi_zesidi_di): Likewise.
16561         (andsi_iorsi3_notsi): Likewise.
16562         (smax_0): Likewise.
16563         (smax_m1): Likewise.
16564         (smin_0): Likewise.
16565         (not_shiftsi): Likewise.
16566         (unaligned_loadsi): Likewise.
16567         (unaligned_loadhis): Likewise.
16568         (unaligned_loadhiu): Likewise.
16569         (unaligned_storesi): Likewise.
16570         (unaligned_storehi): Likewise.
16571         (extv_reg): Likewise.
16572         (extzv_t2): Likewise.
16573         (divsi3): Likewise.
16574         (udivsi3): Likewise.
16575         (arm_zero_extendhisi2addsi): Likewise.
16576         (arm_zero_extendqisi2addsi): Likewise.
16577         (compareqi_eq0): Likewise.
16578         (arm_extendhisi2_v6): Likewise.
16579         (arm_extendqisi2addsi): Likewise.
16580         (arm_movt): Likewise.
16581         (thumb2_ldrd): Likewise.
16582         (thumb2_ldrd_base): Likewise.
16583         (thumb2_ldrd_base_neg): Likewise.
16584         (thumb2_strd): Likewise.
16585         (thumb2_strd_base): Likewise.
16586         (thumb2_strd_base_neg): Likewise.
16587         (arm_negsi2): Add alternative for 16-bit encoding.
16588         (arm_one_cmplsi2): Likewise.
16590 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16592         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
16593         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
16594         (movdfcc): Likewise.
16595         * config/arm/vfp.md (*thumb2_movsf_vfp):
16596         Disable predication for arm_restrict_it.
16597         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
16598         (*thumb2_movdfcc_vfp): Likewise.
16599         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
16600         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
16601         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
16602         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
16603         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
16604         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
16605         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
16606         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
16607         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
16608         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
16609         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
16610         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
16611         Disable predication for arm_restrict_it.
16613 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
16615         * config/i386/bmiintrin.h (_bextr_u32): New.
16616         (_bextr_u64): Ditto.
16618 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
16620         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
16621         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
16622         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
16623         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
16624         * config/mips/n32-elf.h: ...this new file.
16626 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
16628         PR target/57224
16629         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
16630         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
16632 2013-06-27  Catherine Moore  <clm@codesourcery.com>
16634         * config/mips/mips-tables.opt: Regenerate.
16635         * config/mips/mips-cpus.def: Add m14ke and m14kec.
16636         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
16637         * doc/invoke.texi: Add -m14kc.
16639 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
16641         PR target/57623
16642         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
16643         constraints of operand 1 and 2.
16645         PR target/57623
16646         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
16647         to match RTL canonicalization.  Swap predicates and
16648         constraints of operand 1 and 2.
16650 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
16652         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
16653         Process OP_INOUT regs for splitting too.
16655 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
16657         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
16658         decl before the loop, initialize to NULL.
16659         (vectorizable_load): Initialize ptr_incr to NULL.
16661 2013-06-27  Martin Jambor  <mjambor@suse.cz>
16663         PR lto/57208
16664         * ipa-ref.h (ipa_maybe_record_reference): Declare.
16665         * ipa-ref.c (ipa_maybe_record_reference): New function.
16666         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
16667         * ipa-cp.c (create_specialized_node): Record potential references from
16668         aggvals.
16669         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
16671 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
16673         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
16674         parameter 'mode' of type 'enum machine_mode mode'; change to pass
16675         'mode' to force_reg.
16676         (aarch64_add_offset): Update calls to aarch64_force_temporary.
16677         (aarch64_expand_mov_immediate): Likewise.
16679 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
16681         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
16682         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
16684 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16686         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
16687         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
16688         (struct machine_function): Add tbegin_p.
16689         (s390_canonicalize_comparison): Fold CC mode compares to
16690         conditional jump if possible.
16691         (s390_emit_jump): Return the emitted jump.
16692         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
16693         Handle CCRAWmode compares.
16694         (s390_option_override): Default to -mhtm if available.
16695         (s390_reg_clobbered_rtx): Handle floating point regs as well.
16696         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
16697         FPRs instead of df_regs_ever_live_p.
16698         (s390_optimize_nonescaping_tx): New function.
16699         (s390_init_frame_layout): Extend clobbered_regs array to cover
16700         FPRs as well.
16701         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
16702         (s390_expand_tbegin): New function.
16703         (enum s390_builtin): New enum definition.
16704         (code_for_builtin): New array definition.
16705         (s390_init_builtins): New function.
16706         (s390_expand_builtin): New function.
16707         (TARGET_INIT_BUILTINS): Define.
16708         (TARGET_EXPAND_BUILTIN): Define.
16709         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
16710         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
16711         (s390_alc_comparison): Likewise.
16712         * config/s390/s390-modes.def: Add CCRAWmode.
16713         * config/s390/s390.h (processor_flags): Add PF_TX.
16714         (TARGET_CPU_HTM): Define macro.
16715         (TARGET_HTM): Define macro.
16716         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
16717         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
16718         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
16719         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
16720         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
16721         (TBEGIN_MASK, TBEGINC_MASK): New constants.
16722         ("*cc_to_int"): Move up.
16723         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
16724         constants other than 0.
16725         ("*ccraw_to_int"): New insn and splitter definition.
16726         ("tbegin", "tbegin_nofloat", "tbegin_retry")
16727         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
16728         ("tx_assist"): New expander.
16729         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
16730         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
16731         * config/s390/s390.opt: Add -mhtm option.
16732         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
16733         * config/s390/htmxlintrin.h: New file.
16734         * config/s390/htmintrin.h: New file.
16735         * config/s390/s390intrin.h: New file.
16736         * doc/extend.texi: Document htm builtins.
16737         * config.gcc: Add the new header files to extra_headers.
16739 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
16741         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
16742         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
16744 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
16745             Pat Haugen  <pthaugen@us.ibm.com>
16746             Peter Bergner  <bergner@vnet.ibm.com>
16748         * config/rs6000/power8.md: New.
16749         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
16750         setting for power8 entry.
16751         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
16752         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
16753         test for Power4/Power5 only.
16754         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
16755         support.
16756         (force_new_group): Adjust comment.
16757         * config/rs6000/rs6000.md: Include power8.md.
16759 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
16761         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
16762         * config/arm/arm-protos.h (arm_max_conditional_execute): New
16763         declaration.
16764         (tune_params): Update comment.
16765         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
16766         (arm_max_conditional_execute): New function.
16767         (thumb2_final_prescan_insn): Use max_insn_skipped and
16768         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
16770 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
16772         PR tree-optimization/57705
16773         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
16774         SSA_NAME step, provided that it is not defined inside the loop.
16775         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
16776         (get_initial_def_for_induction): Handle SSA_NAME IV step.
16778 2013-06-25  Martin Jambor  <mjambor@suse.cz>
16780         PR middle-end/57670
16781         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
16782         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
16783         calls in the dump.
16784         (ipa_note_param_call): Initialize member_ptr flag.
16785         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
16786         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
16787         (ipa_write_indirect_edge_info): Stream member_ptr flag.
16788         (ipa_read_indirect_edge_info): Likewise.
16790 2013-06-25  Richard Biener  <rguenther@suse.de>
16792         PR middle-end/56977
16793         * passes.c (init_optimization_passes): Move pass_fold_builtins
16794         and pass_dce earlier with -Og.
16796 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
16798         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
16799         <BIT_FIELD_REF>: Remove trailing TAB.
16800         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
16801         remove blank line.
16803 2013-06-24  Martin Jambor  <mjambor@suse.cz>
16805         PR tree-optimization/57358
16806         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
16807         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
16808         (ipa_analyze_params_uses): Generate pessimistic info when true.
16810 2013-06-24  Martin Jambor  <mjambor@suse.cz>
16812         PR tree-optimization/57539
16813         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
16814         global.inlined_to of the new node to it.  All callers changed.
16815         * ipa-inline-transform.c (clone_inlined_nodes): New variable
16816         inlining_into, pass it to cgraph_clone_node.
16817         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
16818         ipa_free_edge_args_substructures.
16819         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
16820         rdesc linked list.  Do not assert rdesc edges have inlined caller.
16821         Assert we have found an rdesc in the rdesc list.
16823 2013-06-24  Richard Biener  <rguenther@suse.de>
16825         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
16826         (pointer_set_lookup): Declare.
16827         (class pointer_map): New template class implementing a
16828         generic pointer to T map.
16829         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
16830         pointer_map<T>::contains, pointer_map<T>::insert,
16831         pointer_map<T>::traverse): New functions.
16832         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
16833         (pointer_set_lookup): New function.
16834         (pointer_set_contains): Use pointer_set_lookup.
16835         (pointer_set_insert): Likewise.
16836         (insert_aux): Remove.
16837         (struct pointer_map_t): Embed a pointer_set_t.
16838         (pointer_map_create): Adjust.
16839         (pointer_map_destroy): Likewise.
16840         (pointer_map_contains): Likewise.
16841         (pointer_map_insert): Likewise.
16842         (pointer_map_traverse): Likewise.
16843         * tree-streamer.h (struct streamer_tree_cache_d): Use a
16844         pointer_map<unsigned> instead of a pointer_map_t.
16845         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
16846         (streamer_tree_cache_lookup): Likewise.
16847         (streamer_tree_cache_create): Likewise.
16848         (streamer_tree_cache_delete): Likewise.
16849         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
16850         pointer_map<unsigned> instead of a pointer_map_t.
16851         (lto_init_tree_ref_encoder): Adjust.
16852         (lto_destroy_tree_ref_encoder): Likewise.
16853         * lto-section-out.c (lto_output_decl_index): Likewise.
16854         (lto_record_function_out_decl_state): Likewise.
16855         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
16857 2013-06-24  Richard Biener  <rguenther@suse.de>
16859         PR tree-optimization/57488
16860         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
16862 2013-06-24  Alan Modra  <amodra@gmail.com>
16864         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
16865         (gen_easy_altivec_constant): Likewise.
16866         * config/rs6000/predicates.md (easy_vector_constant_add_self,
16867         easy_vector_constant_msb): Likewise.
16869 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
16871         PR target/57688
16872         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
16873         add missing return true.
16875 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
16877         PR target/52483
16878         * config/sh/predicates.md (general_extend_operand): Invoke
16879         general_movsrc_operand for memory operands.
16880         (general_movsrc_operand): Allow reg+reg addressing, do not use
16881         general_operand for memory operands.
16883 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
16885         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
16886         when current target options does not apply.
16887         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
16888         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
16889         * config/i386/bmiintrin.h: Pass appropriate target
16890         attributes to header.
16891         * config/i386/mmintrin.h: Ditto.
16892         * config/i386/nmmintrin.h: Ditto.
16893         * config/i386/avx2intrin.h: Ditto.
16894         * config/i386/fxsrintrin.h: Ditto.
16895         * config/i386/tbmintrin.h: Ditto.
16896         * config/i386/xsaveintrin.h: Ditto.
16897         * config/i386/f16cintrin.h: Ditto.
16898         * config/i386/xtestintrin.h: Ditto.
16899         * config/i386/xsaveoptintrin.h: Ditto.
16900         * config/i386/bmi2intrin.h: Ditto.
16901         * config/i386/lzcntintrin.h: Ditto.
16902         * config/i386/smmintrin.h: Ditto.
16903         * config/i386/wmmintrin.h: Ditto.
16904         * config/i386/x86intrin.h: Remove all header include guards.
16905         * config/i386/prfchwintrin.h: Ditto.
16906         * config/i386/pmmintrin.h: Ditto.
16907         * config/i386/tmmintrin.h: Ditto.
16908         * config/i386/xmmintrin.h: Ditto.
16909         * config/i386/popcntintrin.h: Ditto.
16910         * config/i386/rdseedintrin.h: Ditto.
16911         * config/i386/ammintrin.h: Ditto.
16912         * config/i386/emmintrin.h: Ditto.
16913         * config/i386/immintrin.h: Remove all header include guards.
16914         * config/i386/fma4intrin.h: Ditto.
16915         * config/i386/lwpintrin.h: Ditto.
16916         * config/i386/xopintrin.h: Ditto.
16917         * config/i386/ia32intrin.h: Ditto.
16918         * config/i386/avxintrin.h: Ditto.
16919         * config/i386/rtmintrin.h: Ditto.
16920         * config/i386/fmaintrin.h: Ditto.
16921         * config/i386/mm3dnow.h: Ditto.
16923 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
16925         * common/config/i386/i386-common.c: Handle LZCNT.
16927 2013-06-22  Andi Kleen  <ak@linux.intel.com>
16929         * doc/extend.texi: Use __atomic_store_n instead of
16930         __atomic_store in HLE example.
16932 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
16934         * config/sh/sh.c: Remove <cstdlib> workaround.
16936 2013-06-21  Andi Kleen  <ak@linux.intel.com>
16938         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
16940 2013-06-21  Andi Kleen  <ak@linux.intel.com>
16942         * doc/extend.texi: Document that __atomic_clear and
16943           __atomic_test_and_set should only be used with bool.
16945 2013-06-20  Jan Hubicka  <jh@suse.cz>
16947         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
16948         types_same_for_odr.
16949         * tree.c (decls_same_for_odr): New function.
16950         (same_for_edr): New function.
16951         (types_same_for_odr): New function.
16952         (get_binfo_at_offset): Use it.
16953         * tree.h (types_same_for_odr): Declare.
16955 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
16956             Jason Merrill  <jason@redhat.com>
16958         * system.h: Include <cstdlib> as well as <stdlib.h>.
16960 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
16962         PR target/57655
16963         * config/i386/i386.c (construct_container): Report error if
16964         long double is used with disabled x87 float returns.
16966 2013-06-20  Jan Hubicka  <jh@suse.cz>
16968         * lto-cgraph.c (input_symtab): Do not set cgraph state.
16970 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
16972         PR rtl-optimization/57425
16973         PR rtl-optimization/57569
16974         * alias.c (write_dependence_p): Remove parameters mem_mode and
16975         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
16976         Changed all callers.
16977         (canon_anti_dependence): Get comments and semantics in sync.
16978         Add parameter mem_canonicalized.  Changed all callers.
16979         * rtl.h (canon_anti_dependence): Update prototype.
16981 2013-06-20  Richard Biener  <rguenther@suse.de>
16983         * data-streamer-in.c (streamer_read_uhwi): Optimize single
16984         byte case, inline streamer_read_uchar and defer section
16985         overrun check.
16987 2013-06-20  Richard Biener  <rguenther@suse.de>
16989         PR tree-optimization/57584
16990         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
16991         SSA names into the expanded expression that take part in
16992         abnormal coalescing.
16994 2013-06-19  Sharad Singhai  <singhai@google.com>
16996         * gcov.c (print_usage): Handle new option.
16997         (process_args): Ditto.
16998         (get_gcov_intermediate_filename): New function.
16999         (output_intermediate_file): New function.
17000         (output_gcov_file): New function
17001         (generate_results): Handle new option.
17002         (release_function): Relase demangled name.
17003         (read_graph_file): Handle demangled name.
17004         (output_lines): Ditto.
17005         * doc/gcov.texi: Document gcov intermediate format.
17007 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
17009         PR bootstrap/57604
17010         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
17011         (lra_emit_add): Use the functions.  Add comment about Y as an
17012         address segment.
17014 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
17016         PR driver/57652
17017         * collect2.c (collect_atexit): New.
17018         (collect_exit): Delete.
17019         (main): Register collect_atexit with atexit.
17020         (collect_wait): Change collect_exit to exit.
17021         (do_wait): Same.
17022         * collect2.h (collect_exit): Delete.
17023         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
17025 2013-06-19  Wei Mi  <wmi@google.com>
17027         PR rtl-optimization/57518
17028         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
17029         if regno is used in paradoxical subreg.
17030         (update_equiv_regs): Check pdx_subregs[regno] before
17031         set a reg to be equivalent with a mem.
17033 2013-06-19  Matthias Klose  <doko@ubuntu.com>
17035         PR driver/57651
17036         * file-find.h (find_a_file): Add a mode parameter.
17037         * file-find.c (find_a_file): Likewise.
17038         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
17039         with X_OK for the executables.
17040         * collect2.c (main): Call find_a_file with X_OK.
17042 2013-06-19  Steve Ellcey  <sellcey@mips.com>
17044         PR target/56942
17045         * config/mips/mips.md (casesi_internal_mips16_<mode>):
17046         Use NEXT_INSN instead of next_real_insn.
17048 2013-06-19  Jan Hubicka  <jh@suse.cz>
17050         * cgraph.h (const_value_known_p): Replace by ...
17051         (ctor_for_folding): .. this one.
17052         * cgraphunit.c (process_function_and_variable_attributes): Use it.
17053         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
17054         * expr.c (expand_expr_real_1): Likewise.
17055         (string_constant): Likewise.
17056         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
17057         * ipa.c (process_references): Likewise.
17058         (symtab_remove_unreachable_nodes): Likewise.
17059         * ipa-inline-analysis.c (param_change_prob): Likewise.
17060         * gimple-fold.c (canonicalize_constructor_val): Likewise.
17061         (get_base_constructor): Likwise.
17062         * varpool.c (varpool_remove_node): Likewise.
17063         (varpool_remove_initializer): LIkewise.
17064         (dump_varpool_node): LIkwise.
17065         (const_value_known_p): Rewrite to ...
17066         (ctor_for_folding): ... this one.
17068 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
17070         PR driver/57651
17071         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
17072         PERSONALITY in $PATH derived prefixes.
17074 2013-06-19  Jeff Law  <law@redhat.com>
17076         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
17077         in comment.
17079         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
17080         (simplify_bitwise_binary): Use it to simpify certain binary ops on
17081         booleans.
17083 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
17085         * config/arm/vfp.md: Move VFP instruction classification documentation
17086         to ...
17087         * config/arm/arm.md: ... here.  Update instruction classification
17088         documentation.
17090 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
17092         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
17093         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
17094         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
17096 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
17098         PR target/57609
17099         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
17100         with NEXT_INSN.  Use tablejump_p to check for jump table data
17101         insns.
17103 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
17105         PR c++/56544
17106         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
17107         that now in C++ the value is correct per the C++ standards.
17109 2013-06-19  Richard Biener  <rguenther@suse.de>
17111         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
17112         for global context.
17114 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17116         Revert:
17117         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17119         PR target/57609
17120         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
17121         with next_active_insn.
17123 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
17125         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
17126         functions are inlined during failures to flag an error.
17127         * tree-inline.c (expand_call_inline): Allow the error to be flagged
17128         in early inline pass.
17130 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
17132         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
17133         in comments.
17135 2013-06-18  Julian Brown  <julian@codesourcery.com>
17137         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
17138         Permit virtual register pre-reload if !strict.
17139         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
17140         change.
17141         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
17142         prototype.
17143         * config/arm/neon.md (movmisalign<mode>): Use
17144         neon_perm_struct_or_reg_operand instead of
17145         neon_struct_or_register_operand.
17146         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
17147         neon_permissive_struct_operand instead of neon_struct_operand.
17148         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
17149         neon_vector_mem_operand.
17150         * config/arm/predicates.md (neon_struct_operand): Adjust call to
17151         neon_vector_mem_operand.
17152         (neon_permissive_struct_operand): New.
17153         (neon_struct_or_register_operand): Rename to...
17154         (neon_perm_struct_or_reg_operand): This. Adjust call to
17155         neon_vector_mem_operand.
17157 2013-06-18  Richard Biener  <rguenther@suse.de>
17159         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
17160         * lto-streamer.h: Include pointer-set.h.
17161         (struct lto_decl_slot): Remove.
17162         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
17163         Remove next_index entry.
17164         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
17165         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
17166         (lto_init_tree_ref_encoder): Adjust.
17167         (lto_destroy_tree_ref_encoder): Likewise.
17168         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
17169         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
17170         (lto_output_decl_index): Adjust.
17171         (lto_new_out_decl_state): Likewise.
17172         (lto_record_function_out_decl_state): Likewise.
17173         * lto-streamer-out.c (copy_function): Likewise.
17175 2013-06-18  Richard Biener  <rguenther@suse.de>
17177         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
17178         * cgraphunit.c: Include cfgloop.h.
17179         (init_lowered_empty_function): Initialize the loop tree.
17180         (assemble_thunk): Insert new BBs into loops.
17182 2013-06-18  Richard Biener  <rguenther@suse.de>
17184         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
17185         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
17186         the map from cache entry to cache index optional.
17187         (streamer_tree_cache_replace_tree): Adjust accordingly.
17188         (streamer_tree_cache_append): Likewise.
17189         (streamer_tree_cache_delete): Likewise.
17190         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
17191         streamer cache map from cache entry to cache index.
17192         * lto-streamer-out.c (create_output_block): Adjust.
17194 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
17196         * config/arm/arm.md (attribute "insn"): Move multiplication and
17197         division attributes to...
17198         (attribute "type"): ... here.  Remove mult.
17199         (attribute "mul32"): New attribute.
17200         (attribute "mul64"): Add umaal.
17201         (*arm_mulsi3): Update attributes.
17202         (*arm_mulsi3_v6): Likewise.
17203         (*thumb_mulsi3): Likewise.
17204         (*thumb_mulsi3_v6): Likewise.
17205         (*mulsi3_compare0): Likewise.
17206         (*mulsi3_compare0_v6): Likewise.
17207         (*mulsi_compare0_scratch): Likewise.
17208         (*mulsi_compare0_scratch_v6): Likewise.
17209         (*mulsi3addsi): Likewise.
17210         (*mulsi3addsi_v6): Likewise.
17211         (*mulsi3addsi_compare0): Likewise.
17212         (*mulsi3addsi_compare0_v6): Likewise.
17213         (*mulsi3addsi_compare0_scratch): Likewise.
17214         (*mulsi3addsi_compare0_scratch_v6): Likewise.
17215         (*mulsi3subsi): Likewise.
17216         (*mulsidi3adddi): Likewise.
17217         (*mulsi3addsi_v6): Likewise.
17218         (*mulsidi3adddi_v6): Likewise.
17219         (*mulsidi3_nov6): Likewise.
17220         (*mulsidi3_v6): Likewise.
17221         (*umulsidi3_nov6): Likewise.
17222         (*umulsidi3_v6): Likewise.
17223         (*umulsidi3adddi): Likewise.
17224         (*umulsidi3adddi_v6): Likewise.
17225         (*smulsi3_highpart_nov6): Likewise.
17226         (*smulsi3_highpart_v6): Likewise.
17227         (*umulsi3_highpart_nov6): Likewise.
17228         (*umulsi3_highpart_v6): Likewise.
17229         (mulhisi3): Likewise.
17230         (*mulhisi3tb): Likewise.
17231         (*mulhisi3bt): Likewise.
17232         (*mulhisi3tt): Likewise.
17233         (maddhisi4): Likewise.
17234         (*maddhisi4tb): Likewise.
17235         (*maddhisi4tt): Likewise.
17236         (maddhidi4): Likewise.
17237         (*maddhidi4tb): Likewise.
17238         (*maddhidi4tt): Likewise.
17239         (divsi3): Likewise.
17240         (udivsi3): Likewise.
17241         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
17242         (thumb2_mulsi_short_compare0): Likewise.
17243         (thumb2_mulsi_short_compare0_scratch): Likewise.
17244         * config/arm/arm1020e.md (1020mult1): Update attribute change.
17245         (1020mult2): Likewise.
17246         (1020mult3): Likewise.
17247         (1020mult4): Likewise.
17248         (1020mult5): Likewise.
17249         (1020mult6): Likewise.
17250         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
17251         change.
17252         (cortex_a15_mult64): Likewise.
17253         (cortex_a15_sdiv): Likewise.
17254         (cortex_a15_udiv): Likewise.
17255         * config/arm/arm1026ejs.md (mult1): Update attribute change.
17256         (mult2): Likewise.
17257         (mult3): Likewise.
17258         (mult4): Likewise.
17259         (mult5): Likewise.
17260         (mult6): Likewise.
17261         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
17262         (pj4_ir_div): Likewise.
17263         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
17264         (11_mult2): Likewise.
17265         (11_mult3): Likewise.
17266         (11_mult4): Likewise.
17267         (11_mult5): Likewise.
17268         (11_mult6): Likewise.
17269         (11_mult7): Likewise.
17270         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
17271         (cortex_a8_mla): Likewise.
17272         (cortex_a8_mull): Likewise.
17273         (cortex_a8_smulwy): Likewise.
17274         (cortex_a8_smlald): Likewise.
17275         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
17276         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
17277         (cortex_r4_mul_3): Likewise.
17278         (cortex_r4_mla_4): Likewise.
17279         (cortex_r4_mla_3): Likewise.
17280         (cortex_r4_smlald): Likewise.
17281         (cortex_r4_mull): Likewise.
17282         (cortex_r4_sdiv): Likewise.
17283         (cortex_r4_udiv): Likewise.
17284         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
17285         (cortex_a7_idiv): Likewise.
17286         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
17287         (9_mult2): Likewise.
17288         (9_mult3): Likewise.
17289         (9_mult4): Likewise.
17290         (9_mult5): Likewise.
17291         (9_mult6): Likewise.
17292         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
17293         (cortex_a53_sdiv): Likewise.
17294         (cortex_a53_udiv): Likewise.
17295         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
17296         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
17297         (mp626_mult2): Likewise.
17298         (mp626_mult3): Likewise.
17299         (mp626_mult4): Likewise.
17300         * config/arm/fa526.md (526_mult1): Update attribute change.
17301         (526_mult2): Likewise.
17302         * config/arm/arm-generic.md (mult): Update attribute change.
17303         (mult_ldsched_strongarm): Likewise.
17304         (mult_ldsched): Likewise.
17305         (multi_cycle): Likewise.
17306         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
17307         * config/arm/fa606te.md (606te_mult1): Update attribute change.
17308         (606te_mult2): Likewise.
17309         (606te_mult3): Likewise.
17310         (606te_mult4): Likewise.
17311         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
17312         (cortex_a9_mac16): Likewise.
17313         (cortex_a9_multiply): Likewise.
17314         (cortex_a9_mac): Likewise.
17315         (cortex_a9_multiply_long): Likewise.
17316         * config/arm/fa626te.md (626te_mult1): Update attribute change.
17317         (626te_mult2): Likewise.
17318         (626te_mult3): Likewise.
17319         (626te_mult4): Likewise.
17321 2013-06-18  Richard Biener  <rguenther@suse.de>
17323         PR lto/57334
17324         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
17326 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17328         PR target/57609
17329         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
17330         with next_active_insn.
17332 2013-06-18  Alan Modra  <amodra@gmail.com>
17334         * config/rs6000/rs6000.h (enum data_align): New.
17335         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
17336         (DATA_ABI_ALIGNMENT): Define.
17337         (CONSTANT_ALIGNMENT): Correct comment.
17338         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
17339         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
17341 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
17343         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
17344         ATTRIBUTE_UNUSED marking.
17346 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
17348         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
17349         alternative and update.
17350         (aarch64_dup_lanedi): Delete.
17351         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
17352         * config/aarch64/aarch64-simd-builtins.def: Update.
17354 2013-06-17  Richard Biener  <rguenther@suse.de>
17356         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
17357         (lto_input_scc): Declare.
17358         (lto_input_tree_1): Likewise.
17359         (struct lto_stats_d): Add num_tree_bodies_output and
17360         num_pickle_refs_output.
17361         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
17362         (lto_read_tree_1): Split out from ...
17363         (lto_read_tree): ... this.
17364         (lto_input_scc): New function.
17365         (lto_input_tree_1): Split out from ...
17366         (lto_input_tree): ... this.  Handle LTO_tree_scc.
17367         (lto_data_in_create): Create the streamer cache without hashes.
17368         * lto-streamer-out.c (create_output_block): Create the streamer
17369         cache with hashes when not doing WPA.
17370         (lto_write_tree_1): Split out from ...
17371         (lto_write_tree): ... this.
17372         (get_symbol_initial_value): New function.
17373         (lto_output_tree_1): Split out from ...
17374         (lto_output_tree): ... this.  Write trees as series of SCCs
17375         using a DFS walk via DFS_write_tree.
17376         (struct sccs, struct scc_entry): New types.
17377         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
17378         (DFS_write_tree_body): New function.
17379         (DFS_write_tree): Likewise.
17380         (hash_tree): Likewise.
17381         (scc_entry_compare): Likewise.
17382         (hash_scc): Likewise.
17383         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
17384         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
17385         TREE_CHAIN as regular reference.
17386         (streamer_read_integer_cst): Remove.
17387         (streamer_get_pickled_tree): Adjust.
17388         * tree-streamer-out.c (streamer_write_chain): Disable streaming
17389         of DECL_EXTERNALs in BLOCK_VARS for now.
17390         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
17391         reference.
17392         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
17393         Add hash value argument and record that if hashes are recorded
17394         in the cache.
17395         (streamer_tree_cache_insert_1): Adjust.
17396         (streamer_tree_cache_insert): Likewise.
17397         (streamer_tree_cache_insert_at): Rename to ...
17398         (streamer_tree_cache_replace_tree): ... this and adjust.
17399         (streamer_tree_cache_append): Adjust.
17400         (record_common_node): Likewise.
17401         (streamer_tree_cache_create): Add argument whether to
17402         record hash values together with trees.
17403         (streamer_tree_cache_delete): Adjust.
17404         * tree-streamer.h (struct streamer_tree_cache_d): Add
17405         vector of hashes.
17406         (streamer_read_integer_cst): Remove.
17407         (streamer_tree_cache_insert): Adjust.
17408         (streamer_tree_cache_append): Likewise.
17409         (streamer_tree_cache_insert_at): Rename to ...
17410         (streamer_tree_cache_replace_tree): ... this and adjust.
17411         (streamer_tree_cache_create): Add argument whether to record hashes.
17412         (streamer_tree_cache_get): Rename to ...
17413         (streamer_tree_cache_get_tree): ... this.
17414         (streamer_tree_cache_get_hash): New function.
17415         * tree.c (cache_integer_cst): New function.
17416         * tree.h (cache_integer_cst): Declare.
17417         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
17418         * lto-symtab.c (lto_varpool_replace_node): Only release
17419         DECL_INITIAL of non-prevailing decls.
17420         * varpool.c (varpool_remove_initializer): Do not release
17421         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
17423 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
17425         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
17426         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
17427         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
17428         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
17429         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
17430         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
17431         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
17432         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
17433         instead of TARGET_64BIT.
17434         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
17435         Require ISA_HAS_<D>DIV.
17437 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
17439         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
17440         (mips*-*-linux*): Move default with_llsc setting to where other
17441         defaults are set.
17442         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
17443         with_arch block.
17444         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
17445         Likewise.  Remove default with_tune setting.  Move default float
17446         setting to its own block.  Handle with_llsc in the same block as above.
17448 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
17450         PR rtl-optimization/57425
17451         PR rtl-optimization/57569
17452         * alias.c (write_dependence_p): Add new parameters mem_mode,
17453         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
17454         Changed all callers.
17455         (canon_anti_dependence): New function.
17456         * cse.c (check_dependence): Use canon_anti_dependence.
17457         * cselib.c (cselib_invalidate_mem): Likewise.
17458         * rtl.h (canon_anti_dependence): Declare.
17460 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
17462         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
17463         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
17464         ".set mips3" for 64-bit targets.
17466 2013-06-15  Dehao Chen  <dehao@google.com>
17468         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
17469         * gimple-low.c (gimple_check_call_matching_types): Likewise.
17470         (gimple_check_call_args): Likewise.
17471         * value-prof.c (check_ic_target): Likewise.
17472         * ipa-inline.c (early_inliner): Likewise.
17473         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
17474         * cgraph.c (cgraph_create_edge_1): Likewise.
17475         (cgraph_make_edge_direct): Likewise.
17477 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
17479         PR target/57615
17480         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
17481         rs6000_output_move_128bit to handle emitting quad memory
17482         operations.  Set attribute length to 8 bytes.
17484 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
17486         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
17487         New pattern.
17488         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
17489         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
17490         (aarch64_<su>mlsl<mode>): Likewise.
17492 2013-06-14  Mike Stump  <mikestump@comcast.net>
17494         * Makefile.in (TARGET_H): Add insn-codes.h.
17496 2013-06-14  Alan Modra  <amodra@gmail.com>
17498         PR middle-end/57134
17499         PR middle-end/57586
17500         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
17501         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
17502         bitfield expansion when EXPAND_MEMORY.
17503         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
17505 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
17507         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
17508         test for clearing quad memory on 32-bit later.
17510 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
17512         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
17513         (fold_negate_expr): Likewise.
17514         (fold_real_zero_addition_p): Handle vectors.
17515         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
17517 2013-06-14  Alan Modra  <amodra@gmail.com>
17519         * varasm.c (force_const_mem): Revert 2013-06-07 change.
17521 2013-06-13  Jan Hubicka  <jh@suse.cz>
17523         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
17524         Local comdats are not externally visible.
17525         * symtab.c (dump_symtab_base): Dump externally visible.
17526         (verify_symtab_base): Verify back links in the symtab hash.
17528 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
17530         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
17531         CONVERT_EXPR as equal nodes.
17533 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
17535         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
17537 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
17539         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
17540         Generalize to complex and vector.
17541         * tree.c (build_all_ones_cst): New function.
17542         * tree.h (build_all_ones_cst): Declare it.
17544 2013-06-13  Alan Modra  <amodra@gmail.com>
17546         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
17547         * config/rs6000/rs6000.md (signbittf2): New insn.
17548         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
17549         (abstf2_internal, cmptf_internal2): Likewise.
17550         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
17552 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
17553             Pat Haugen  <pthaugen@us.ibm.com>
17554             Peter Bergner  <bergner@vnet.ibm.com>
17556         * config/rs6000/rs6000.c (emit_load_locked): Add support for
17557         power8 byte, half-word, and quad-word atomic instructions.
17558         (emit_store_conditional): Likewise.
17559         (rs6000_expand_atomic_compare_and_swap): Likewise.
17560         (rs6000_expand_atomic_op): Likewise.
17562         * config/rs6000/sync.md (larx): Add new modes for power8.
17563         (stcx): Likewise.
17564         (AINT): New mode iterator to include TImode as well as normal
17565         integer modes on power8.
17566         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
17567         that VSX registers are not considered.  Use AINT mode iterator
17568         instead of INT1 to allow inclusion of quad word atomic operations
17569         on power8.
17570         (load_locked<mode>): Likewise.
17571         (store_conditional<mode>): Likewise.
17572         (atomic_compare_and_swap<mode>): Likewise.
17573         (atomic_exchange<mode>): Likewise.
17574         (atomic_nand<mode>): Likewise.
17575         (atomic_fetch_<fetchop_name><mode>): Likewise.
17576         (atomic_nand_fetch<mode>): Likewise.
17577         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
17578         each type.
17579         (ATOMIC): On power8, add QImode, HImode modes.
17580         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
17581         modes that promote to SImode.
17582         (load_lockedti): Convert TImode arguments to PTImode, so that we
17583         get a guaranteed even/odd register pair.
17584         (load_lockedpti): Likewise.
17585         (store_conditionalti): Likewise.
17586         (store_conditionalpti): Likewise.
17588         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
17589         atomic load/store instructions.
17590         (HSI): Likewise.
17592 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
17594         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
17595         loads.
17596         (insn_count): New attribute, with most cases extracted from...
17597         (length): ...here.  Redefine most cases in terms of insn_count.
17598         (single_insn): Delete.
17599         (can_delay): Use insn_count to check for single instructions.
17600         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
17601         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
17602         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
17603         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
17604         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
17605         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
17606         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
17607         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
17608         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
17609         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
17610         rather than "length".
17611         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
17612         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
17613         Use "insn_count" rather than "length".
17614         * config/mips/mips-dsp.md
17615         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
17616         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
17617         length attributes.
17619 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
17621         PR tree-optimization/57361
17622         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
17624 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
17626         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
17627         to split.
17628         (aarch64_simd_combine<mode>): New instruction expansion.
17629         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
17630         function prototype.
17631         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
17632         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
17634 2013-06-12  Jan Hubicka  <jh@suse.cz>
17636         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
17637         decl has when in streaming stage.
17638         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
17639         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
17641 2013-06-12  Roland Stigge  <stigge@antcom.de>
17643         PR target/57578
17644         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
17646 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
17648         PR tree-optimization/57537
17649         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
17650         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
17652 2013-06-12  Richard Biener  <rguenther@suse.de>
17654         * data-streamer.h (streamer_write_char_stream): CSE
17655         obs->current_pointer.
17656         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
17657         streamer_write_char_stream manually and optimize the resulting loop.
17658         (streamer_write_hwi_stream): Likewise.
17660 2013-06-12  Jan Hubicka  <jh@suse.cz>
17662         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
17663         * cgraph.h (varpool_create_empty_node): Declare.
17664         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
17665         duplicated nodes.
17666         * symtab.c (symtab_unregister_node): Be lax about missin entries
17667         in node hash.
17668         (symtab_get_node): Update comment.
17669         * varpool.c (varpool_create_empty_node): Break out from ...
17670         (varpool_node_for_decl): ... here.
17671         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
17673 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
17675         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
17676         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
17677         part.  Use straight-line flow at the end.
17678         <COMPONENT_REF>: Remove superfluous else.
17679         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
17681 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
17683         PR target/56564
17684         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
17685         target hook even for !TREE_PUBLIC decls.  If no resolution info
17686         is available, return false for common and external decls.
17688 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
17690         * config/rl78/constraints.md (U): New constraint.
17691         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
17692         valloc attribute.
17694 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
17696         PR target/57589
17697         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
17698         to allow returning address to AT_PLATFORM name.
17700 2013-06-11  Jan Hubicka  <jh@suse.cz>
17702         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
17703         * cgraph.h (symtab_node_base): Add weakref flag.
17704         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
17705         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
17706         (output_weakrefs): Use weakref flag.
17707         * fold-const.c (simple_operand_p): Handle WEAK.
17708         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
17709         * ipa.c (varpool_externally_visible_p): Drop weakref.
17710         (function_and_variable_visibility): Update comment; fix weakref
17711         sanity checks; do not clear DECL_WEAK on them.
17712         * lto-cgraph.c (lto_output_node): update.
17713         (lto_output_varpool_node): Update.
17714         (input_overwrite_node): Update.
17715         (input_node): Update.
17716         (input_varpool_node): Update.
17717         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
17718         (lto_symtab_merge_symbols): Add sanity check.
17719         (lto_symtab_prevailing_decl): Do not special case weakrefs.
17720         * passes.c (rest_of_decl_compilation): Set static flag, too.
17721         * symtab.c (dump_symtab_base): Dump weakref.
17722         (verify_symtab_base): Sanity check weakrefs.
17723         (symtab_make_decl_local): Remove duplicated code.
17724         (symtab_alias_ultimate_target): Simplify.
17725         * varpool.c (varpool_create_variable_alias): Set weakref flag.
17727 2013-06-11  Tom de Vries  <tom@codesourcery.com>
17729         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
17730         sequence_vect == NULL.
17732 2013-06-11  DJ Delorie  <dj@redhat.com>
17734         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
17735         (rl78_unwind_word_mode): New.
17737 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
17739         * final.c (debug_prefix_maps): Make static.
17741 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
17743         * function.c (initial_trampoline): Remove stray copy.
17745 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
17747         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
17749 2013-06-11  Martin Jambor  <mjambor@suse.cz>
17751         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
17752         within bounds at the beginning of the function.
17754 2013-06-11  Alan Modra  <amodra@gmail.com>
17756         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
17757         reporting.
17758         (get_named_section): Don't NULL !DECL_P decl.
17760 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
17762         * doc/invoke.texi (core-avx2): Document.
17763         (slm): Likewise.
17764         (atom): Updated with MOVBE.
17766 2013-06-11  Richard Biener  <rguenther@suse.de>
17768         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
17770 2013-06-11  Anton Blanchard  <anton@samba.org>
17772         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
17773         correct shift value in little-endian mode.
17775 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
17777         PR target/56564
17778         * varasm.c (get_variable_align): Move #endif to the right place.
17780 2013-06-10  Cary Coutant  <ccoutant@google.com>
17782         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
17783         for hash so that hash table traversal order is deterministic.
17785 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
17786             Pat Haugen  <pthaugen@us.ibm.com>
17787             Peter Bergner  <bergner@vnet.ibm.com>
17789         * config/rs6000/vector.md (GPR move splitter): Do not split moves
17790         of vectors in GPRS if they are direct moves or quad word load or
17791         store moves.
17793         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
17794         declaration.
17795         (direct_move_p): Likewise.
17796         (quad_load_store_p): Likewise.
17798         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
17799         classes into bins based on the physical register type.
17800         (reg_class_to_reg_type): Likewise.
17801         (IS_STD_REG_TYPE): Likewise.
17802         (IS_FP_VECT_REG_TYPE): Likewise.
17803         (reload_fpr_gpr): Arrays to determine what insn to use if we can
17804         use direct move instructions.
17805         (reload_gpr_vsx): Likewise.
17806         (reload_vsx_gpr): Likewise.
17807         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
17808         information that is a simplification of register classes.  Also
17809         precalculate direct move reload helpers.
17810         (direct_move_p): New function to return true if the operation can
17811         be done as a direct move instruciton.
17812         (quad_load_store_p): New function to return true if the operation
17813         is a quad memory operation.
17814         (rs6000_legitimize_address): If quad memory, only allow register
17815         indirect for TImode addresses.
17816         (rs6000_legitimate_address_p): Likewise.
17817         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
17818         (rs6000_reload_register_type): Likewise.
17819         (register_to_reg_type): Return register type.
17820         (rs6000_secondary_reload_simple_move): New helper function for
17821         secondary reload and secondary memory needed to identify anything
17822         that is a simple move, and does not need reloading.
17823         (rs6000_secondary_reload_direct_move): New helper function for
17824         secondary reload to identify cases that can be done with several
17825         instructions via the direct move instructions.
17826         (rs6000_secondary_reload_move): New helper function for secondary
17827         reload to identify moves between register types that can be done.
17828         (rs6000_secondary_reload): Add support for quad memory operations
17829         and for direct move.
17830         (rs6000_secondary_memory_needed): Likewise.
17831         (rs6000_debug_secondary_memory_needed): Change argument names.
17832         (rs6000_output_move_128bit): New function to return the move to
17833         use for 128-bit moves, including knowing about the various
17834         limitations of quad memory operations.
17836         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
17837         memory operations.  call rs6000_output_move_128bit for the actual
17838         instruciton(s) to generate.
17839         (vsx_movti_64bit): Likewise.
17841         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
17842         (UNSPEC_P8V_MTVSRWZ): Likewise.
17843         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
17844         (UNSPEC_P8V_MTVSRD): Likewise.
17845         (UNSPEC_P8V_XXPERMDI): Likewise.
17846         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
17847         (UNSPEC_FUSION_GPR): Likewise.
17848         (FMOVE128_GPR): New iterator for direct move.
17849         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
17850         (f32_sv): Likewise.
17851         (f32_dm): Likewise.
17852         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
17853         loads and direct move instructions.
17854         (zero_extendsidi2_lfiwzx): Likewise.
17855         (extendsidi2_lfiwax): Likewise.
17856         (extendsidi2_nocell): Likewise.
17857         (floatsi<mode>2_lfiwax): Likewise.
17858         (lfiwax): Likewise.
17859         (floatunssi<mode>2_lfiwzx): Likewise.
17860         (lfiwzx): Likewise.
17861         (fix_trunc<mode>_stfiwx): Likewise.
17862         (fixuns_trunc<mode>_stfiwx): Likewise.
17863         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
17864         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
17865         (parity<mode>2_cmpb): Set length/type attr.
17866         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
17867         for 'mr.' to fast_compare.
17868         (bpermd_<mode>): Change type attr to popcnt.
17869         (p8_fmrgow_<mode>): New insns for power8 direct move support.
17870         (p8_mtvsrwz_1): Likewise.
17871         (p8_mtvsrwz_2): Likewise.
17872         (reload_fpr_from_gpr<mode>): Likewise.
17873         (p8_mtvsrd_1): Likewise.
17874         (p8_mtvsrd_2): Likewise.
17875         (p8_xxpermdi_<mode>): Likewise.
17876         (reload_vsx_from_gpr<mode>): Likewise.
17877         (reload_vsx_from_gprsf): Likewise.
17878         (p8_mfvsrd_3_<mode>): LIkewise.
17879         (reload_gpr_from_vsx<mode>): Likewise.
17880         (reload_gpr_from_vsxsf): Likewise.
17881         (p8_mfvsrd_4_disf): Likewise.
17882         (multi-word GPR splits): Do not split direct moves or quad memory
17883         operations.
17885 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
17887         * tree-into-ssa.c (interesting_blocks): Make static.
17889 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
17891         PR target/56564
17892         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
17893         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
17894         Use DATA_ABI_ALIGNMENT for that case instead if defined.
17895         (get_variable_align): New function.
17896         (get_variable_section, emit_bss, emit_common,
17897         assemble_variable_contents, place_block_symbol): Use
17898         get_variable_align instead of DECL_ALIGN.
17899         (assemble_noswitch_variable): Add align argument, use it
17900         instead of DECL_ALIGN.
17901         (assemble_variable): Adjust caller.  Use get_variable_align
17902         instead of DECL_ALIGN.
17903         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
17904         caller.
17905         (DATA_ABI_ALIGNMENT): Define.
17906         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
17907         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
17908         opt is false, only return the psABI mandated alignment increase.
17909         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
17910         (DATA_ABI_ALIGNMENT): ... this.
17911         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
17912         (DATA_ABI_ALIGNMENT): ... this.
17913         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
17914         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
17915         (DATA_ABI_ALIGNMENT): ... this.
17916         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
17917         * doc/tm.texi: Regenerated.
17919 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
17921         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
17922         cmp_code to construct REG_EQUAL note.
17924 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
17926         PR target/57568
17927         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
17928         that operands[2] doesn't overlap with operands[0].
17930 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
17931             Jan Hubicka  <jh@suse.cz>
17933         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
17934         hack to mark symbols as used.
17936 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
17938         PR rtl-optimization/57559
17939         * lra-constraints.c (process_alt_operands): Don't discourage
17940         memory with known offset for offsetable memory constraint.
17941         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
17943 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
17945         * varasm.c (struct oc_local_state): Reorder fields.
17946         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
17947         and adjust accordingly.
17948         (output_constructor): Reorder initialization code and adjust call to
17949         output_constructor_bitfield.
17951 2013-06-07  Jan Hubicka  <jh@suse.cz>
17953         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
17955 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
17957         * tree-object-size.c (unknown): Make const.
17959 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17961         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
17962         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
17963         for last alternative in the cpu_facility attribute.
17965 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17967         PR target/56315
17968         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
17969         (xordi3): Change operand 2 constraint to arm_xordi_operand.
17970         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
17971         * config/arm/constraints.md (Dg): New constraint.
17972         * config/arm/neon.md (xordi3_neon): Remove.
17973         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
17974         * config/arm/predicates.md (arm_xordi_operand): New predicate.
17976 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17978         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
17979         Clean up alternatives.
17981 2013-06-07  Alan Modra  <amodra@gmail.com>
17983         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
17984         va_list_gpr_size.
17986 2013-06-07  Alan Modra  <amodra@gmail.com>
17988         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
17990 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17992         * config/arm/constraints.md (Df): New constraint.
17993         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
17994         Correct length attribute for last two alternatives.
17996 2013-06-07  Alan Modra  <amodra@gmail.com>
17998         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
17999         override user -mfp-in-toc.
18000         (offsettable_ok_by_alignment): Consider just the current access
18001         rather than the whole object, unless BLKmode.  Handle
18002         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
18003         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
18004         for -mcmodel=medium.
18005         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
18006         override user -mfp-in-toc or -msum-in-toc.  Default to
18007         -mno-fp-in-toc for -mcmodel=medium.
18009 2013-06-06  DJ Delorie  <dj@redhat.com>
18011         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
18012         TARGET_VALID_POINTER_MODE.
18014 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
18015             Pat Haugen  <pthaugen@us.ibm.com>
18016             Peter Bergner  <bergner@vnet.ibm.com>
18018         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
18019         Document new power8 builtins.
18021         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
18022         condition code register, to allow 128-bit logical operations to be
18023         done in the VSX or GPR registers.
18024         (nor<mode>3): Use the canonical form for nor.
18025         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
18026         vclz*, and vpopcnt* vector instructions.
18027         (nand<mode>3): Likewise.
18028         (orc<mode>3): Likewise.
18029         (clz<mode>2): LIkewise.
18030         (popcount<mode>2): Likewise.
18032         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
18033         that only the GPRs are recognized.
18035         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18036         support for new power8 builtins.
18038         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
18039         builtin functions.
18040         (xscvdpspn): Likewise.
18041         (vclz): Likewise.
18042         (vclzb): Likewise.
18043         (vclzh): Likewise.
18044         (vclzw): Likewise.
18045         (vclzd): Likewise.
18046         (vpopcnt): Likewise.
18047         (vpopcntb): Likewise.
18048         (vpopcnth): Likewise.
18049         (vpopcntw): Likewise.
18050         (vpopcntd): Likewise.
18051         (vgbbd): Likewise.
18052         (vmrgew): Likewise.
18053         (vmrgow): Likewise.
18054         (eqv): Likewise.
18055         (eqv_v16qi3): Likewise.
18056         (eqv_v8hi3): Likewise.
18057         (eqv_v4si3): Likewise.
18058         (eqv_v2di3): Likewise.
18059         (eqv_v4sf3): Likewise.
18060         (eqv_v2df3): Likewise.
18061         (nand): Likewise.
18062         (nand_v16qi3): Likewise.
18063         (nand_v8hi3): Likewise.
18064         (nand_v4si3): Likewise.
18065         (nand_v2di3): Likewise.
18066         (nand_v4sf3): Likewise.
18067         (nand_v2df3): Likewise.
18068         (orc): Likewise.
18069         (orc_v16qi3): Likewise.
18070         (orc_v8hi3): Likewise.
18071         (orc_v4si3): Likewise.
18072         (orc_v2di3): Likewise.
18073         (orc_v4sf3): Likewise.
18074         (orc_v2df3): Likewise.
18076         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
18077         allow power8 quad mode in 64-bit.
18078         (rs6000_builtin_vectorized_function): Add support to vectorize
18079         ISA 2.07 count leading zeros, population count builtins.
18080         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
18081         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
18082         (builtin_function_type): Add vgbbd builtin function which takes an
18083         unsigned argument.
18084         (altivec_expand_vec_perm_const): Add support for new power8 merge
18085         instructions.
18087         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
18088         that does not include TImdoe for use with 32-bit.
18089         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
18090         instructions.
18091         (UNSPEC_VSX_CVDPSPN): Likewise.
18092         (vsx_xscvdpspn): Likewise.
18093         (vsx_xscvspdpn): Likewise.
18094         (vsx_xscvdpspn_scalar): Likewise.
18095         (vsx_xscvspdpn_directmove): Likewise.
18096         (vsx_and<mode>3): Split logical operations into 32-bit and
18097         64-bit. Add support to do logical operations on TImode as well as
18098         VSX vector types.  Allow logical operations to be done in either
18099         VSX registers or in general purpose registers in 64-bit mode.  Add
18100         splitters if GPRs were used. For AND, add clobber of CCmode to
18101         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
18102         encoding.
18103         (vsx_and<mode>3_32bit): Likewise.
18104         (vsx_and<mode>3_64bit): Likewise.
18105         (vsx_ior<mode>3): Likewise.
18106         (vsx_ior<mode>3_32bit): Likewise.
18107         (vsx_ior<mode>3_64bit): Likewise.
18108         (vsx_xor<mode>3): Likewise.
18109         (vsx_xor<mode>3_32bit): Likewise.
18110         (vsx_xor<mode>3_64bit): Likewise.
18111         (vsx_one_cmpl<mode>2): Likewise.
18112         (vsx_one_cmpl<mode>2_32bit): Likewise.
18113         (vsx_one_cmpl<mode>2_64bit): Likewise.
18114         (vsx_nor<mode>3): Likewise.
18115         (vsx_nor<mode>3_32bit): Likewise.
18116         (vsx_nor<mode>3_64bit): Likewise.
18117         (vsx_andc<mode>3): Likewise.
18118         (vsx_andc<mode>3_32bit): Likewise.
18119         (vsx_andc<mode>3_64bit): Likewise.
18120         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
18121         and xxlorc instructions.
18122         (vsx_eqv<mode>3_64bit): Likewise.
18123         (vsx_nand<mode>3_32bit): Likewise.
18124         (vsx_nand<mode>3_64bit): Likewise.
18125         (vsx_orc<mode>3_32bit): Likewise.
18126         (vsx_orc<mode>3_64bit): Likewise.
18128         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
18130         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
18131         instruction.
18132         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
18133         (p8_vmrgow): Likewise.
18134         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
18135         GPRs to be split under VSX.
18136         (p8v_clz<mode>2): Add power8 count leading zero support.
18137         (p8v_popcount<mode>2): Add power8 population count support.
18138         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
18139         support.
18141         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
18142         instruction.
18144         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
18145         builtin functions.
18146         (vec_nand): Likewise.
18147         (vec_vclz): Likewise.
18148         (vec_vclzb): Likewise.
18149         (vec_vclzd): Likewise.
18150         (vec_vclzh): Likewise.
18151         (vec_vclzw): Likewise.
18152         (vec_vgbbd): Likewise.
18153         (vec_vmrgew): Likewise.
18154         (vec_vmrgow): Likewise.
18155         (vec_vpopcnt): Likewise.
18156         (vec_vpopcntb): Likewise.
18157         (vec_vpopcntd): Likewise.
18158         (vec_vpopcnth): Likewise.
18159         (vec_vpopcntw): Likewise.
18161 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
18163         PR rtl-optimization/57468
18164         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
18165         spilled pseudos.
18167 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
18169         PR rtl-optimization/57459
18170         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
18171         type when setting live regs.
18173 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
18175         * config/s390/s390.opt (mlra): New option.
18176         * config/s390/s390.c (s390_decompose_address): Check displacement
18177         for all registers for LRA.
18178         (s390_secondary_reload): Don't used secondary reloads for LRA.
18179         (s390_lra_p): New function.
18180         (TARGET_LRA_P): Define.
18181         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
18182         of attribute cpu_facility to zarch for the last alternative.
18183         (*cmpmem_short): Ditto.
18185 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
18187         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
18188         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
18189         (arm_expand_prologue): Likewise.
18191 2013-06-06  Teresa Johnson  <tejohnson@google.com>
18193         PR c++/53743
18194         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
18195         as this is now done by redirect_edge_and_branch_force.
18196         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
18197         barriers, and fix interaction with splitting.
18198         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
18199         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
18200         reflect changes made in the routine.
18201         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
18202         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
18203         since this is called in cfglayout mode, and replace partition fixup
18204         with assert as that is now done by force_nonfallthru_and_redirect.
18205         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
18206         already be marked with region crossing note.
18207         (insert_section_boundary_note): Make non-static, gate on flag
18208         has_bb_partition, rewrite to also check for multiple partitions.
18209         (rest_of_handle_reorder_blocks): Remove call to
18210         insert_section_boundary_note, now done later during free_cfg.
18211         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
18212         * bb-reorder.h (insert_section_boundary_note): Declare.
18213         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
18214         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
18215         invoke insert_section_boundary_note.
18216         (try_redirect_by_replacing_jump): Remove unnecessary
18217         check for region crossing note.
18218         (fixup_partition_crossing): New function.
18219         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
18220         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
18221         in non-cfglayout mode.
18222         (force_nonfallthru_and_redirect): Fixup partition boundaries,
18223         remove old code that tried to do this. Emit barrier correctly
18224         when we are in cfglayout mode.
18225         (last_bb_in_partition): New function.
18226         (rtl_split_edge): Correctly fixup partition boundaries.
18227         (commit_one_edge_insertion): Remove old code that tried to
18228         fixup region crossing edge since this is now handled in
18229         split_block, and set up insertion point correctly since
18230         block may now end in a jump.
18231         (verify_hot_cold_block_grouping): Guard against checking when not in
18232         linearized RTL mode.
18233         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
18234         notes.
18235         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
18236         rtl_verify_flow_info, so not called in cfglayout mode.
18237         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
18238         (fixup_reorder_chain): Remove old code that attempted to fixup region
18239         crossing note as this is now handled in force_nonfallthru_and_redirect.
18240         (duplicate_insn_chain): Don't duplicate switch section notes.
18241         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
18242         note.
18243         * basic-block.h (emit_barrier_after_bb): Declare.
18245 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18247         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
18248         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
18249         arm_usatsihi): Adjust alternatives for arm_restrict_it.
18251 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18253         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
18254         where appropriate.
18255         * config/arm/ldmstm.md: Regenerate.
18257 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18259         * config/arm/sync.md (atomic_loaddi_1):
18260         Disable predication for arm_restrict_it.
18261         (arm_load_exclusive<mode>): Likewise.
18262         (arm_load_exclusivesi): Likewise.
18263         (arm_load_exclusivedi): Likewise.
18264         (arm_load_acquire_exclusive<mode>): Likewise.
18265         (arm_load_acquire_exclusivesi): Likewise.
18266         (arm_load_acquire_exclusivedi): Likewise.
18267         (arm_store_exclusive<mode>): Likewise.
18268         (arm_store_exclusive<mode>): Likewise.
18269         (arm_store_release_exclusivedi): Likewise.
18270         (arm_store_release_exclusive<mode>): Likewise.
18272 2013-06-06  Richard Biener  <rguenther@suse.de>
18274         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
18275         after LTO_null.
18276         (lto_tag_is_tree_code_p): Adjust.
18277         (lto_tag_is_gimple_code_p): Likewise.
18278         (lto_gimple_code_to_tag): Likewise.
18279         (lto_tag_to_gimple_code): Likewise.
18280         (lto_tree_code_to_tag): Likewise.
18281         (lto_tag_to_tree_code): Likewise.
18282         * data-streamer.h (streamer_write_hwi_in_range): Use
18283         uhwi streaming to stream the normalized range.
18284         (streamer_read_hwi_in_range): Likewise.
18286 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18288         * config/arm/arm.md (enabled_for_depr_it): New attribute.
18289         (predicable_short_it): Likewise.
18290         (predicated): Likewise.
18291         (enabled): Handle above.
18292         (define_cond_exec): Set predicated attribute to yes.
18294 2013-06-05  Mike Stump  <mikestump@comcast.net>
18296         * gdbinit.in (__FUNCTION__): Add.
18298 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
18300         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
18301         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
18303 2013-06-05  Jan Hubicka  <jh@suse.cz>
18305         * varasm.c (mark_decl_referenced): Revert the removal until targets
18306         are fixed.
18308 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
18310         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
18311         instead of mark_decl_referenced.
18313 2013-06-05  Jan Hubicka  <jh@suse.cz>
18315         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
18316         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
18317         and symtab_used_from_object_file_p.
18318         (cgraph_make_node_local_1): Clear forced_by_abi.
18319         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
18320         * cgraph.h (symtab_node_base): Add forced_by_abi.
18321         (decide_is_variable_needed): Remove.
18322         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
18323         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
18324         (decide_is_symbol_needed): ... this one; handle symbols in general;
18325         always analyze virtuals; honnor forced_by_abi.
18326         (cgraph_finalize_function): Update.
18327         (varpool_finalize_decl): Update.
18328         (symbol_defined_and_needed): Remove.
18329         (analyze_functions): Update.
18330         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
18331         output_refs, input_overwrite_node): Handle forced_by_abi.
18332         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
18333         (address_taken_from_non_vtable_p): ... this one.
18334         (comdat_can_be_unshared_p_1): New function.
18335         (cgraph_comdat_can_be_unshared_p): Rename to ...
18336         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
18337         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
18338         (function_and_variable_visibility): Clear forced_by_abi as needed.
18339         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
18340         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
18341         * symtab.c (dump_symtab_base): Dump forced_by_abi.
18342         * varpool.c (decide_is_variable_needed): Remove.
18344 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18346         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
18347         (arm_option_override): Override arm_restrict_it where appropriate.
18348         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
18349         * config/arm/arm.opt (mrestrict-it): New command-line option.
18350         * doc/invoke.texi: Document -mrestrict-it.
18352 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
18354         * tsan.c (tsan_atomic_table): Make const.
18356 2013-06-05  Richard Biener  <rguenther@suse.de>
18358         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
18359         index associated with the tree we are supposed to replace.
18360         * tree-streamer-out.c (pack_ts_base_value_fields): Output
18361         TREE_ASM_WRITTEN as zero for everything but SSA names.
18363 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
18365         * tree-ssa-structalias.c (call_stmt_vars): Make static.
18367 2013-06-04  Jan Hubicka  <jh@suse.cz>
18369         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
18370         (input_node, input_varpool_node): Handle correctly external same
18371         body aliases.
18372         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
18373         nodes at ltrans stage.
18375 2013-06-04  Jan Hubicka  <jh@suse.cz>
18377         * ipa-inline.c (update_caller_keys): Fix availability test.
18378         (update_callee_keys): Likewise.
18379         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
18380         to follow ELF standard.
18382 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
18384         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
18385         (mips64r5900el-*-elf*): New configurations.
18386         * config/mips/mips-cpus.def (r5900): New processor.
18387         * config/mips/mips-tables.opt: Regenerate.
18388         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
18389         (mips_issue_rate): Handle PROCESSOR_R5900.
18390         (mips_reorg_process_insns): Force reorder mode for the R5900.
18391         * config/mips/mips.h (TARGET_MIPS5900): Define.
18392         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
18393         TARGET_MIPS5900.
18394         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
18395         TARGET_MIPS5900.
18396         * config/mips/mips.md (processor): Add r5900.
18397         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
18399 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
18401         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
18402         into function to generate MOVI instruction.
18403         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
18404         (aarch64_preferred_simd_mode): Turn into wrapper.
18405         (aarch64_output_scalar_simd_mov_immediate): New function.
18406         * config/aarch64/aarch64-protos.h: Add prototype for above.
18408 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
18410         * config/aarch64/aarch64.c (simd_immediate_info): Remove
18411         element_char member.
18412         (sizetochar): Return signed char.
18413         (aarch64_simd_valid_immediate): Remove elchar and other
18414         unnecessary variables.
18415         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
18416         Calculate element_char as required.
18417         * config/aarch64/aarch64-protos.h: Update and move prototype
18418         for aarch64_output_simd_mov_immediate.
18419         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
18420         Update arguments.
18422 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
18424         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
18425         information completed by aarch64_simd_valid_immediate.
18426         (aarch64_legitimate_constant_p): Update arguments.
18427         (aarch64_simd_valid_immediate): Work with struct rather than many
18428         pointers.
18429         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
18430         (aarch64_simd_make_constant): Update arguments.
18431         (aarch64_output_simd_mov_immediate): Work with struct rather than
18432         many pointers.  Output immediate directly rather than as operand.
18433         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
18434         Update prototype.
18435         * config/aarch64/constraints.md (Dn): Update arguments.
18437 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
18439         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
18440         longer static.
18441         (aarch64_simd_immediate_valid_for_move): Remove.
18442         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
18443         (aarch64_simd_make_constant): Update call.
18444         (aarch64_output_simd_mov_immediate): Update call.
18445         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
18446         Add prototype.
18447         * config/aarch64/constraints.md (Dn): Update call.
18449 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
18451         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
18452         return type to bool for prototype.
18453         (aarch64_legitimate_constant_p): Check for true instead of not -1.
18454         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
18455         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
18457 2013-06-04  Catherine Moore  <clm@codesourcery.com>
18459         * config/mips/mips.opt (meva): New.
18460         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
18461         (ASM_SPEC): Handle -meva.
18462         * doc/invoke.texi (meva):  Document.
18464 2013-06-04  Alan Modra  <amodra@gmail.com>
18466         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
18467         constant output.
18469 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18471         * rtl.def: Add extra fourth optional field to define_cond_exec.
18472         * gensupport.c (process_one_cond_exec): Process attributes from
18473         define_cond_exec.
18474         * doc/md.texi: Document fourth field in define_cond_exec.
18476 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
18478         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
18479         out the processing order as in store_bit_field_1.
18481 2013-06-04  Jan Hubicka  <jh@suse.cz>
18483         PR middle-end/57500
18484         * cgraphunit.c (cgraph_process_same_body_aliases): Create
18485         non-VAR_DECL node if it does not exist yet.
18487 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
18489         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
18490         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
18491         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
18492         target_cpu_default setting.
18494 2013-06-03  Teresa Johnson  <tejohnson@google.com>
18496         * dumpfile.c (opt_info_switch_p): Change -fopt-info
18497         default to -fopt-info=optimized instead of all.
18498         * doc/invoke.texi: Ditto.
18499         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
18500         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
18501         (execute_vect_slp): Emit BB vectorization success under
18502         MSG_OPTIMIZED_LOCATIONS.
18503         * tree-vect-slp.c (vect_slp_transform_bb): Change
18504         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
18505         * tree-vect-loop.c (vect_transform_loop): Ditto.
18507 2013-06-03  Jason Merrill  <jason@redhat.com>
18509         PR c++/57415
18510         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18511         Use TARGET_EXPR for C++.
18513 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
18515         PR rtl-optimization/57268
18516         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
18517         if DEBUG_INSN_P (insn).
18519         Reapply
18520         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
18522         PR rtl-optimization/57268
18523         * sched-deps.c (sched_analyze_2): Flush dependence lists if
18524         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
18526 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
18528         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
18529         (ix86_avoid_lea_for_addr): Likewise.
18530         (exact_dependency_1): Likewise.
18531         (ix86_adjust_cost): Likewise.
18532         (swap_top_of_ready_list): Fix formatting and !reload_completed check
18533         removed.
18534         (do_reorder_for_imul): Fix typo, formatting and
18535         !reload_completed check removed.
18536         (ix86_sched_reorder): Fix typo and formatting.
18537         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
18538         list.
18540 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
18542         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
18544 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
18546         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
18547         <STRING_CST>: Likewise.
18548         <VECTOR_CST>: Likewise.
18550 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
18551             Mikael Morin  <mikael@gcc.gnu.org>
18553         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
18554         * config.in: Regenerated.
18555         * configure: Regenerated.
18557 2013-06-01  Jan Hubicka  <jh@suse.cz>
18559         PR middle-end/57366
18560         * cgraphunit.c (compile): When weakref is not supported,
18561         set up transparent aliases before final output pass.
18562         * varasm.c (assemble_alias): Do not try to do it here.
18564 2013-06-01  Jan Hubicka  <jh@suse.cz>
18566         PR middle-end/57467
18567         * passes.c (for_per_function): Skip unanalyzed functions.
18569 2013-06-01  Jan Hubicka  <jh@suse.cz>
18571         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
18572         (lto_symtab_merge_symbols_1): ... this one.
18573         (lto_symtab_merge_cgraph_nodes): Rename to ...
18574         (lto_symtab_merge_symbols): ... this one; simplify.
18575         * cgraph.c (same_body_aliases_done): Rename to ...
18576         (cpp_implicit_aliases_done): ... this one.
18577         (cgraph_create_function_alias): Update.
18578         (cgraph_same_body_alias): Update.
18579         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
18580         (verify_edge_corresponds_to_fndecl): Simplify.
18581         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
18582         (cgraph_node): Remove same_body_alias.
18583         (varpool_node): Remove alias_of and extra_name_alias.
18584         (same_body_aliases_done): Rename to ..
18585         (cpp_implicit_aliases_done): ... this one.
18586         (symtab_alias_ultimate_target): Add default parameter.
18587         (symtab_resolve_alias): New function.
18588         (fixup_same_cpp_alias_visibility): Declare.
18589         (cgraph_function_node): Add default parameter.
18590         (cgraph_node_asm_name): Likewise.
18591         (cgraph_function_or_thunk_node): Add default parameter; do
18592         not ICE when it is NULL.
18593         (varpool_variable_node): Likewise.
18594         * tree-emutls.c (create_emultls_var): Update.
18595         (ipa_lower_emutls): Update.
18596         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
18597         (cgraph_reset_node): Reset alias info.
18598         (cgraph_finalize_function): Update.
18599         (fixup_same_cpp_alias_visibility): Move to symtab.c.
18600         (analyze_function): Simplify.
18601         (cgraph_process_same_body_aliases): Simplify.
18602         (analyze_functions): Fixup same body aliases.
18603         (handle_alias_pairs): Simplify.
18604         (assemble_thunk): Update.
18605         (assemble_thunks_and_aliases): Update.
18606         (output_weakrefs): Rewrite.
18607         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
18608         (lto_output_varpool_node): Likewise.
18609         (compute_ltrans_boundary): Remve assert.
18610         (get_alias_symbol): New functoin.
18611         (input_node): Rewrite alias handling.
18612         (input_varpool_node): Likewise.
18613         * ipa-pure-const.c (propagate_pure_const): Fix formating.
18614         * ipa.c (process_references): Handle weakrefs correctly.
18615         (symtab_remove_unreachable_nodes): Likewise.
18616         * trans-mem.c (get_cg_data): Update.
18617         (ipa_tm_create_version_alias): Update.
18618         (ipa_tm_execute): Update.
18619         * symtab.c (dump_symtab_base): Dump aliases.
18620         (verify_symtab_base): Verify aliases.
18621         (symtab_node_availability): New function.
18622         (symtab_alias_ultimate_target): Simplify.
18623         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
18624         handle all the fixup cases.
18625         (symtab_resolve_alias): New function.
18626         * passes.c (ipa_write_summaries): Handle weakrefs.
18627         * varpool.c (varpool_analyze_node): Simplify.
18628         (assemble_aliases): Update.
18629         (varpool_create_variable_alias): Simplify.
18630         (varpool_extra_name_alias): Simplify.
18631         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
18632         (lto_symtab_merge_symbols): ... this one.
18634 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
18636         Revert
18637         PR rtl-optimization/57268
18638         * sched-deps.c (sched_analyze_2): Flush dependence lists if
18639         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
18641 2013-06-01  Tobias Burnus  <burnus@net-b.de>
18643         Partially reverted:
18644         2013-05-31  Tobias Burnus  <burnus@net-b.de>
18646         PR middle-end/57073
18647         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
18648         further up.
18650 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
18652         PR rtl-optimization/57268
18653         * sched-deps.c (sched_analyze_2): Flush dependence lists if
18654         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
18656 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
18658         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
18659         unordered comparison operators when -fno-trapping-math is in effect
18660         on the e500.
18661         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
18662         and implement unordered comparison operators properly on the e500.
18664 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
18666         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
18667         for constant scalar integers.
18668         (simplify_relational_operation_1): Likewise.
18670 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
18672         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
18673         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
18674         Fix comment.
18676 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
18677             Igor Zamyatin  <igor.zamyatin@intel.com>
18679         Silvermont (SLM) architecture performance tuning.
18680         * config/i386/i386.h (enum ix86_tune_indices): Add
18681         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
18682         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
18684         * config/i386/i386.c (initial_ix86_tune_features)
18685         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
18686         (ix86_lea_outperforms): Handle Silvermont tuning.
18687         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
18688         call.
18689         (ix86_use_lea_for_mov): Likewise.
18690         (ix86_avoid_lea_for_addr): Likewise.
18691         (ix86_lea_for_add_ok): Likewise.
18692         (exact_dependency_1): New function.
18693         (exact_store_load_dependency): Likewise.
18694         (ix86_adjust_cost): Handle Silvermont tuning.
18695         (do_reoder_for_imul): Likewise.
18696         (swap_top_of_ready_list): New function.
18697         (ix86_sched_reorder): Changed to handle Silvermont tuning.
18699         * config/i386/i386.md (peepholes that split memory operand in fp
18700         converts): New.
18702 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
18704         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
18705         Remove un-necessary braces.
18707 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
18709         * config/aarch64/aarch64.c (aarch64_classify_symbol):
18710         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
18712 2013-05-31  Tobias Burnus  <burnus@net-b.de>
18714         PR middle-end/57073
18715         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
18717 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18719         PR target/56315
18720         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
18721         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
18722         * config/arm/neon.md (iordi3_neon): Remove.
18723         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
18724         * config/arm/predicates.md (imm_for_neon_logic_operand):
18725         Move to earlier in the file.
18726         (neon_logic_op2): Likewise.
18727         (arm_iordi_operand_neon): New predicate.
18729 2013-05-31  Richard Biener  <rguenther@suse.de>
18731         PR tree-optimization/57478
18732         PR tree-optimization/57453
18733         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
18734         are life as well.
18736 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
18738         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
18739         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
18741 2013-05-30  Tobias Burnus  <burnus@net-b.de>
18742             Thomas Koenig  <tkoenig@gcc.gnu.org>
18744         PR middle-end/57073
18745         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
18746         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
18748 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
18750         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
18752 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
18754         * target.def (register_usage_leveling_p): New hook.
18755         * targhooks.c (default_register_usage_leveling_p): New.
18756         * targhooks.h (default_register_usage_leveling_p): New prototype.
18757         * lra-assigns.c (register_usage_leveling_p): Use the hook.
18758         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
18759         * doc/tm.texi: Update.
18760         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
18762 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
18764         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
18765         (*insv_reg<mode>): New define_insn.
18767 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
18769         PR rtl-optimization/57439
18770         * postreload.c (move2add_valid_value_p): Check that we have
18771         a zero subreg_regno_offset when accessing the register in
18772         the requested mode.
18774 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
18775             Igor Zamyatin  <igor.zamyatin@intel.com>
18777         Silvermont (SLM) architecture pipeline model, tuning and
18778         insn selection.
18779         * config.gcc: Add slm config options and target.
18781         * config/i386/slm.md: New.
18783         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
18785         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
18786         PROCESSOR_SLM.
18787         (ix86_target_macros_internal): Likewise.
18789         * config/i386/i386.c (slm_cost): New cost.
18790         (m_SLM): New macro flag.
18791         (initial_ix86_tune_features): Set m_SLM.
18792         (x86_accumulate_outgoing_args): Likewise.
18793         (x86_arch_always_fancy_math_387): Likewise.
18794         (processor_target_table): Add slm cost.
18795         (cpu_names): Add slm cpu name.
18796         (x86_option_override_internal): Set SLM ISA.
18797         (ix86_issue_rate): New case PROCESSOR_SLM.
18798         (ia32_multipass_dfa_lookahead): Likewise.
18799         (fold_builtin_cpu): Add slm.
18801         * config/i386/i386.h (TARGET_SLM): New target macro.
18802         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
18803         (processor_type): Add PROCESSOR_SLM.
18805         * config/i386/i386.md (cpu): Add new value "slm".
18806         (slm.md): Include slm.md.
18808 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
18809             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
18811         * config/arm/arm-protos.h: Add and update function protos.
18812         * config/arm/arm.c (use_simple_return_p): New added.
18813         (thumb2_expand_return): Check simple_return flag.
18814         * config/arm/arm.md: Add simple_return and conditional simple_return.
18815         * config/arm/iterators.md: Add iterator for return and simple_return.
18817 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
18819         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
18820         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
18821         (arm_emit_vfp_multi_reg_pop): Likewise.
18822         (thumb2_emit_ldrd_pop): Likewise.
18823         (arm_expand_epilogue): Add misc REG_CFA notes.
18824         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
18826 2013-05-29  Lawrence Crowl  <crowl@google.com>
18828         * config/arm/t-arm: Update for below.
18830         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
18831         Change type to hash_table.  Update dependent calls and types.
18833         * config/i386/t-cygming: Update for below.
18835         * config/i386/t-interix: Update for below.
18837         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
18838         Change type to hash_table.  Update dependent calls and types.
18839         (i386_find_on_wrapper_list::wrappers): Likewise.
18841         * config/ia64/t-ia64: Update for below.
18843         * config/ia64/ia64.c (bundle_state_table):
18844         Change type to hash_table.  Update dependent calls and types.
18846         * config/mips/mips.c (mips_reorg_process_insns::htab):
18847         Change type to hash_table.  Update dependent calls and types.
18849         * config/sol2.c (solaris_comdat_htab):
18850         Change type to hash_table.  Update dependent calls and types.
18852         * config/t-sol2: Update for above.
18854 2013-05-29  Teresa Johnson  <tejohnson@google.com>
18856         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
18857         functions are not yet marked as defined.
18859 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
18860             Pat Haugen  <pthaugen@us.ibm.com>
18861             Peter Bergner  <bergner@vnet.ibm.com>
18863         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
18864         instructions.
18865         (VEC_A): Likewise.
18866         (VEC_C): Likewise.
18867         (vrotl<mode>3): Likewise.
18868         (vashl<mode>3): Likewise.
18869         (vlshr<mode>3): Likewise.
18870         (vashr<mode>3): Likewise.
18872         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18873         support for power8 V2DI builtins.
18875         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
18876         power8 V2DI builtins.
18877         (vupkhsw): Likewise.
18878         (vupklsw): Likewise.
18879         (vaddudm): Likewise.
18880         (vminsd): Likewise.
18881         (vmaxsd): Likewise.
18882         (vminud): Likewise.
18883         (vmaxud): Likewise.
18884         (vpkudum): Likewise.
18885         (vpksdss): Likewise.
18886         (vpkudus): Likewise.
18887         (vpksdus): Likewise.
18888         (vrld): Likewise.
18889         (vsld): Likewise.
18890         (vsrd): Likewise.
18891         (vsrad): Likewise.
18892         (vsubudm): Likewise.
18893         (vcmpequd): Likewise.
18894         (vcmpgtsd): Likewise.
18895         (vcmpgtud): Likewise.
18896         (vcmpequd_p): Likewise.
18897         (vcmpgtsd_p): Likewise.
18898         (vcmpgtud_p): Likewise.
18899         (vupkhsw): Likewise.
18900         (vupklsw): Likewise.
18901         (vaddudm): Likewise.
18902         (vmaxsd): Likewise.
18903         (vmaxud): Likewise.
18904         (vminsd): Likewise.
18905         (vminud): Likewise.
18906         (vpksdss): Likewise.
18907         (vpksdus): Likewise.
18908         (vpkudum): Likewise.
18909         (vpkudus): Likewise.
18910         (vrld): Likewise.
18911         (vsld): Likewise.
18912         (vsrad): Likewise.
18913         (vsrd): Likewise.
18914         (vsubudm): Likewise.
18916         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
18917         support for power8 V2DI instructions.
18919         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
18920         power8 V2DI instructions.  Combine pack and unpack insns to use an
18921         iterator for each mode.  Check whether a particular mode supports
18922         Altivec instructions instead of just checking TARGET_ALTIVEC.
18923         (UNSPEC_VPKUWUM): Likewise.
18924         (UNSPEC_VPKSHSS): Likewise.
18925         (UNSPEC_VPKSWSS): Likewise.
18926         (UNSPEC_VPKUHUS): Likewise.
18927         (UNSPEC_VPKSHUS): Likewise.
18928         (UNSPEC_VPKUWUS): Likewise.
18929         (UNSPEC_VPKSWUS): Likewise.
18930         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
18931         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
18932         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
18933         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
18934         (UNSPEC_VUPKHSB): Likewise.
18935         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
18936         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
18937         (UNSPEC_VUPKHSH): Likewise.
18938         (UNSPEC_VUPKLSB): Likewise.
18939         (UNSPEC_VUPKLSH): Likewise.
18940         (VI2): Likewise.
18941         (VI_char): Likewise.
18942         (VI_scalar): Likewise.
18943         (VI_unit): Likewise.
18944         (VP): Likewise.
18945         (VP_small): Likewise.
18946         (VP_small_lc): Likewise.
18947         (VU_char): Likewise.
18948         (add<mode>3): Likewise.
18949         (altivec_vaddcuw): Likewise.
18950         (altivec_vaddu<VI_char>s): Likewise.
18951         (altivec_vadds<VI_char>s): Likewise.
18952         (sub<mode>3): Likewise.
18953         (altivec_vsubcuw): Likewise.
18954         (altivec_vsubu<VI_char>s): Likewise.
18955         (altivec_vsubs<VI_char>s): Likewise.
18956         (altivec_vavgs<VI_char>): Likewise.
18957         (altivec_vcmpbfp): Likewise.
18958         (altivec_eq<mode>): Likewise.
18959         (altivec_gt<mode>): Likewise.
18960         (altivec_gtu<mode>): Likewise.
18961         (umax<mode>3): Likewise.
18962         (smax<mode>3): Likewise.
18963         (umin<mode>3): Likewise.
18964         (smin<mode>3): Likewise.
18965         (altivec_vpkuhum): Likewise.
18966         (altivec_vpkuwum): Likewise.
18967         (altivec_vpkshss): Likewise.
18968         (altivec_vpkswss): Likewise.
18969         (altivec_vpkuhus): Likewise.
18970         (altivec_vpkshus): Likewise.
18971         (altivec_vpkuwus): Likewise.
18972         (altivec_vpkswus): Likewise.
18973         (altivec_vpks<VI_char>ss): Likewise.
18974         (altivec_vpks<VI_char>us): Likewise.
18975         (altivec_vpku<VI_char>us): Likewise.
18976         (altivec_vpku<VI_char>um): Likewise.
18977         (altivec_vrl<VI_char>): Likewise.
18978         (altivec_vsl<VI_char>): Likewise.
18979         (altivec_vsr<VI_char>): Likewise.
18980         (altivec_vsra<VI_char>): Likewise.
18981         (altivec_vsldoi_<mode>): Likewise.
18982         (altivec_vupkhsb): Likewise.
18983         (altivec_vupkhs<VU_char>): Likewise.
18984         (altivec_vupkls<VU_char>): Likewise.
18985         (altivec_vupkhsh): Likewise.
18986         (altivec_vupklsb): Likewise.
18987         (altivec_vupklsh): Likewise.
18988         (altivec_vcmpequ<VI_char>_p): Likewise.
18989         (altivec_vcmpgts<VI_char>_p): Likewise.
18990         (altivec_vcmpgtu<VI_char>_p): Likewise.
18991         (abs<mode>2): Likewise.
18992         (vec_unpacks_hi_v16qi): Likewise.
18993         (vec_unpacks_hi_v8hi): Likewise.
18994         (vec_unpacks_lo_v16qi): Likewise.
18995         (vec_unpacks_hi_<VP_small_lc>): Likewise.
18996         (vec_unpacks_lo_v8hi): Likewise.
18997         (vec_unpacks_lo_<VP_small_lc>): Likewise.
18998         (vec_pack_trunc_v8h): Likewise.
18999         (vec_pack_trunc_v4si): Likewise.
19000         (vec_pack_trunc_<mode>): Likewise.
19002         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
19003         V2DI builtins.
19004         (vec_vmaxsd): Likewise.
19005         (vec_vmaxud): Likewise.
19006         (vec_vminsd): Likewise.
19007         (vec_vminud): Likewise.
19008         (vec_vpksdss): Likewise.
19009         (vec_vpksdus): Likewise.
19010         (vec_vpkudum): Likewise.
19011         (vec_vpkudus): Likewise.
19012         (vec_vrld): Likewise.
19013         (vec_vsld): Likewise.
19014         (vec_vsrad): Likewise.
19015         (vec_vsrd): Likewise.
19016         (vec_vsubudm): Likewise.
19017         (vec_vupkhsw): Likewise.
19018         (vec_vupklsw): Likewise.
19020 2013-05-29  Jan Hubicka  <jh@suse.cz>
19022         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
19023         flags; reorder rest of fields in more consistent way.
19024         (varpool_node): Remove analyzed, finalized and alias.
19025         (cgraph_ndoe): Likewise.
19026         (symtab_alias_ultimate_target): New function.
19027         (cgraph_function_node): Move offline.
19028         (cgraph_reset_node): Declare.
19029         (cgraph_comdat_can_be_unshared_p): Remove.
19030         (varpool_remove_initializer): Declare.
19031         (varpool_first_defined_variable, varpool_next_defined_variable
19032         cgraph_first_defined_function, cgraph_next_defined_function): Update.
19033         (cgraph_function_with_gimple_body_p): Update.
19034         (varpool_all_refs_explicit_p): Update.
19035         (symtab_alias_target): New function.
19036         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
19037         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
19038         (cgraph_function_or_thunk_node): Simplify using
19039         symtab_alias_ultimate_target.
19040         (varpool_variable_node): Likewise.
19041         * cgraph.c (cgraph_create_function_alias): Update.
19042         (cgraph_add_thunk): Update.
19043         (cgraph_remove_node): Update.
19044         (dump_cgraph_node): Do not dump removed flags.
19045         (cgraph_function_body_availability): Update.
19046         (cgraph_propagate_frequency): Update.
19047         (verify_cgraph_node): Check sanity of local flag.
19048         (cgraph_function_node): Move here from cgraph.h; revamp for
19049         cgraph_function_or_thunk_node.
19050         * lto-symtab.c (lto_varpool_replace_node): Update.
19051         (lto_symtab_resolve_can_prevail_p): Update.
19052         (lto_symtab_merge_cgraph_nodes): Update.
19053         * ipa-cp.c (determine_versionability, initialize_node_lattices,
19054         propagate_constants_accross_call, devirtualization_time_bonus,
19055         ipcp_propagate_stage): Update.
19056         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
19057         * ipa-inline-transform.c (clone_inlined_nodes,
19058         preserve_function_body_p): Update.
19059         * ipa-reference.c (propagate): Update.
19060         (write_node_summary_p): Update.
19061         * toplev.c (wrapup_global_declaration_2): Update.
19062         * cgraphunit.c (cgraph_analyze_function): Rename to ...
19063         (analyze_function) ... this one.
19064         (cgraph_process_new_functions): Update.
19065         (cgraph_reset_node): Export.
19066         (cgraph_finalize_function): Update.
19067         (cgraph_add_new_function): Update.
19068         (process_function_and_variable_attributes): Update.
19069         (varpool_finalize_decl): Update.
19070         (symbol_finalized): Remove.
19071         (symbol_finalized_and_needed): Rename to ...
19072         (symbol_defined_and_needed): ... update.
19073         (cgraph_analyze_functions): Update.
19074         (handle_alias_pairs): Update.
19075         (mark_functions_to_output): Update.
19076         (assemble_thunk): Update.
19077         (output_in_order): Update.
19078         (output_weakrefs): Update.
19079         (finalize_compilation_unit): Update.
19080         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
19081         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
19082         input_node, input_varpool_node): Update.
19083         * dbxout.c (dbxout_expand_expr): Update.
19084         * cgraphclones.c (cgraph_clone_node): Update.
19085         (cgraph_copy_node_for_versioning): Update.
19086         (cgraph_materialize_clone): Update.
19087         (cgraph_materialize_all_clones): Update.
19088         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
19089         propagate_pure_const, propagate_nothrow): Update.
19090         * lto-streamer-out.c (lto_output, write_symbol): Update.
19091         * ipa-utils.c (ipa_reverse_postorder): Update.
19092         * ipa-inline.c (can_inline_edge_p): Update.
19093         (update_caller_keys, ipa_inline): Update.
19094         * dwarf2out.c (reference_to_unused,
19095         premark_types_used_by_global_vars_helper): Update.
19096         * tree-eh.c (tree_could_trap_p): Update.
19097         * ipa-split.c (consider_split, execute_split_functions): Update.
19098         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
19099         has_addr_references_p): Update; move ahead in file for better
19100         readability.
19101         (process_references): Simplify.
19102         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
19103         bodies are removed.
19104         (cgraph_comdat_can_be_unshared_p): Make static.
19105         (cgraph_externally_visible_p): Update.
19106         (varpool_externally_visible_p): Update.
19107         (function_and_variable_visibility): Update.
19108         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
19109         ipa_tm_mark_force_output_node): Update.
19110         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
19111         estimate_edge_devirt_benefit, inline_generate_summary,
19112         inline_write_summary): Update.
19113         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
19114         * ipa-prop.c (ipa_compute_jump_functions): Update.
19115         (ipa_print_node_params, ipa_prop_read_section,
19116         ipa_update_after_lto_read, read_replacements_section): Update.
19117         * varasm.c (mark_decl_referenced): Update.
19118         (assemble_alias, dump_tm_clone_pairs): Update.
19119         * tree-inline.c (copy_bb): Update.
19120         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
19121         Update.
19122         * symtab.c (dump_symtab_base): Print new flags.
19123         (verify_symtab_base): Verify new flags.
19124         (symtab_alias_ultimate_target): New function.
19125         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
19126         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
19127         Update.
19128         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
19129         Update.
19130         * i386.c (ix86_get_function_versions_dispatcher,
19131         ix86_generate_version_dispatcher_body): Update.
19132         (fold_builtin_cpu): Use varpool_add_new_variable.
19133         * varpool.c (varpool_remove_initializer): Break out from ...
19134         (varpool_remove_node): ... this one.
19135         (dump_varpool_node, varpool_node_for_asm,
19136         cgraph_variable_initializer_availability, varpool_analyze_node,
19137         varpool_assemble_decl, varpool_remove_unreferenced_decls,
19138         varpool_finalize_named_section_flags, varpool_create_variable_alias):
19139         Update.
19141 2013-05-29  Jan Hubicka  <jh@suse.cz>
19143         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
19145 2013-05-29  Easwaran Raman  <eraman@google.com>
19147         PR tree-optimization/57442
19148         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
19149         when control exits the main loop.
19151 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
19153         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
19154         and RX600.
19155         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
19156         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
19157         * rx/t-rx: Add rx100 under multi library matches option for nofpu
19158         option.
19160 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
19162         PR tree-optimization/57441
19163         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
19164         Don't limit size of incr_vec to number of candidates.
19166 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
19168         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
19169         and mips16 directories.
19170         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
19171         (MULTILIB_DIRNAMES): Ditto.
19172         (MULTILIB_EXCEPTIONS): Add new exceptions.
19173         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
19174         (MULTILIB_DIRNAMES): Ditto.
19175         (MULTILIB_EXCEPTIONS): Add new exceptions.
19177 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
19178             Marcus Shawcroft  <marcus.shawcroft@arm.com>
19180         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
19181         SYMBOL_TINY_ABSOLUTE.
19182         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
19183         SYMBOL_TINY_ABSOLUTE.
19184         (aarch64_expand_mov_immediate): Likewise.
19185         (aarch64_classify_symbol): Likewise.
19186         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
19187         Permit SYMBOL_TINY_ABSOLUTE.
19188         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
19190 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
19191             Marcus Shawcroft  <marcus.shawcroft@arm.com>
19193         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
19194         Refactor if/switch.  Replace gcc_assert with if.
19196 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
19198         * config/i386/i386.c (initial_ix86_tune_features): Enable
19199         FP Reassociation for AMD bdver1 and bdver2.
19201 2013-05-29  Martin Jambor  <mjambor@suse.cz>
19203         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
19204         and IMAGPART_EXPR do not occur within other handled_components.
19206 2013-05-29  Richard Biener  <rguenther@suse.de>
19208         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
19209         access on whether the use is in the BB we currently try to
19210         vectorize.
19211         (vect_bb_vectorization_profitable_p): Pass the BB we currently
19212         vectorize to vect_bb_slp_scalar_cost.
19214 2013-05-29  Richard Biener  <rguenther@suse.de>
19216         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
19217         computing scalar cost offsetted by stmts that are kept live
19218         by scalar uses.
19219         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
19220         for computation of scalar cost.
19222 2013-05-28  Steve Ellcey  <sellcey@mips.com>
19224         * config/mips/mips-cpus.def (mips32r2): Change processor type.
19226 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
19228         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
19229         array notation built-in reduction functions.
19230         * doc/passes.texi (Passes): Added documentation about changes done
19231         for Cilk Plus.
19232         * doc/invoke.texi (C Dialect Options): Added documentation about
19233         the -fcilkplus flag.
19234         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
19235         (BUILTINS_DEF): Depend on cilkplus.def.
19236         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
19237         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
19238         * cilkplus.def: New file.
19240 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
19242         PR rtl-optimization/57439
19243         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
19245 2013-05-28  Easwaran Raman  <eraman@google.com>
19247         PR tree-optimization/57337
19248         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
19249         (find_insert_point): Correctly identify the insertion point
19250         when two statements with the same UID is compared.
19252 2013-05-28  Richard Biener  <rguenther@suse.de>
19254         PR tree-optimization/56787
19255         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
19256         from the list of data references.
19257         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
19258         clobbers.
19259         (vect_analyze_loop_operations): Likewise.
19260         (vect_transform_loop): Remove clobbers.
19262 2013-05-28  Martin Jambor  <mjambor@suse.cz>
19264         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
19265         and REALPART_EXPRs have scalar type.
19267 2013-05-28  Richard Biener  <rguenther@suse.de>
19269         PR tree-optimization/57411
19270         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
19271         virtual operands.
19272         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
19273         virtual operand propagation.
19275 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
19277         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
19278         destination register for bmasksi_vis.
19279         (vector_init_bshuffle): Likewise.
19280         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
19282 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
19284         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
19285         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
19286         mode if the instruction isn't available in the original mode.
19287         * config/sparc/sparc.opt (mfix-ut699): New option.
19288         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
19289         (divdf3): Turn into expander.
19290         (divdf3_nofix): New insn.
19291         (divdf3_fix): Likewise.
19292         (divsf3): Disable if -mfix-ut699.
19293         (sqrtdf2): Turn into expander.
19294         (sqrtdf2_nofix): New insn.
19295         (sqrtdf2_fix): Likewise.
19296         (sqrtsf2): Disable if -mfix-ut699.
19298 2013-05-27  Richard Biener  <rguenther@suse.de>
19300         PR middle-end/57412
19301         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
19302         block for the new loop.
19304 2013-05-27  Richard Biener  <rguenther@suse.de>
19306         PR tree-optimization/57343
19307         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
19308         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
19309         (number_of_iterations_cond): Do not build the folded tree.
19311 2013-05-27  Richard Biener  <rguenther@suse.de>
19313         Revert
19314         PR middle-end/57381
19315         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
19316         OEP_CONSTANT_ADDRESS_OF retained.
19318         PR tree-optimization/57417
19319         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
19320         for unchanged base.
19321         (set_ssa_val_to): Compare addresses using
19322         get_addr_base_and_unit_offset.
19324 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
19326         PR rtl-optimization/56833
19327         * postreload.c (move2add_record_mode): New function.
19328         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
19329         (move2add_use_add2_insn): Use move2add_record_sym_value.
19330         (move2add_use_add3_insn): Likewise.
19331         (reload_cse_move2add): Use move2add_valid_value_p and
19332         move2add_record_mode.  Invalidate call-clobbered and REG_INC
19333         affected regs by setting reg_mode to VOIDmode.
19334         (move2add_note_store): Don't pretend the inside of a SUBREG is
19335         the actual destination.  Invalidate single/leading registers by
19336         setting reg_mode to VOIDmode.
19337         Use move2add_record_sym_value, move2add_valid_value_p and
19338         move2add_record_mode.
19340 2013-05-27  Richard Biener  <rguenther@suse.de>
19342         PR tree-optimization/57396
19343         * tree-affine.c (double_int_constant_multiple_p): Properly
19344         return false for val == 0 and div != 0.
19346 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
19348         * config/mips/mips.h: Use #elif in preprocessor conditions.
19350 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
19352         PR target/53916
19353         * config/mips/constraints.md (kl): New constraint.
19354         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
19355         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
19356         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
19357         from LO for MIPS16.
19358         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
19360 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
19362         PR target/55777
19363         * config/mips/mips.c (mips_can_inline_p): New function.
19364         (TARGET_CAN_INLINE_P): Define.
19366 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
19368         * sched-int.h (ds_t, dw_t): Make unsigned int.
19369         Fix documentation that describes how all the ds_t bits are used.
19370         Reserve the last bit for delayed-branch scheduling.
19371         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
19372         (BITS_PER_DEP_WEAK): Fix definition and documentation.
19373         (gen_dep_weak_1): Remove prototype.
19374         * sched-deps.c (get_dep_weak_1): Make static.
19375         * target.def (speculate_insn, needs_block_p, gen_spec_check,
19376         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
19377         * doc/tm.texi: Regenerate.
19378         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
19380 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
19382         PR debug/56950
19383         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
19385 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
19386             Sandra Loosemore  <sandra@codesourcery.com>
19388         * config.gcc (powerpc-*): Allow native for with-cpu.
19390 2013-05-24  Jeff Law  <law@redhat.com>
19392         PR tree-optimization/57124
19393         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
19394         conversion feeding a condition if the range has an overflow
19395         if -fstrict-overflow.  Add warnings for when we do make the
19396         transformation.
19398 2013-05-24  Dehao Chen  <dehao@google.com>
19400         * tree-cfg.c (locus_discrim_map): Fix the typo.
19401         (locus_discrim_hasher): Likewise.
19402         (locus_discrim_hasher::hash): Likewise.
19403         (locus_discrim_hasher::equal): Likewise.
19405 2013-05-24  Martin Jambor  <mjambor@suse.cz>
19407         PR tree-optimization/57294
19408         * cgraph.h (ipa_record_stmt_references): Declare.
19409         * cgraphbuild.c (ipa_record_stmt_references): New function.
19410         (build_cgraph_edges): Use ipa_record_stmt_references.
19411         (rebuild_cgraph_edges): Likewise.
19412         (cgraph_rebuild_references): Likewise.
19413         * ipa-prop.c (ipa_modify_call_arguments): Discard references
19414         associated with the old statement and build references from the
19415         newly built statements.
19416         * ipa-ref.c (ipa_remove_stmt_references): New function.
19417         * ipa-ref.h (ipa_remove_stmt_references): Declare.
19419 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
19421         * lra-constraints.c (emit_spill_move): Use smaller mode for
19422         mem-mem moves.
19423         (check_and_process_move): Consider mem-reg moves for secondary
19424         too.
19425         (curr_insn_transform): Don't lose insns emitted before for
19426         secondary memory moves.
19427         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
19428         reg set up in the current insn.
19430 2013-05-24  Dehao Chen  <dehao@google.com>
19432         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
19433         hash function.
19434         (locus_descrim_hasher::equal): Likewise.
19435         (build_gimple_cfg): New discriminator assignment algorithm.
19436         (make_edges): Likewise.
19437         (next_discriminator_for_locus): Likewise.
19438         (same_line_p): Likewise.
19439         (assign_discriminators): Likewise.
19440         (make_cond_expr_edges): Likewise.
19441         (make_gimple_switch_edges): Likewise.
19442         (make_goto_expr_edges): Likewise.
19443         (make_gimple_asm_edges): Likewise.
19445 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
19447         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
19448         X format specifier to only display bottom 16 bits.
19449         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
19450         immediate to match for operand 2, since it will be masked.
19452 2013-05-24  Richard Biener  <rguenther@suse.de>
19454         PR tree-optimization/57287
19455         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
19456         all SSA names that occur in abnormal PHIs.
19458 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
19460         PR tree-ssa/57385
19461         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
19462         that index is not negative.
19464 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
19466         PR rtl-optimization/55177
19467         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
19468         (simplify_byte_swapping_operation): New.
19469         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
19470         (simplify_relational_operation_1): Deal with BSWAP.
19472 2013-05-23  Richard Henderson  <rth@redhat.com>
19474         PR target/56742
19475         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
19476         (ix86_reorg): Call it.
19478 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
19480         PR target/57379
19481         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
19482         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
19483         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
19485 2013-05-23  Christian Bruel  <christian.bruel@st.com>
19487         PR debug/57351
19488         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
19490 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
19491             Marcus Shawcroft  <marcus.shawcroft@arm.com>
19493         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
19494         * config/aarch64/constraints.md (Usa): Remove.
19495         * doc/md.texi (AArch64 Usa): Remove.
19497 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
19498             Marcus Shawcroft  <marcus.shawcroft@arm.com>
19500         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
19501         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
19502         * config/aarch64/predicates.md (aarch64_const_address): Remove.
19503         (aarch64_mov_operand): Use aarch64_mov_operand_p.
19505 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
19507         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
19508         instruction (AdvSIMD).
19509         * config/aarch64/aarch64-builtins.c
19510         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
19511         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
19513 2013-05-23  Martin Jambor  <mjambor@suse.cz>
19515         PR middle-end/57347
19516         * tree.h (contains_bitfld_component_ref_p): Declare.
19517         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
19518         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
19519         caller.
19520         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
19521         not access a bit-field.  Assert all final offsets are byte-aligned.
19523 2013-05-23  Richard Biener  <rguenther@suse.de>
19525         PR tree-optimization/57380
19526         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
19527         least one invariant or re-used load.
19528         * passes.c (init_optimization_passes): Move pass_phiprop before
19529         pass_forwprop.
19531 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
19533         * config/aarch64/aarch64-simd.md
19534         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
19536 2013-05-23  Richard Biener  <rguenther@suse.de>
19538         PR middle-end/57381
19539         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
19540         OEP_CONSTANT_ADDRESS_OF retained.
19542 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
19544         PR middle-end/57344
19545         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
19546         don't lower unit.  Handle unit not being always BITS_PER_WORD.
19548 2013-05-23  Richard Biener  <rguenther@suse.de>
19550         PR rtl-optimization/57341
19551         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
19552         instead of true_dependence.
19554 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
19556         * bb-reorder.c (branch_threshold): Make const.
19557         (exec_threshold): Ditto.
19559 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
19560             Pat Haugen  <pthaugen@us.ibm.com>
19561             Peter Bergner  <bergner@vnet.ibm.com>
19563         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
19564         documentation for the power8 crypto builtins.
19566         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
19568         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
19569         macros for defining power8 builtin functions.
19570         (BU_P8V_AV_2): Likewise.
19571         (BU_P8V_AV_P): Likewise.
19572         (BU_P8V_VSX_1): Likewise.
19573         (BU_P8V_OVERLOAD_1): Likewise.
19574         (BU_P8V_OVERLOAD_2): Likewise.
19575         (BU_CRYPTO_1): Likewise.
19576         (BU_CRYPTO_2): Likewise.
19577         (BU_CRYPTO_3): Likewise.
19578         (BU_CRYPTO_OVERLOAD_1): Likewise.
19579         (BU_CRYPTO_OVERLOAD_2): Likewise.
19580         (XSCVSPDP): Fix typo, point to the correct instruction.
19581         (VCIPHER): Add power8 crypto builtins.
19582         (VCIPHERLAST): Likewise.
19583         (VNCIPHER): Likewise.
19584         (VNCIPHERLAST): Likewise.
19585         (VPMSUMB): Likewise.
19586         (VPMSUMH): Likewise.
19587         (VPMSUMW): Likewise.
19588         (VPERMXOR_V2DI): Likewise.
19589         (VPERMXOR_V4SI: Likewise.
19590         (VPERMXOR_V8HI: Likewise.
19591         (VPERMXOR_V16QI: Likewise.
19592         (VSHASIGMAW): Likewise.
19593         (VSHASIGMAD): Likewise.
19594         (VPMSUM): Likewise.
19595         (VPERMXOR): Likewise.
19596         (VSHASIGMA): Likewise.
19598         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
19599         __CRYPTO__ if the crypto instructions are available.
19600         (altivec_overloaded_builtins): Add support for overloaded power8
19601         builtins.
19603         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
19604         support for power8 crypto builtins.
19605         (builtin_function_type): Likewise.
19606         (altivec_init_builtins): Add support for builtins that take vector
19607         long long (V2DI) arguments.
19609         * config/rs6000/crypto.md: New file, define power8 crypto
19610         instructions.
19612 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
19613             Pat Haugen  <pthaugen@us.ibm.com>
19614             Peter Bergner  <bergner@vnet.ibm.com>
19616         * doc/invoke.texi (Option Summary): Add power8 options.
19617         (RS/6000 and PowerPC Options): Likewise.
19619         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
19620         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
19621         wm, wn, wr documentation.
19623         * config/rs6000/constraints.md (wm): New constraint for VSX
19624         registers if direct move instructions are enabled.
19625         (wn): New constraint for no registers.
19626         (wq): New constraint for quad word even GPR registers.
19627         (wr): New constraint if 64-bit instructions are enabled.
19628         (wv): New constraint if power8 vector instructions are enabled.
19629         (wQ): New constraint for quad word memory locations.
19631         * config/rs6000/predicates.md (const_0_to_15_operand): New
19632         constraint for 0..15 for crypto instructions.
19633         (gpc_reg_operand): If VSX allow registers in VSX registers as well
19634         as GPR and floating point registers.
19635         (int_reg_operand): New predicate to match only GPR registers.
19636         (base_reg_operand): New predicate to match base registers.
19637         (quad_int_reg_operand): New predicate to match even GPR registers
19638         for quad memory operations.
19639         (vsx_reg_or_cint_operand): New predicate to allow vector logical
19640         operations in both GPR and VSX registers.
19641         (quad_memory_operand): New predicate for quad memory operations.
19642         (reg_or_indexed_operand): New predicate for direct move support.
19644         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
19645         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
19646         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
19647         (POWERPC_MASKS): Add power8 options.
19648         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
19649         various options.
19651         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
19652         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
19654         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
19655         (-mpower8-fusion): New power8 options.
19656         (-mpower8-fusion-sign): Likewise.
19657         (-mpower8-vector): Likewise.
19658         (-mcrypto): Likewise.
19659         (-mdirect-move): Likewise.
19660         (-mquad-memory): Likewise.
19662         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
19663         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
19664         registers.
19665         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
19666         (rs6000_debug_vector_unit): Add p8_vector.
19667         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
19668         definitions.  Also print fusion state.
19669         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
19670         (rs6000_builtin_mask_calculate): Add power8 builtin support.
19671         (rs6000_option_override_internal): Add support for power8.
19672         (rs6000_common_init_builtins): Add debugging for skipped builtins
19673         if -mdebug=builtin.
19674         (rs6000_adjust_cost): Add power8 support.
19675         (rs6000_issue_rate): Likewise.
19676         (insn_must_be_first_in_group): Likewise.
19677         (insn_must_be_last_in_group): Likewise.
19678         (force_new_group): Likewise.
19679         (rs6000_register_move_cost): Likewise.
19680         (rs6000_opt_masks): Likewise.
19682         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
19683         power8 capable assembler, default to power7 options.
19684         (TARGET_DIRECT_MOVE): Likewise.
19685         (TARGET_CRYPTO): Likewise.
19686         (TARGET_P8_VECTOR): Likewise.
19687         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
19688         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
19689         (VECTOR_MEM_P8_VECTOR_P): Likewise.
19690         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
19691         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
19692         (TARGET_XSCVDPSPN): Likewise.
19693         (TARGET_XSCVSPDPN): Likewsie.
19694         (TARGET_SYNC_HI_QI): Likewise.
19695         (TARGET_SYNC_TI): Likewise.
19696         (MASK_CRYPTO): Likewise.
19697         (MASK_DIRECT_MOVE): Likewise.
19698         (MASK_P8_FUSION): Likewise.
19699         (MASK_P8_VECTOR): Likewise.
19700         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
19701         temporary used by some of the direct move instructions to get two FP
19702         temporary registers does not force creation of a stack frame.
19703         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
19704         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
19705         that any VSX registers are tieable, even if they are also an
19706         Altivec vector mode.
19707         (r6000_reg_class_enum): Add wm, wr, wv constraints.
19708         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
19709         (RS6000_BTM_CRYPTO): Likewise.
19710         (RS6000_BTM_COMMON): Likewise.
19712         * config/rs6000/rs6000.md (cpu attribute): Add power8.
19713         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
19714         (enum rs6000_vector): Add power8 vector support.
19716 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
19718         PR target/19599
19719         PR target/57340
19720         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
19721         (any_sibcall_could_use_r3): this and handle indirect calls.
19722         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
19724 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
19726         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
19728 2013-05-22  Richard Biener  <rguenther@suse.de>
19730         PR middle-end/57349
19731         * profile.c (branch_prob): Do not split blocks that are
19732         abnormally receiving from ECF_RETURNS_TWICE functions.
19734 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
19736         * recog.c (offsettable_address_addr_space_p): Fix calculation of
19737         address mode.  Move pointer mode initialization to the same place.
19739 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
19741         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
19742         while it has any effect.
19744 2013-05-21  Easwaran Raman  <eraman@google.com>
19746         PR tree-optimization/57322
19747         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
19748         UID of the statement added to the BB to be 1.
19750 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
19752         PR tree-optimization/57331
19753         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
19754         of conversion from pointer type to integral type with integer.
19756 2013-05-21  Martin Jambor  <mjambor@suse.cz>
19758         PR lto/57289
19759         * ipa-prop.c (ipa_read_node_info): Process param_used and
19760         controlled_uses in the same order as when writing.
19762 2013-05-21  Magnus Granberg  <baldrick@free.fr>
19764         PR plugins/56754
19765         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
19767 2013-05-21  Richard Biener  <rguenther@suse.de>
19769         PR tree-optimization/57318
19770         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
19771         estimate stmts with side-effects as likely eliminated.
19773 2013-05-21  Richard Biener  <rguenther@suse.de>
19775         PR tree-optimization/57330
19776         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
19777         preserve the call stmts fntype.
19779 2013-05-21  Richard Biener  <rguenther@suse.de>
19781         PR tree-optimization/57303
19782         * tree-ssa-sink.c (statement_sink_location): Improve killing
19783         stmt detection and properly handle self-assignments.
19785 2013-05-21  Christian Bruel  <christian.bruel@st.com>
19787         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
19788         spanning registers. LEAF_REG_REMAP is supported only for contiguous
19789         registers. Set register size out of the PARALLEL loop.
19791 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
19793         PR target/56547
19794         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
19795         (*fmasf4, *fmasf4_media): New insns.
19797 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
19799         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
19800         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
19801         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
19802         (mips_idiv_insns): Update the comments to say that the returned
19803         instruction counts are in units of BASE_INSN_LENGTH.
19804         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
19805         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
19806         using 2 rather than 4 as the length of indirect MIPS16 and
19807         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
19808         length of a NOP.  Don't divide MIPS16 lengths by 2.
19809         (mips16_split_long_branches): Assume a branch is long if the
19810         length is greater than 4 rather than 8.
19811         * config/mips/mips.md (length): Give MIPS16 lengths directly,
19812         rather than multiplying them by 2.  Multiply instruction counts
19813         by BASE_INSN_LENGTH rather than 4.
19814         (*jump_mips16, tls_get_tp_mips16_<mode>)
19815         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
19817 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
19819         * config/mips/mips.md (extended_mips16): Remove branch case.
19820         (length): Remove duplicated extended_mips16 test.
19822 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
19824         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
19826 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
19828         * recog.h (Recog_data): Rename to...
19829         (recog_data_d): ...this.
19830         (recog_data): Update accordingly.
19831         * recog.c (recog_data): Likewise.
19832         * reload.c (save_recog_data): Likewise.
19833         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
19834         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
19836 2013-05-17  Julian Brown  <julian@codesourcery.com>
19838         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
19839         found in a REG_EQUAL note, invalidate it.
19841 2013-05-17  Easwaran Raman  <eraman@google.com>
19843         * tree-ssa-reassoc.c (find_insert_point): New function.
19844         (insert_stmt_after): Likewise.
19845         (get_def_stmt): Likewise.
19846         (ensure_ops_are_available): Likewise.
19847         (not_dominated_by): Likewise.
19848         (rewrite_expr_tree): Do not move statements beyond what is
19849         necessary. Remove call to swap_ops_for_binary_stmt...
19850         (reassociate_bb): ... and move it here.
19851         (build_and_add_sum): Assign UIDs for new statements.
19852         (linearize_expr): Likewise.
19853         (do_reassoc): Renumber gimple statement UIDs.
19855 2013-05-17  Jan Hubicka  <jh@suse.cz>
19857         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
19858         weakrefs.
19859         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
19860         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
19861         weakrefs.
19862         (output_weakrefs): Update.
19864 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
19865             Martin Jambor  <mjambor@suse.cz>
19867         PR middle-end/57276
19868         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
19869         value that corresponds to the given aggval is found in values vector.
19871 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
19873         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
19874         sse, sse2, sse3, ssse3 and sse4a flags to options.
19876 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
19878         * gengtype-state.c: (s_expr_writer): New class, to handle
19879         prettifying of output layout of s-expressions.
19880         (state_writer): New class, to write out gtype.state.
19881         (state_written_type_count): Move this variable into member data of
19882         state_writer.
19883         (s_expr_writer::s_expr_writer): New code: constructor for new class
19884         (state_writer::state_writer(): ditto
19885         (s_expr_writer::write_new_line): New function
19886         (s_expr_writer::write_any_indent): ditto
19887         (s_expr_writer::begin_s_expr): ditto
19888         (s_expr_writer::end_s_expr): ditto
19889         (write_state_fileloc): convert to method of state_writer...
19890         (state_writer:: write_state_fileloc): ...and use methods of
19891         s_expr_writer to write indentation into the gtype.state output file
19892         to visually represent the hierarchical structure of the list
19893         structures
19894         (write_state_fields): ditto, renaming to...
19895         (state_writer::write_state_fields)
19896         (write_state_a_string): ditto, renaming to...
19897         (state_writer::write_state_a_string)
19898         (write_state_string_option): ditto, renaming to...
19899         (state_writer::write_state_string_option)
19900         (write_state_type_option): ditto, renaming to...
19901         (state_writer::write_state_type_option)
19902         (write_state_nested_option): ditto, renaming to...
19903         (state_writer::write_state_nested_option)
19904         (write_state_option): ditto, renaming to...
19905         (state_writer::write_state_option)
19906         (write_state_options): ditto, renaming to...
19907         (state_writer::write_state_options)
19908         (write_state_lang_bitmap): ditto, renaming to...
19909         (state_writer::write_state_lang_bitmap)
19910         (write_state_version): ditto, renaming to...
19911         (state_writer::write_state_version)
19912         (write_state_scalar_type): ditto, renaming to...
19913         (state_writer::write_state_scalar_type)
19914         (write_state_string_type): ditto, renaming to...
19915         (state_writer::write_state_string_type)
19916         (write_state_undefined_type): ditto, renaming to...
19917         (state_writer::write_state_undefined_type)
19918         (write_state_struct_union_type): ditto, renaming to...
19919         (state_writer::write_state_struct_union_type)
19920         (write_state_struct_type): ditto, renaming to...
19921         (state_writer::write_state_struct_type)
19922         (write_state_user_struct_type): ditto, renaming to...
19923         (state_writer::write_state_user_struct_type)
19924         (write_state_lang_struct_type): ditto, renaming to...
19925         (state_writer::write_state_lang_struct_type)
19926         (write_state_param_struct_type): ditto, renaming to...
19927         (state_writer::write_state_param_struct_type)
19928         (write_state_pointer_type): ditto, renaming to...
19929         (state_writer::write_state_pointer_type)
19930         (write_state_array_type): ditto, renaming to...
19931         (state_writer::write_state_array_type)
19932         (write_state_gc_used): ditto, renaming to...
19933         (state_writer::write_state_gc_used)
19934         (write_state_common_type_content): ditto, renaming to...
19935         (state_writer::write_state_common_type_content)
19936         (write_state_type): ditto, renaming to...
19937         (state_writer::write_state_type)
19938         (write_state_pair_list): ditto, renaming to...
19939         (state_writer::write_state_pair_list)
19940         (write_state_pair): ditto, renaming to...
19941         (state_writer::write_state_pair)
19942         (write_state_typedefs): ditto, renaming to...
19943         (state_writer::write_state_typedefs)
19944         (write_state_structures): ditto, renaming to...
19945         (state_writer::write_state_structures)
19946         (write_state_param_structs): ditto, renaming to...
19947         (state_writer::write_state_param_structs)
19948         (write_state_variables): ditto, renaming to...
19949         (state_writer::write_state_variables)
19950         (write_state_srcdir): ditto, renaming to...
19951         (state_writer::write_state_srcdir)
19952         (write_state_files_list): ditto, renaming to...
19953         (state_writer::write_state_files_list)
19954         (write_state_languages): ditto, renaming to...
19955         (state_writer::write_state_languages)
19956         (write_state): create a state_writer instance and use it when
19957         writing out the state file
19959 2013-05-17  Mike Stump  <mikestump@comcast.net>
19961         PR rtl-optimization/57304
19962         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
19963         accessing DF_REF_REAL_LOC.
19965 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
19967         PR rtl-optimization/57281
19968         PR rtl-optimization/57300
19969         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
19970         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
19971         what the other splitter did if the registers are dead.
19973 2013-05-17  Richard Biener  <rguenther@suse.de>
19975         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
19976         MEM_REF offsets.
19978 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
19980         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
19981         linking.
19983 2013-05-17  Marek Polacek  <polacek@redhat.com>
19985         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
19986         length when doing non-zero store of storing '\0' to '\0'.
19988 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
19990         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
19991         vect_external_def oprnd1 with loop_vinfo, try to emit
19992         optional cast, negation and and stmts on the loop preheader
19993         edge instead of into the pattern def seq.
19995         PR tree-optimization/57051
19996         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
19997         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
19999 2013-05-16  Nick Clifton  <nickc@redhat.com>
20001         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
20002         (rl78_is_naked_func): New function.
20003         (rl78_expand_prologue): Skip prologue generation for naked functions.
20004         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
20005         * doc/extend.texi (naked): Add RL78 to the list of processors
20006         that supports this attribute.
20008 2013-05-16  Jeff Law  <law@redhat.com>
20010         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
20012 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
20014         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
20015         cache parameters using detect_caches_amd also for CYRIX,
20016         NSC and TM2 signatures.
20018 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
20019             Dzianis Kahanovich  <mahatma@eu.by>
20021         PR target/45359
20022         PR target/46396
20023         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
20024         VIA/Centaur processors and determine their cache parameters
20025         using detect_caches_amd.
20027 2013-05-16  Teresa Johnson  <tejohnson@google.com>
20029         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
20030         (rtl_verify_edges): New function.
20031         (rtl_verify_bb_insns): Ditto.
20032         (rtl_verify_bb_pointers): Ditto.
20033         (rtl_verify_bb_insn_chain): Ditto.
20034         (rtl_verify_fallthru): Ditto.
20035         (rtl_verify_bb_layout): Ditto.
20036         (rtl_verify_flow_info_1): Outline checks into new functions.
20037         (rtl_verify_flow_info): Ditto.
20039 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
20041         * cfghooks.c (copy_bbs): Add update_dominance argument.
20042         * cfghooks.h (copy_bbs): Update prototype.
20043         * tree-cfg.c (gimple_duplicate_sese_region):
20044         Add update_dominance argument.
20045         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
20046         * tree-ssa-loop-ch.c (copy_loop_headers): Update
20047         gimple_duplicate_sese_region call.
20048         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
20049         Update copy_bbs call.
20050         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
20051         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
20053 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
20055         * tree-vectorizer.h (NUM_PATTERNS): Increment.
20056         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
20057         vect_recog_rotate_pattern.
20058         (vect_recog_rotate_pattern): New function.
20060 2013-05-16  Jason Merrill  <jason@redhat.com>
20062         * Makefile.in (LLINKER): New variable.
20063         (mostlyclean): Remove link mutex.
20064         * configure.ac: Handle --enable-link-mutex.
20065         * lock-and-run.sh: New script.
20067 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
20069         PR target/19599
20070         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
20071         for NULL decl.
20073 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20075         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
20077 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
20079         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
20080         * config/arm/arm.c (next_consecutive_mem): New function.
20081         (gen_movmem_ldrd_strd): Likewise.
20082         * config/arm/arm.md (movmemqi): Update condition and code.
20083         (unaligned_loaddi, unaligned_storedi): New patterns.
20085 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20087         * config.gcc: Obsolete *-*-solaris2.9*.
20088         * doc/install.texi (Specific, *-*-solaris2*): Document it.
20090 2013-05-16  Richard Biener  <rguenther@suse.de>
20092         * passes.c (init_optimization_passes): Move pass_parallelize_loops
20093         earlier, after GRAPHITE transforms and IV canonicalization.
20095 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
20097         * omp-low.c (extract_omp_for_data): For collapsed loops,
20098         if at least one of the loops is known at compile time to
20099         iterate zero times, set count to 0.
20100         (expand_omp_regimplify_p): New function.
20101         (expand_omp_for_generic): For collapsed loops, if at least
20102         one of the loops isn't known to iterate at least once,
20103         add runtime check with setting count to 0.
20104         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
20105         For unsigned types if it isn't known at compile time that
20106         the loop will iterate at least once, add runtime check to bypass
20107         the whole loop if initial condition isn't true.
20109 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
20111         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
20113 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
20115         PR middle-end/57286
20116         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
20117         transformations to avoid an infinite loop.
20119 2013-05-16  Marek Polacek  <polacek@redhat.com>
20121         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
20123 2013-05-15  Leif Ekblad  <leif@rdos.net>
20125         * config/i386/i386.c (ix86_decompose_address): Use
20126         DEFAULT_TLS_SEG_REG to access TLS segment register.
20127         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
20128         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
20129         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
20131 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
20133         PR target/57260
20134         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
20135         sibling calls to functions that would normally be lazily bound,
20136         unless $gp is call-clobbered.
20138 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
20140         * config/i386/i386.c (ix86_option_override_internal): Update
20141         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
20142         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
20143         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
20144         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
20145         of TARGET_3DNOW.
20146         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
20148 2013-05-15  Andreas Schwab  <schwab@suse.de>
20150         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
20151         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
20152         third operand.
20154 2013-05-15  Teresa Johnson  <tejohnson@google.com>
20156         * loop-unroll.c (report_unroll_peel): Check decision before
20157         emitting unroll/peel message.
20159 2013-05-15  Teresa Johnson  <tejohnson@google.com>
20161         * function.h (has_bb_partition): New rtl_data flag.
20162         (bb_reorder_complete): Ditto.
20163         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
20164         instead of flag_reorder_blocks_and_partition.
20165         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
20166         with some enhancements.
20167         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
20168         * bb-reorder.c (connect_traces): Check for has_bb_partition
20169         instead of flag_reorder_blocks_and_partition.
20170         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
20171         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
20172         verify_hot_cold_block_grouping.
20173         (partition_hot_cold_basic_blocks): Set has_bb_partition.
20175 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
20177         PR target/19599
20178         * config/arm/predicates.md (call_insn_operand): New predicate.
20179         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
20180         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
20181         if insn is not a tail call.
20182         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
20183         registers.
20184         * config/arm/arm.h (enum reg_class): New caller save register class.
20185         (REG_CLASS_NAMES): Likewise.
20186         (REG_CLASS_CONTENTS): Likewise.
20187         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
20188         without decls.
20190 2013-05-15  Richard Biener  <rguenther@suse.de>
20192         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
20193         of MSG_OPTIMIZED_LOCATIONS.
20194         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
20195         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
20196         message.
20197         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
20198         of MSG_OPTIMIZED_LOCATIONS.
20199         (execute_vect_slp): Likewise.
20200         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
20201         (vect_create_cond_for_alias_checks): Likewise.
20202         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
20203         (vect_recog_widen_mult_pattern): Likewise.
20204         (vect_recog_widen_sum_pattern): Likewise.
20205         (vect_recog_over_widening_pattern): Likewise.
20206         (vect_recog_widen_shift_pattern): Likewise.
20207         (vect_recog_vector_vector_shift_pattern): Likewise.
20208         (vect_recog_divmod_pattern): Likewise.
20209         (vect_recog_mixed_size_cond_pattern): Likewise.
20210         (vect_recog_bool_pattern): Likewise.
20211         (vect_pattern_recog_1): Likewise.
20213 2013-05-15  Martin Jambor  <mjambor@suse.cz>
20215         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
20216         non-functions to builtin_unreachable.
20217         * ipa-inline-transform.c (inline_call): Do not assert estimates were
20218         correct when new direct edges were discovered.
20220 2013-05-15  Martin Jambor  <mjambor@suse.cz>
20222         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
20223         header, print symbol order instead of node uid, print more information
20224         about indirect edge targets.
20225         (ipa_make_edge_direct_to_target): Print symbol order instead of node
20226         uids.
20227         (ipa_make_edge_direct_to_target): Likewise.
20228         (remove_described_reference): Likewise.
20229         (propagate_controlled_uses): Likewise.
20230         (ipa_print_node_params): Also print symbol order.
20231         (ipcp_transform_function): Print symbol order instead of node uids.
20232         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
20233         (cgraph_get_create_real_symbol_node): Likewise.
20234         * ipa-cp.c (print_lattice): Likewise.
20235         (print_all_lattices): Likewise.
20236         (determine_versionability): Likewise.
20237         (initialize_node_lattices): Likewise.
20238         (estimate_local_effects): Likewise.
20239         (update_profiling_info): Likewise.
20240         (create_specialized_node): Likewise.
20241         (perhaps_add_new_callers): Likewise.
20242         (decide_about_value): Likewise.
20243         (decide_whether_version_node): Likewise.
20244         (identify_dead_nodes): Likewise.
20245         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
20246         (dump_inline_summary): Likewise.
20247         (estimate_node_size_and_time): Likewise.
20248         (inline_analyze_function): Likewise.
20249         * ipa-inline.c (report_inline_failed_reason): Likewise.
20250         (want_early_inline_function_p): Likewise.
20251         (edge_badness): Likewise.
20252         (update_edge_key): Likewise.
20253         (inline_small_functions): Likewise.  Add dumping of order to two other
20254         dumps.
20255         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
20256         instead of node uids.
20257         (propagate_pure_const): Likewise.
20258         (propagate_pure_const): Likewise.
20259         * ipa-utils.c (dump_cgraph_node_set): Likewise.
20260         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
20261         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
20262         of node uids.
20263         * tree-pretty-print.c (dump_function_header): Likewise.
20264         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
20265         Print symbol order instead of node uids.
20267 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20269         * config/s390/s390.c (s390_register_move_cost): Don't impose the
20270         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
20272 2013-05-15  Richard Biener  <rguenther@suse.de>
20274         PR tree-optimization/57275
20275         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
20276         return value for fail to do runtime alias checks for gather loads.
20278 2013-05-15  Jan Hubicka  <jh@suse.cz>
20280         PR lto/57038
20281         PR lto/47375
20282         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
20283         weakrefs are not external.
20284         (lto_symtab_merge_decls): Fix thinko when dealing with
20285         non-lto_symtab decls.
20286         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
20287         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
20288         * varpool.c (dump_varpool_node): Dump more flags.
20290 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
20292         * config/i386/i386.c (processor_alias_table): Add instruction
20293         FSGSBASE for AMD bdver3 architecture.
20295 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
20297         * tree.c (warn_deprecated_use): Print file:line using locus color.
20298         * diagnostic.c (diagnostic_report_current_module): Print file:line
20299         and file:line:column using locus color.
20301 2013-05-14  Mike Stump  <mikestump@comcast.net>
20303         * gdbinit.in: Add __null.
20305 2013-05-14  Mike Stump  <mikestump@comcast.net>
20307         * recog.h: Rename struct recog_data to Recog_data.
20308         * recog.c: Likewise.
20309         * reload.c (can_reload_into): Likewise.
20310         * config/picochip/picochip.c: Likewise.
20312 2013-05-14  Mike Stump  <mikestump@comcast.net>
20314         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
20316 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
20318         * resource.h (struct resources): Remove unch_memory member.
20319         (CLEAR_RESOURCE): Don't clear unch_memory.
20320         * resource.c (mark_referenced_resources): Don't set it.
20321         (mark_set_resources): Likewise.
20322         (mark_target_live_regs): Don't clear it.
20323         (init_resource_info): Likewise.
20324         * reorg.c (resource_conflicts_p): Don't compare it.
20325         (redundant_insn): Don't set it.
20327         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
20328         Remove prototypes.
20329         * emit-rtl.c (next_label): Remove unused function.
20330         (skip_consecutive_labels, link_cc0_insns): Move to ...
20331         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
20332         only place where these functions are used, and make them static.
20334 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
20336         * fold-const.c (fold_negate_expr): Handle vectors.
20337         (fold_truth_not_expr): Make it static.
20338         (fold_invert_truthvalue): New static function.
20339         (invert_truthvalue_loc): Handle vectors. Do not call
20340         fold_truth_not_expr directly.
20341         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
20342         <TRUTH_NOT_EXPR>: Do not cast to boolean.
20343         (fold_comparison): Handle vector constants.
20344         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
20345         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
20346         * tree.h (fold_truth_not_expr): Remove declaration.
20348 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
20350         * config/aarch64/aarch64-simd.md
20351         (aarch64_vcond_internal<mode>): Rename to...
20352         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
20353         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
20354         float modes. Clarify all iterator modes.
20355         (vcond<mode><mode>): Use new name for vcond expanders.
20356         (vcond<v_cmp_result><mode>): Likewise.
20357         (vcondu<mode><mode>: Likewise.
20358         * config/aarch64/iterators.md (VDQF_COND): New.
20360 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
20362         PR bootstrap/57266
20363         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
20364         variable for the shift amount. Check that we shift by non-negative
20365         amounts.
20367 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
20369         PR target/42017
20370         * config/arm/arm.h (EPILOGUE_USES): Only return true
20371         for LR_REGNUM after epilogue_completed.
20373 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
20375         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
20376         is error_mark_node.
20378 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20380         PR target/57261
20381         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
20382         and Solaris 11+/x86 with gld.
20383         * configure: Regenerate.
20385 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
20387         * expmed.c (expand_shift_1): Canonicalize rotates by
20388         constant bitsize / 2 to bitsize - 1.
20389         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
20390         case ROTATERT>: Likewise.
20392         Revert:
20393         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
20395         * config/i386/i386.md (rotateinv): New code attr.
20396         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
20397         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
20398         roll $31, %eax, etc.
20400 2013-05-14  Richard Biener  <rguenther@suse.de>
20402         PR middle-end/57235
20403         * tree-eh.c (sink_clobbers): Give up for successors with
20404         multiple predecessors and no virtual uses.
20406 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
20408         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
20409         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
20411 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
20413         PR middle-end/57251
20414         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
20415         the case when both op0 and op1 have VOIDmode.
20417 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
20419         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
20420         in multiply-accumulate mode.
20422 2013-05-13  Guozhi Wei  <carrot@google.com>
20424         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
20426 2013-05-13  Kai Tietz  <ktietz@redhat.com>
20428         PR target/56975
20429         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
20430         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
20431         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
20432         * config/i386/i386.c (ix86_option_override_internal): Likewise.
20433         (ix86_expand_prologue): Likewise.
20434         (ix86_expand_split_stack_prologue): Likewise.
20435         (legitimate_pic_address_disp_p): Likewise.
20436         (legitimize_pic_address): Likewise.
20437         (legitimize_tls_address): Likewise.
20438         (legitimize_pe_coff_symbol): Likewise.
20439         (output_pic_addr_const): Likewise.
20440         (construct_plt_address): Likewise.
20441         (ix86_expand_call): Likewise.
20442         (x86_output_mi_thunk): Likewise.
20443         (x86_function_profiler): Likewise.
20445 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
20447         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
20448         similar switch cases.
20449         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
20450         (aarch64_simd_mov_to_<mode>low): Delete.
20451         (aarch64_simd_mov_to_<mode>high): Delete.
20452         (move_lo_quad_<mode>): Add w<-r alternative.
20453         (aarch64_simd_move_hi_quad_<mode>): Likewise.
20454         (aarch64_simd_mov_from_*): Update type attribute.
20455         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
20456         statement.
20458 2013-05-13  Jan Hubicka  <jh@suse.cz>
20460         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
20461         * config/i386/i386.c (ix86_compute_frame_layout,
20462         ix86_expand_epilogue, emit_i387_cw_initialization,
20463         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
20464         ix86_local_alignment): Fix use of size/speed predicates.
20466 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
20468         PR tree-optimization/45216
20469         PR tree-optimization/57157
20470         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
20471         the (-Y) & (B - 1) variant if OP is |.
20472         * expmed.c (expand_shift_1): For rotations by const0_rtx just
20473         return shifted.  Use (-op1) & (prec - 1) as other_amount
20474         instead of prec - op1.
20476 2013-05-13  Martin Jambor  <mjambor@suse.cz>
20478         PR middle-end/42371
20479         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
20480         (ipa_constant_data): New type.
20481         (ipa_jump_func): Use ipa_constant_data to hold information about
20482         constant jump functions.
20483         (ipa_get_jf_constant): Adjust to jump function type changes.
20484         (ipa_get_jf_constant_rdesc): New function.
20485         (ipa_param_descriptor): New field controlled_uses.
20486         (ipa_get_controlled_uses): New function.
20487         (ipa_set_controlled_uses): Likewise.
20488         * ipa-ref.h (ipa_find_reference): Declare.
20489         * ipa-prop.c (ipa_cst_ref_desc): New type.
20490         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
20491         changes.
20492         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
20493         New parameter cs.  Adjust all callers.
20494         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
20495         (remove_described_reference): New function.
20496         (jfunc_rdesc_usable): Likewise.
20497         (try_make_edge_direct_simple_call): Decrement controlled use count,
20498         attempt to remove reference if it hits zero.
20499         (combine_controlled_uses_counters): New function.
20500         (propagate_controlled_uses): Likewise.
20501         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
20502         (ipa_edge_duplication_hook): Duplicate reference descriptions.
20503         (ipa_print_node_params): Print described use counter.
20504         (ipa_write_jump_function): Adjust to jump function type changes.
20505         (ipa_read_jump_function): New parameter CS, pass it to
20506         ipa_set_jf_constant.  Adjust caller.
20507         (ipa_write_node_info): Stream controlled use count
20508         (ipa_read_node_info): Likewise.
20509         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
20510         asserting.
20511         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
20512         count.  Remove cloning-added reference if it reaches zero.
20513         * ipa-ref.c (ipa_find_reference): New function.
20515 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
20517         * config/i386/i386.c (processor_target_table): Modified default
20518         alignment values for AMD BD and BT architectures.
20520 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
20522         * tree-vect-generic.c (uniform_vector_p): Move ...
20523         * tree.c (uniform_vector_p): ... here.
20524         * tree.h (uniform_vector_p): Declare it.
20525         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
20526         into a scalar.
20528 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
20530         PR tree-optimization/57230
20531         * tree-ssa-strlen.c (handle_char_store): Record length for
20532         array store from STRING_CST.
20534         PR tree-optimization/57230
20535         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
20536         check.
20538 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
20540         * config/epiphany/epiphany.c (epiphany_init): Check size of
20541         NUM_MODES_FOR_MODE_SWITCHING.
20542         (epiphany_expand_prologue):
20543         Remove CONFIG_REGNUM initial value handling code.
20544         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
20545         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
20546         (emit_set_fp_mode, epiphany_mode_after): Likewise.
20547         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
20548         Don't return 1 for FP_MODE_NONE.
20549         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
20550         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
20551         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
20552         * config/epiphany/epiphany.md (save_config): New pattern.
20554 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
20556         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
20558 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
20560         * config/i386/i386.md (memory): Handle sseishft1.
20561         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
20562         (*vec_extractv2di_1): Ditto.
20564 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
20566         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
20567         saved registers.
20569 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
20571         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
20572         Add mthumb/march=armv7-a multilib.
20573         Add mthumb/march=armv7-r multilib.
20574         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
20576 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
20578         * config/v850/t-rtems: Add more multilibs.
20580 2013-05-10  Richard Biener  <rguenther@suse.de>
20582         PR tree-optimization/57214
20583         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
20584         not propagate from SSA names that occur in abnormal PHI nodes.
20586 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
20588         * stor-layout.c (element_precision): New function.
20589         * machmode.h (element_precision): Declare it.
20590         * tree.c (build_minus_one_cst): New function.
20591         (element_precision): Likewise.
20592         * tree.h (build_minus_one_cst): Declare new function.
20593         (element_precision): Likewise.
20594         * fold-const.c (operand_equal_p): Use element_precision.
20595         (fold_binary_loc): Handle vector types.
20596         * convert.c (convert_to_integer): Use element_precision.
20597         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
20598         separately.
20600 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
20602         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
20603         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
20604         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
20605         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
20606         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
20607         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
20608         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
20609         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
20610         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
20611         (Uuw8): New constraints.
20612         (Usb4): Move into alphabetical order.
20613         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
20614         (sd8_operand, ub8_operand, uw8_operand): New predicates.
20615         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
20616         previously unnamed patterns.
20617         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
20618         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
20619         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
20620         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
20621         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
20622         of specific lengths.
20624 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
20626         * config/i386/i386.md (rotateinv): New code attr.
20627         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
20628         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
20629         roll $31, %eax, etc.
20631         PR tree-optimization/45216
20632         PR tree-optimization/57157
20633         * tree-ssa-forwprop.c (simplify_rotate): New function.
20634         (ssa_forward_propagate_and_combine): Call it.
20636 2013-05-10  Richard Biener  <rguenther@suse.de>
20638         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
20639         disable peeling when we version for aliasing.
20640         (vector_alignment_reachable_p): Honor explicit user alignment.
20641         (vect_supportable_dr_alignment): Likewise.
20642         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
20643         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
20644         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
20645         then peeling to arrange for the cost-model check to come first.
20647 2013-05-10  Alan Modra  <amodra@gmail.com>
20649         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
20650         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
20651         * configure: Regenerate.
20653 2013-05-10  Alan Modra  <amodra@gmail.com>
20655         PR target/55033
20656         * varasm.c (default_elf_select_section): Move !DECL_P check..
20657         (get_named_section): ..to here before calling get_section_name.
20658         Adjust assertion.
20659         (default_section_type_flags): Add DECL_P check.
20660         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
20661         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
20663 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
20665         * config/epiphany/epiphany.c (epiphany_expand_prologue):
20666         When using gen_stack_adjust_str with a register offset, add a
20667         REG_FRAME_RELATED_EXPR note.
20669 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
20671         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
20672         (*vec_extractv4si_zext_mem): Ditto.
20673         (*vec_extractv2di): Add 0->x and x->x alternatives.
20674         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
20675         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
20677 2013-05-09  Jason Merrill  <jason@redhat.com>
20679         N3639 C++1y VLA support
20680         * gimplify.c (gimplify_vla_decl): Don't touch an existing
20681         DECL_VALUE_EXPR.
20683         * tree.c (build_constructor_va): New.
20684         * tree.h: Declare it.
20686 2013-05-09  Martin Jambor  <mjambor@suse.cz>
20688         PR lto/57084
20689         * gimple-fold.c (canonicalize_constructor_val): Call
20690         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
20692 2013-05-09  Jan Hubicka  <jh@suse.cz>
20693             Richard Biener  <rguenther@suse.de>
20695         PR lto/54095
20696         * symtab.c (symtab_make_decl_local): Do not add private names.
20698 2013-05-09  Jan Hubicka  <jh@suse.cz>
20700         PR lto/54095
20701         * symtab.c (insert_to_assembler_name_hash): Handle clones.
20702         (unlink_from_assembler_name_hash): Likewise.
20703         (symtab_prevail_in_asm_name_hash, symtab_register_node,
20704         symtab_unregister_node, symtab_initialize_asm_name_hash,
20705         change_decl_assembler_name): Update.
20707 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
20709         * config/aarch64/aarch64.md: New movtf split.
20710         (*movtf_aarch64): Update.
20711         (aarch64_movdi_tilow): Handle TF modes and rename to
20712         aarch64_movdi_<mode>low.
20713         (aarch64_movdi_tihigh): Handle TF modes and rename to
20714         aarch64_movdi_<mode>high
20715         (aarch64_movtihigh_di): Handle TF modes and rename to
20716         aarch64_mov<mode>high_di
20717         (aarch64_movtilow_di): Handle TF modes and rename to
20718         aarch64_mov<mode>low_di
20719         (aarch64_movtilow_tilow): Remove spurious whitespace.
20720         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
20721         splits.
20722         (aarch64_print_operand): Update.
20724 2013-05-09  Alan Modra  <amodra@gmail.com>
20726         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
20727         powerpc64le.
20728         * configure: Regenerate.
20730 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
20732         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
20733         splitter preparation statements.
20734         * config/i386/sse.md (*vec_extract* splitters): Ditto.
20735         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
20736         adjust_address_nv.
20738 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20740         * gimple-ssa-strength-reduction.c (count_candidates): Change
20741         return value to int.
20742         (analyze_candidates_and_replace): Change type of length to int.
20744 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
20746         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
20747         (*vec_extract<mode>): Use VI12_128 mode iterator.
20748         (*vec_extract<mode>_mem): Ditto.
20749         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
20750         attribute.
20752 2013-05-08  Diego Novillo  <dnovillo@google.com>
20754         PR bootstrap/54659
20756         Revert:
20757         2012-08-17  Diego Novillo  <dnovillo@google.com>
20759         PR bootstrap/54281
20760         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
20761         * config.in: Regenerate.
20762         * configure: Regenerate.
20763         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
20765 2013-05-08  Jan Hubicka  <jh@suse.cz>
20767         PR lto/54095
20768         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
20769         * cgraph.h (symtab_node_base): Add unique_name.
20770         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
20771         input_overwrite_node, input_varpool_node): Stream unique_name.
20772         * cgraphclones.c (cgraph_create_virtual_clone,
20773         cgraph_function_versioning): Set unique_name.
20774         * ipa.c (function_and_variable_visibility): Set unique_name.
20776 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20778         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
20779         (alloc_cand_and_find_basis): Restrict conditional candidate
20780         processing to CAND_MULTs.
20782 2013-05-08  Jan Hubicka  <jh@suse.cz>
20784         PR lto/54095
20785         lto-symtab.c (lto_symtab_symbol_p): New function.
20786         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
20787         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
20788         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
20789         Skip static symbols.
20791 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
20793         PR tree-optimization/57200
20794         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
20795         Only call inform if the preceding warning_at returns true.
20797 2013-05-07  Han Shen  <shenhan@google.com>
20799         * cfgexpand.c (record_or_union_type_has_array_p): New function.
20800         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
20801         * common.opt (fstack-protector-strong): New option.
20802         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
20803         * doc/invoke.texi (Optimization Options): Document
20804         "-fstack-protector-strong".
20805         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
20807 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
20809         * config/mips/mips.c (mips_machine_reorg2): Return 0.
20811 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
20813         * ira.c (update_equiv_regs): Add insn having equiv memory even if
20814         it is not lhs of the insn.
20815         (setup_reg_equiv): Remove insn having equiv memory which it is not
20816         lhs of the insn.
20817         * lra-constraints.c (process_address): Try to improve generation
20818         code for address base + disp.
20819         (lra_constraints): Make correct the code for checking insn setting
20820         up backward equivalence.  Remove insn only if it is in the init
20821         insn list.
20822         * lra-eliminations.c (update_reg_eliminate): Change return value.
20823         (lra_eliminate): Use the result.
20825 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
20827         * config/i386/sse.md (ssescalarnummask): New mode attribute.
20828         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
20829         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
20830         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
20831         register target operands.
20832         (*vec_extractv8hi_sse2): New pattern.
20833         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
20834         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
20835         (*vec_extract<mode>_mem): New insn and split pattern.
20837 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
20839         * config/arm/arm.c (arm_asan_shadow_offset): New function.
20840         (TARGET_ASAN_SHADOW_OFFSET): Define.
20841         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
20842         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
20844 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20846         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
20847         (incr_vec_index): Return -1 if increment not found.
20848         (create_add_on_incoming_edge): Assert if increment not found.
20849         (record_increment): Limit number of increments recorded.
20850         (all_phi_incrs_profitable): Return false if an increment not found.
20851         (replace_profitable_candidates): Don't process increments that were
20852         not recorded.
20853         (analyze_candidates_and_replace): Limit size of incr_vec.
20855 2013-05-07  Richard Biener  <rguenther@suse.de>
20857         * calls.c (special_function_p): setjmp-like functions are leaf.
20858         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
20859         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
20861 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
20863         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
20864         (aarch64_simd_mov<mode>): New expander.
20865         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
20866         (aarch64_simd_mov_to_<mode>high): Likewise.
20867         (aarch64_simd_mov_from_<mode>low): Likewise.
20868         (aarch64_simd_mov_from_<mode>high): Likewise.
20869         (aarch64_dup_lane<mode>): Update.
20870         (aarch64_dup_lanedi): New instruction pattern.
20871         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
20872         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
20874 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20876         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
20877         (replace_mult_candidate): Remove unnecessary argument; remove
20878         unnecessary parameter from call to introduce_cast_before_cand.
20879         (replace_unconditional_candidate): Remove unnecessary parameter
20880         from call to replace_mult_candidate.
20881         (replace_conditional_candidate): Likewise.
20882         (insert_initializers): Use make_temp_ssa_name.
20883         (introduce_cast_before_cand): Remove unnecessary argument; use
20884         make_temp_ssa_name.
20885         (replace_one_candidate): Remove unnecessary argument; remove
20886         unnecessary parameter from calls to introduce_cast_before_cand.
20887         (replace_profitable_candidates): Remove unnecessary parameters
20888         from calls to replace_one_candidate.
20890 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20892         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
20893         phi def as possibly hiding a basis for a CAND_ADD whose operands
20894         have been commuted in the analysis.
20895         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
20897 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20899         * config/aarch64/aarch64.md
20900         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
20901         shift value between 0-4.
20903 2013-05-07  Richard Biener  <rguenther@suse.de>
20905         * double-int.h (rshift): New overload.
20906         * double-int.c (rshift): New function.
20907         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
20908         (create_reference_ops_from_ref): Remove.
20909         (vn_reference_insert): Use shared ops for constructing the
20910         reference and copy it.
20912 2013-05-07  Richard Biener  <rguenther@suse.de>
20914         PR middle-end/57190
20915         * tree-eh.c (sink_clobbers): Properly propagate
20916         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
20918 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
20920         PR tree-optimization/57149
20921         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
20922         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
20923         collect_phi_def_edges, execute_late_warn_uninitialized): Use
20924         uninit_undefined_value_p instead of ssa_undefined_value_p.
20926         PR debug/57184
20927         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
20928         for modifier == EXPAND_INITIALIZER.
20930 2013-05-07  Anton Blanchard  <anton@samba.org>
20932         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
20933         for powerpc64 little endian.
20934         * configure: Regenerate.
20936 2013-05-06  Graham Stott  <grahams@btinternet.com>
20938         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
20939         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
20940         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
20941         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
20943 2013-05-06  Graham Stott  <grahams@btinternet.com>
20945         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
20946         codes which allow non-lvalues.
20948 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
20950         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
20951         components are all 1s.
20952         (integer_minus_onep): New function.
20953         * tree.h (integer_minus_onep): Declare it.
20954         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
20955         integer_minus_onep instead of integer_all_onesp.
20957 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
20959         PR target/52933
20960         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
20961         variations of these patterns.
20963 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
20965         * config/i386/i386.md (isa): Add x64_sse4 member.
20966         (enabled): Handle x64_sse4.
20967         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
20968         instruction for 64bit SSE4_1 targets.  Update insn attributes.
20969         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
20970         instruction for SSE4_1 targets.  Update insn attributes.
20971         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
20972         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
20973         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
20974         const_1 selector.
20975         (*vec_extractv4si): Rename from *sse4_1_pextrd.
20976         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
20977         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
20979 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
20981         PR target/57108
20982         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
20984 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
20986         * final.c (do_assembler_dialects): Don't handle curly braces and
20987         vertical bar escaped by % as dialect delimiters.
20988         (output_asm_insn): Print curly braces and vertical bar if escaped
20989         by % and ASSEMBLER_DIALECT defined.
20990         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
20991         * doc/tm.texi: Regenerated.
20993 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
20995         * config/mips/mips.c: Include tree-pass.h.
20996         (mips_reorg): Split in pre- and post-dbr_schedule parts.
20997         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
20998         (pass_mips_machine_reorg2): New machine specific pass.
20999         (insert_pass_mips_machine_reorg2): New pass plugin definition.
21000         (mips_option_override): Register the new pass.
21001         * rtl.h (cleanup_barriers): Remove prototype.
21002         (dbr_schedule): Likewise.
21003         * jump.c (cleanup_barriers): Make static.
21004         * reorg.c (dbr_schedule): Likewise.
21006 2013-05-06  Richard Biener  <rguenther@suse.de>
21008         PR tree-optimization/57185
21009         * tree-parloops.c (add_field_for_reduction): Handle anonymous
21010         SSA names properly.
21012 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
21014         PR target/57106
21015         * config/i386/i386.c (add_parameter_dependencies): Add dependence
21016         between "first_arg" and "insn", not "last" and "insn".
21018 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
21020         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
21021         (find_candidates_in_block): Re-enable slsr_process_phi.
21022         (create_phi_basis): Fix double counting of candidate adjustment.
21024 2013-05-06  Richard Biener  <rguenther@suse.de>
21026         PR middle-end/57147
21027         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
21028         the edge is also fallthru, preserve it and just clear the
21029         abnormal flag.
21030         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
21031         also complex, preserve that and just clear the fallthru flag.
21032         * tree-inline.c (update_ssa_across_abnormal_edges): Also
21033         update virtual operands.
21035 2013-05-06  Alan Modra  <amodra@gmail.com>
21037         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
21038         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
21039         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
21040         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
21041         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
21042         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
21044 2013-05-06  Alan Modra  <amodra@gmail.com>
21046         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
21047         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
21048         (DEFAULT_ASM_ENDIAN): Define.
21049         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
21050         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
21051         Update -K PIC clause from sysv4.h.
21052         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
21053         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
21055 2013-05-06  Alan Modra  <amodra@gmail.com>
21057         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
21058         twice for little-endian.
21059         (ashrdi3_no_power, ashrdi3): Support little-endian.
21061 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
21063         PR target/55303
21064         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
21065         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
21066         related expanders.
21067         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
21068         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
21069         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
21070         New predicates.
21072 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
21073             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21075         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
21076         * config/pa/pa.opt: Make mbig-switch a no-op.
21077         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
21078         (CASE_VECTOR_MODE): Always return SImode.
21079         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
21080         for the !TARGET_BIG_SWITCH case.
21081         * config/pa/pa-linux.h: Likewise.
21082         * config/pa/pa-openbsd.h: Likewise.
21083         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
21084         * config/pa/pa.md (short_jump): Remove define_insn.
21085         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
21086         (casesi0): Remove define_insn.
21087         (type): Remove btable_branch.
21088         (pa_combine_type): Likewise.
21089         (in_nullified_branch_delay): Likewise.
21090         (in_call_delay): Likewise.
21091         (define_delay): Likewise.
21092         (define_insn_reservation "Z3"): Likewise.
21093         (define_insn_reservation "Z4"): Likewise.
21094         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
21095         (pa_adjust_insn_length): Remove adjustment for btable branches.
21096         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
21097         and mno-big-switch
21099 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
21101         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
21102         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
21103         Add m->r,x alternatives.
21104         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
21105         splitters using SWI48x mode iterator.
21106         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
21107         TARGET_64BIT.  Add m->x alternative.
21108         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
21109         Add o->x alternative.  Enable for TARGET_SSE.
21110         (sse_storeq): Remove expander.
21111         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
21112         with memory input operand.
21113         (*vec_extractv2di_1 splitter): New.
21114         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
21115         * config/i386/i386.md (ssevecmodelower): New mode attribute.
21117 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
21119         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
21120         (INT_LOWPART): Delete.
21121         (extract_MB): Adjust.
21122         (extract_ME): Adjust.
21123         (print_operand): Adjust.
21125 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
21127         * config/rs6000/predicates.md (reg_or_add_cint_operand,
21128         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
21129         (reg_or_logical_cint_operand, easy_fp_constant,
21130         logical_const_operand): Delete "CONST_DOUBLE" case.
21131         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
21132         "HOST_BITS_PER_WIDE_INT == 64" test.
21133         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
21134         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
21135         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
21136         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
21137         test.
21138         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
21139         CONST_DOUBLE DImode/VOIDmode case.
21140         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
21141         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
21142         CONST_DOUBLE VOIDmode case.
21143         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
21144         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
21145         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
21146         Delete CONST_DOUBLE case.
21147         (splitters for mov FMOVE64 const_double): Delete
21148         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
21149         "HOST_BITS_PER_WIDE_INT >= 64" test.
21150         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
21151         case.
21152         (mov DI const_double): Delete.
21154 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
21156         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
21157         on op shows all bits zero in mode of a lowpart subreg, return zero.
21159 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
21161         PR target/57150
21162         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
21163         to save TFmode registers and DImode to save TImode registers for
21164         caller save operations.
21165         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
21166         mark being partially clobbered since they only use the first
21167         double word.
21169         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
21170         and TDmode only use the upper 64-bits of each VSX register.
21172 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
21174         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
21175         (find_candidates_in_block): Disable slsr_process_phi.
21177 2013-05-03  Guozhi Wei  <carrot@google.com>
21179         * coverage.c (coverage_obj_init): Move the construction of gcov
21180         constructor to ...
21181         (build_init_ctor): ... here.
21183 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
21185         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
21186         (slsr_cand_d): Redefine def_phi.
21187         (stride_status, phi_adjust_status, count_phis_status): New enums.
21188         (find_phi_def): New.
21189         (find_basis_for_base_expr): New.
21190         (find_basis_for_candidate): Handle hidden bases.
21191         (alloc_cand_and_find_basis): Handle phi candidates.
21192         (slsr_process_phi): New.
21193         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
21194         (create_mul_imm_cand): Likewise.
21195         (create_add_ssa_cand): Exclude phi base candidates.
21196         (create_add_imm_cand): Likewise.
21197         (slsr_process_cast): Likewise.
21198         (slsr_process_copy): Likewise.
21199         (find_candidates_in_block): Handle phi candidates.
21200         (dump_candidate): Likewise.
21201         (unconditional_cands): Delete.
21202         (unconditional_cands_with_known_stride_p): Delete.
21203         (phi_dependent_cand_p): New.
21204         (cand_increment): Handle phi-dependent candidates.
21205         (replace_dependent): Delete.
21206         (replace_mult_candidate): New.
21207         (replace_unconditional_candidate): New.
21208         (incr_vec_index): Move to avoid forward reference.
21209         (create_add_on_incoming_edge): New.
21210         (create_phi_basis): New.
21211         (replace_dependents): Delete.
21212         (replace_conditional_candidate): New.
21213         (phi_add_costs): New.
21214         (replace_uncond_cands_and_profitable_phis): New.
21215         (record_increment): Handle phi adjustments.
21216         (record_phi_increments): New.
21217         (record_increments): Handle phi adjustments.
21218         (phi_incr_cost): New.
21219         (lowest_cost_path): Handle phis.
21220         (total_savings): Likewise.
21221         (analyze_increments): Likewise.
21222         (ncd_with_phi): New.
21223         (ncd_of_cand_and_phis): New.
21224         (nearest_common_dominator_for_cands): Handle phi increments.
21225         (all_phi_incrs_profitable): New.
21226         (replace_profitable_candidates): Handle phi-dependent candidates.
21227         (analyze_candidates_and_replace): Likewise.
21229 2013-05-03  Teresa Johnson  <tejohnson@google.com>
21231         PR bootstrap/57154
21232         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
21233         do not exceed REG_BR_PROB_BASE.
21235 2013-05-03  Jeff Law  <law@redhat.com>
21237         PR tree-optimization/57144
21238         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
21239         operand of the condition will bit into the new type when eliminating
21240         a cast feeding a condition.
21242 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
21244         PR rtl-optimization/57130
21245         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
21246         of COMPARE as in_code to the recursive call if needed.
21248 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
21250         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
21251         (enabled): Handle new members.
21252         * config/i386/sse.md (*vec_concatv2si): Merge from
21253         *vec_concatv2si_sse2 and vec_concatv2si_sse.
21254         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
21256 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
21258         PR tree-optimization/57027
21259         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
21260         for fnms opportunity, check we got the prerequisite kind
21261         of tree / gimple before using accessor functions.
21263 2013-05-03  Richard Biener  <rguenther@suse.de>
21265         * double-int.h (lshift): New overload without precision
21266         and arith argument.
21267         (operator *=, operator +=, operator -=): Move ...
21268         * double-int.c (operator *=, operator +=, operator -=): ... here
21269         and implement more efficiently.
21270         (mul_double_with_sign): Remove.
21271         (lshift_double): Adjust to take unsinged shift argument, push
21272         dispatching code to callers.
21273         (mul_double_wide_with_sign): Add early out for callers that
21274         are not interested in high parts or overflow.
21275         (lshift): New function.
21276         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
21277         dispatch code here.
21278         (lrotate, rrotate): Use logical shifts.
21279         * expr.c (get_inner_reference): Use lshift.
21280         * fixed-value.c (do_fixed_divide): Likewise.
21281         * tree-dfa.c (get_ref_base_and_extent): Likewise.
21282         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
21283         (indirect_refs_may_alias_p): Likewise.
21284         (stmt_kills_ref_p_1): Likewise.
21286 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
21288         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
21290 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
21292         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
21293         scalar form of FABD instruction.
21295 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
21297         * lra-constraints.c (process_alt_operands): Add checking alt
21298         number to choose the best alternative.
21300 2013-05-02  Richard Biener  <rguenther@suse.de>
21302         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
21303         bitmap and its handling.
21304         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
21306 2013-05-02  Richard Biener  <rguenther@suse.de>
21308         PR middle-end/57140
21309         * tree-inline.c (copy_loops): Properly handle removed loops.
21310         (copy_cfg_body): Mark destination loops for fixup if source
21311         loops needed fixup.
21313 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
21315         PR target/56732
21316         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
21317         generating simple_return for naked functions.
21319 2013-05-02  Martin Jambor  <mjambor@suse.cz>
21321         PR middle-end/56988
21322         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
21323         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
21324         flags match.
21325         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
21326         ipa_agg_replacement_value structures.
21327         (known_aggs_to_agg_replacement_list): Likewise.
21328         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
21329         (read_agg_replacement_chain): Likewise.
21330         (ipcp_transform_function): Also check that by_ref flags match.
21332 2013-05-02  Richard Biener  <rguenther@suse.de>
21334         * graphds.h (struct graph): Add obstack member.
21335         * graphds.c (new_graph): Initialize obstack and allocate
21336         vertices from it.
21337         (add_edge): Allocate edge from the obstack.
21338         (free_graph): Free the obstack instead of all edges and vertices.
21340 2013-05-02  Teresa Johnson  <tejohnson@google.com>
21342         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
21343         divides.
21344         * cfg.c (update_bb_profile_for_threading): Ditto.
21345         * tree-inline.c (copy_bb): Ditto.
21346         (copy_edges_for_bb): Ditto.
21347         (initialize_cfun): Ditto.
21348         (copy_cfg_body): Ditto.
21349         (expand_call_inline): Ditto.
21350         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
21351         (estimate_node_size_and_time): Ditto.
21352         (inline_merge_summary): Ditto.
21353         * cgraphclones.c (cgraph_clone_edge): Ditto.
21354         (cgraph_clone_node): Ditto.
21355         * sched-rgn.c (compute_dom_prob_ps): Ditto.
21356         (compute_trg_info): Ditto.
21358 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
21360         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
21361         S reg when fp attribute set.
21362         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
21364 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
21366         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
21367         New pattern.
21368         (*and_one_cmplsi3_compare0_uxtw): Likewise.
21369         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
21370         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
21372 2013-05-02  Richard Biener  <rguenther@suse.de>
21374         * tree-scalar-evolution.c (scev_info_hasher): Remove.
21375         (struct instantiate_cache_entry): New type.
21376         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
21377         (struct instantiate_cache_type): New type.
21378         (set_instantiated_value, get_instantiated_value): Remove.
21379         (get_instantiated_value_entry): New function.
21380         (instantiate_scev_name): Use the new cache and adjust.
21381         (instantiate_scev_poly): Adjust.
21382         (instantiate_scev_binary): Likewise.
21383         (instantiate_array_ref): Likewise.
21384         (instantiate_scev_convert): Likewise.
21385         (instantiate_scev_not): Likewise.
21386         (instantiate_scev_3): Likewise.
21387         (instantiate_scev_2): Likewise.
21388         (instantiate_scev_r): Likewise.
21389         (instantiate_scev): Likewise.
21390         (resolve_mixers): Likewise.
21392 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
21394         PR target/57091
21395         * lra-constraints.c (best_small_class_operands_num): Remove.
21396         (process_alt_operands): Remove small_class_operands_num.  Take
21397         small classes operands into losers and only if the operand is not
21398         matched.  Modify debugging output.
21399         (curr_insn_transform): Remove best_small_class_operands_num.
21400         Print insn name.
21402 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21404         * config/aarch64/aarch64-builtins.c
21405         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
21406         * config/aarch64/aarch64-simd-builtins.def
21407         (reduc_splus_): Add new modes.
21408         (reduc_uplus_): New.
21409         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
21410         (reduc_uplus_v4sf): Likewise.
21411         (reduc_splus_v4sf): Likewise.
21412         (aarch64_addv<mode>): Likewise.
21413         (reduc_uplus_<mode>): Likewise.
21414         (reduc_splus_<mode>): Likewise.
21415         (aarch64_addvv2di): Likewise.
21416         (reduc_uplus_v2di): Likewise.
21417         (reduc_splus_v2di): Likewise.
21418         (aarch64_addvv2si): Likewise.
21419         (reduc_uplus_v2si): Likewise.
21420         (reduc_splus_v2si): Likewise.
21421         (reduc_<sur>plus_<mode>): New.
21422         (reduc_<sur>plus_v2di): Likewise.
21423         (reduc_<sur>plus_v2si): Likewise.
21424         (reduc_<sur>plus_v4sf): Likewise.
21425         (aarch64_addpv4sf): Likewise.
21426         * config/aarch64/arm_neon.h
21427         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
21428         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
21429         add UNSPEC_SADDV, UNSPEC_UADDV.
21430         (SUADDV): New.
21431         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
21433 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21435         * config/aarch64/arm_neon.h
21436         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
21438 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21440         * config/aarch64/aarch64-builtins
21441         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
21443 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21445         * config/aarch64/aarch64-simd-builtins.def
21446         (reduc_smax_): New.
21447         (reduc_smin_): Likewise.
21448         (reduc_umax_): Likewise.
21449         (reduc_umin_): Likewise.
21450         (reduc_smax_nan_): Likewise.
21451         (reduc_smin_nan_): Likewise.
21452         (fmax): Remove.
21453         (fmin): Likewise.
21454         (smax): Update for V2SF, V4SF and V2DF modes.
21455         (smin): Likewise.
21456         (smax_nan): New.
21457         (smin_nan): Likewise.
21458         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
21459         (<su><maxmin><mode>3): ...This, refactor.
21460         (s<maxmin><mode>3): New.
21461         (<maxmin_uns><mode>3): Likewise.
21462         (reduc_<maxmin_uns>_<mode>): Refactor.
21463         (reduc_<maxmin_uns>_v4sf): Likewise.
21464         (reduc_<maxmin_uns>_v2si): Likewise.
21465         (aarch64_<fmaxmin><mode>: Remove.
21466         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
21467         new builtin names.
21468         (vmin<q>_f<32,64>): Likewise.
21469         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
21470         (FMAXMIN): New.
21471         (su): Add mappings for smax, smin, umax, umin.
21472         (maxmin): New.
21473         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
21474         (FMAXMIN): Rename as...
21475         (FMAXMIN_UNS): ...This.
21476         (maxminv): Remove.
21477         (fmaxminv): Likewise.
21478         (fmaxmin): Likewise.
21479         (maxmin_uns): New.
21480         (maxmin_uns_op): Likewise.
21482 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21484         * config/aarch64/arm_neon.h
21485         (vac<ge, gt><sd>_f<32, 64>): Rename to...
21486         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
21487         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
21489 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21491         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
21492         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
21494 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21496         * config/aarch64/aarch64-simd.md
21497         (vcond<mode>_internal): Handle special cases for constant masks.
21498         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
21499         (vcondu<mode><mode>): Likewise.
21500         (vcond<v_cmp_result><mode>): New.
21502 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21504         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
21505         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
21506         * config/aarch64/aarch64-simd-builtins.def
21507         (cmeq): Update to BUILTIN_VALLDI.
21508         (cmgt): Likewise.
21509         (cmge): Likewise.
21510         (cmle): Likewise.
21511         (cmlt): Likewise.
21512         * config/aarch64/arm_neon.h
21513         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
21514         to builtins or C as appropriate.
21516 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21518         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
21519         (cmgeu): ...This.
21520         (cmhi): Rename to...
21521         (cmgtu): ...This.
21522         * config/aarch64/aarch64-simd.md
21523         (simd_mode): Add SF.
21524         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
21525         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
21526         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
21527         (cstore<mode>_neg): ...This.
21528         * config/aarch64/iterators.md
21529         (VALLF): new.
21530         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
21531         (COMPARISONS): New.
21532         (UCOMPARISONS): Likewise.
21533         (optab): Add missing comparisons.
21534         (n_optab): New.
21535         (cmp_1): Likewise.
21536         (cmp_2): Likewise.
21537         (CMP): Likewise.
21538         (cmp): Remove.
21539         (VCMP_S): Likewise.
21540         (VCMP_U): Likewise.
21541         (V_cmp_result): Add DF, SF modes.
21542         (v_cmp_result): Likewise.
21543         (v): Likewise.
21544         (vmtype): Likewise.
21545         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
21547 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
21549         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
21550         define_insn to define_insn_and_split.
21551         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
21552         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
21553         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
21554         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
21555         (thumb2_negscc): Likewise.
21557 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
21559         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
21561 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
21563         * config/arm/thumb2.md: Remove trailing whitespaces.
21565 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
21567         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
21568         Use gen_int_mode rather than GEN_INT.
21570 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
21572         * value-prof.c (stream_in_histogram_value): Remove the strayed
21573         debug_gimple_stmt.
21575 2013-04-30  Richard Biener  <rguenther@suse.de>
21577         PR middle-end/57122
21578         * cfghooks.c (split_edge): Properly check for the loop latch edge.
21580 2013-04-30  Richard Biener  <rguenther@suse.de>
21582         PR middle-end/57107
21583         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
21585 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
21587         PR rtl-optimization/56957
21588         PR rtl-optimization/57105
21589         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
21590         variable.  Use just INSN_UID for determining whether an insn
21591         should be only disconnected from the insn stream.
21592         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
21594 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
21596         PR tree-optimization/57104
21597         * tsan.c (instrument_expr): Don't instrument accesses to
21598         DECL_HARD_REGISTER VAR_DECLs.
21600 2013-04-30  Richard Biener  <rguenther@suse.de>
21602         * function.h (loops_for_fn): New inline function.
21603         (set_loops_for_fn): Likewise.
21604         * cfgloop.h (place_new_loop): Add struct function parameter.
21605         (get_loop): Likewise.
21606         (get_loops): Likewise.
21607         (number_of_loops): Likewise.
21608         (fel_next): Adjust.
21609         (fel_init): Likewise.
21610         * cfg.c (get_loop_copy): Adjust.
21611         * cfgloop.c (flow_loops_dump): Likewise.
21612         (record_loop_exits): Likewise.
21613         (verify_loop_structure): Likewise.
21614         * cfgloopanal.c (mark_irreducible_loops): Likewise.
21615         (estimate_reg_pressure_cost): Likewise.
21616         (mark_loop_exit_edges): Likewise.
21617         * cfgloopmanip.c (place_new_loop): Likewise.
21618         (add_loop): Likewise.
21619         (duplicate_loop): Likewise.
21620         * graph.c (draw_cfg_nodes): Likewise.
21621         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
21622         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
21623         (extract_affine_chrec): Likewise.
21624         (build_scop_iteration_domain): Likewise.
21625         * graphite.c (graphite_initialize): Likewise.
21626         * ira-build.c (create_loop_tree_nodes): Likewise.
21627         (more_one_region_p): Likewise.
21628         (rebuild_regno_allocno_maps): Likewise.
21629         (mark_loops_for_removal): Likewise.
21630         (mark_all_loops_for_removal): Likewise.
21631         (remove_unnecessary_regions): Likewise.
21632         (ira_build): Likewise.
21633         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
21634         * loop-init.c (fix_loop_structure): Likewise.
21635         (gate_rtl_move_loop_invariants): Likewise.
21636         (gate_rtl_unswitch): Likewise.
21637         (gate_rtl_unroll_and_peel_loops): Likewise.
21638         (rtl_doloop): Likewise.
21639         * lto-streamer-in.c (input_cfg): Likewise.
21640         * lto-streamer-out.c (output_cfg): Likewise.
21641         * modulo-sched.c (sms_schedule): Likewise.
21642         * predict.c (tree_estimate_probability): Likewise.
21643         (tree_estimate_probability_driver): Likewise.
21644         (estimate_loops): Likewise.
21645         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
21646         (move_sese_region_to_fn): Likewise.
21647         (debug_loop_num): Likewise.
21648         * tree-chrec.c (chrec_evaluate): Likewise.
21649         (hide_evolution_in_other_loops_than_loop): Likewise.
21650         (chrec_component_in_loop_num): Likewise.
21651         (reset_evolution_in_loop): Likewise.
21652         (evolution_function_is_invariant_rec_p): Likewise.
21653         * tree-if-conv.c (main_tree_if_conversion): Likewise.
21654         * tree-inline.c (copy_loops): Likewise.
21655         (copy_cfg_body): Likewise.
21656         (tree_function_versioning): Likewise.
21657         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
21658         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
21659         Likewise.
21660         (add_to_evolution_1): Likewise.
21661         (scev_const_prop): Likewise.
21662         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
21663         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
21664         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
21665         (tree_ssa_lim_initialize): Likewise.
21666         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
21667         (verify_loop_closed_ssa): Likewise.
21668         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
21669         (tree_ssa_loop_im): Likewise.
21670         (tree_ssa_loop_unswitch): Likewise.
21671         (tree_vectorize): Likewise.
21672         (check_data_deps): Likewise.
21673         (tree_ssa_loop_ivcanon): Likewise.
21674         (tree_ssa_loop_bounds): Likewise.
21675         (tree_complete_unroll): Likewise.
21676         (tree_complete_unroll_inner): Likewise.
21677         (tree_parallelize_loops): Likewise.
21678         (tree_ssa_loop_prefetch): Likewise.
21679         (tree_ssa_loop_ivopts): Likewise.
21680         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
21681         * tree-vectorizer.c (vectorize_loops): Likewise.
21683 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
21685         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
21686         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
21687         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
21688         with EABI_LINK_SPEC.
21690 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
21692         PR target/44578
21693         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
21694         alternative.
21696 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
21698         PR target/57097
21699         * lra-constraints.c (process_alt_operands): Discourage a bit more
21700         using memory for pseudos.  Print cost dump for alternatives.
21701         Modify cost values for conflicts with early clobbers.
21702         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
21704 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
21706         PR target/57098
21707         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
21709 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
21711         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
21712         from/to S register.
21713         (movdi_aarch64): Support LDR/STR from/to D register.
21715 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
21717         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
21718         or higher by default.
21720 2013-04-29  Richard Biener  <rguenther@suse.de>
21722         PR middle-end/57075
21723         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
21724         even if not adding abnormal edges for calls that can make
21725         abnormal gotos.
21727 2013-04-29  Richard Biener  <rguenther@suse.de>
21729         PR middle-end/57103
21730         * tree-cfg.c (move_stmt_op): Fix condition under which to update
21731         TREE_BLOCK.
21732         (move_stmt_r): Remove redundant checking.
21734 2013-04-29  Teresa Johnson  <tejohnson@google.com>
21736         PR bootstrap/57077
21737         * basic-block.h (apply_scale): New function.
21738         (apply_probability): Use apply_scale.
21739         * gimple-streamer-in.c (input_bb): Ditto.
21740         * lto-streamer-in.c (input_cfg): Ditto.
21741         * lto-cgraph.c (merge_profile_summaries): Ditto.
21742         * tree-optimize.c (execute_fixup_cfg): Ditto.
21743         * tree-inline.c (copy_bb): Update comment to use apply_scale.
21744         (copy_edges_for_bb): Ditto.
21745         (copy_cfg_body): Ditto.
21747 2013-04-29  Tom de Vries  <tom@codesourcery.com>
21749         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
21750         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
21751         (tail_merge_optimize): Handle current_loops == NULL.
21753 2013-04-26  Jeff Law  <law@redhat.com>
21755         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
21756         (simplify_cond_using_ranges): Generalize code to simplify
21757         COND_EXPRs where one argument is a constant and the other
21758         is an SSA_NAME created by an integral type conversion.
21760 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21762         * config/arm/arm.md (store_minmaxsi): Use only when
21763         optimize_insn_for_size_p.
21765 2013-04-29  Christian Bruel  <christian.bruel@st.com>
21767         PR target/57108
21768         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
21770 2013-04-29  Richard Biener  <rguenther@suse.de>
21772         PR middle-end/57089
21773         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
21774         loop tree make sure to schedule a fixup for the child as well.
21775         (expand_omp_for_generic): Properly add loops.
21776         (expand_omp_for_static_nochunk): Likewise.
21777         (expand_omp_for_static_chunk): Likewise.
21778         (expand_omp_for): For the degenerate case fixup loops.
21779         (expand_omp_sections): Fix default bb placement in loops.
21780         (expand_omp_atomic_pipeline): Properly add loops.
21782 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21784         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
21786 2013-04-29  Tom de Vries  <tom@codesourcery.com>
21788         * tree-ssa-tail-merge.c: Update header comment.
21790 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21792         * config/aarch64/arm_neon.h
21793         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
21794         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
21795         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
21796         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
21797         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
21798         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
21799         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
21800         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
21802 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21804         * config/aarch64/aarch64-simd.md
21805         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
21806         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
21807         fix_trunc, fixuns_trunc.
21808         (ftrunc<VDQF:mode>2): New.
21809         * config/aarch64/iterators.md (optab): Add fix, fixuns.
21810         (fix_trunc_optab): New.
21812 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21814         * config/aarch64/aarch64-builtins.c
21815         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
21816         iceilf, lround, iroundf.
21818 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
21820         PR target/54349
21821         * config/i386/i386.h (enum ix86_tune_indices)
21822         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
21823         New, split from X86_TUNE_INTER_UNIT_MOVES.
21824         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
21825         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
21826         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
21827         (TARGET_INTER_UNIT_MOVES): Remove.
21828         * config/i386/i386.c (initial_ix86_tune_features): Update.
21829         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
21830         (ix86_expand_convert_uns_didf_sse): Use
21831         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
21832         (ix86_expand_vector_init_one_nonzero): Ditto.
21833         (ix86_expand_vector_init_interleave): Ditto.
21834         (inline_secondary_memory_needed): Return true for moves from SSE class
21835         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
21836         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
21837         * config/i386/constraints.md (Yi, Ym): Depend on
21838         TARGET_INTER_UNIT_MOVES_TO_VEC.
21839         (Yj, Yn): New constraints.
21840         * config/i386/i386.md (*movdi_internal): Change constraints of
21841         operand 1 from Yi to Yj and from Ym to Yn.
21842         (*movsi_internal): Ditto.
21843         (*movdf_internal): Ditto.
21844         (*movsf_internal): Ditto.
21845         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
21846         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
21847         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
21848         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
21849         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
21850         * config/i386/sse.md (movdi_to_sse): Ditto.
21851         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
21852         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
21853         TARGET_INTER_UNIT_MOVES.
21854         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
21855         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
21856         instead of TARGET_INTER_UNIT_MOVES.
21857         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
21858         operand 1 from Yi to Yj and from Ym to Yn.
21860 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21862         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
21863         (float_truncate_hi_): Likewise.
21864         (float_extend_lo_): Likewise.
21865         (float_truncate_lo_): Likewise.
21866         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
21867         (aarch64_float_extend_lo_v2df): Likewise.
21868         (vec_unpacks_hi_v4sf): Likewise.
21869         (aarch64_float_truncate_lo_v2sf): Likewise.
21870         (aarch64_float_truncate_hi_v4sf): Likewise.
21871         (vec_pack_trunc_v2df): Likewise.
21872         (vec_pack_trunc_df): Likewise.
21874 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21876         * config/aarch64/aarch64-builtins.c
21877         (aarch64_fold_builtin): Fold float conversions.
21878         * config/aarch64/aarch64-simd-builtins.def
21879         (floatv2si, floatv4si, floatv2di): New.
21880         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
21881         * config/aarch64/aarch64-simd.md
21882         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
21883         * config/aarch64/iterators.md (FLOATUORS): New.
21884         (optab): Add float, floatuns.
21885         (su_optab): Likewise.
21887 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21889         * config/aarch64/aarch64-builtins.c
21890         (aarch64_builtin_vectorized_function): Use new names for
21891         fcvt builtins.
21892         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
21893         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
21894         (fcvtzu): Split as...
21895         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
21896         (fcvtas): Split as...
21897         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
21898         (fcvtau): Split as...
21899         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
21900         (fcvtps): Split as...
21901         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
21902         (fcvtpu): Split as...
21903         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
21904         (fcvtms): Split as...
21905         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
21906         (fcvtmu): Split as...
21907         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
21908         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
21909         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
21910         (lfrintnusf, lfrintnudf): Likewise.
21911         * config/aarch64/aarch64-simd.md
21912         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
21913         define_insn.
21914         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
21915         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
21916         (fcvt_pattern): Likewise.
21918 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21920         * config/aarch64/aarch64-simd.md
21921         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
21922         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
21924 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21926         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
21927         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
21928         (vrnd<a,m,n,p>_f32): Implement using builtins.
21929         (vrnd<i,x><q>_f<32, 64>): New.
21931 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21933         * config/aarch64/aarch64-builtins.c
21934         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
21935         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
21936         (frintz): Rename to...
21937         (btrunc): ...this.
21938         (frintp): Rename to...
21939         (ceil): ...this.
21940         (frintm): Rename to...
21941         (floor): ...this.
21942         (frinti): Rename to...
21943         (nearbyint): ...this.
21944         (frintx): Rename to...
21945         (rint): ...this.
21946         (frinta): Rename to...
21947         (round): ...this.
21948         * config/aarch64/aarch64-simd.md
21949         (aarch64_frint<frint_suffix><mode>): Delete.
21950         (<frint_pattern><mode>2): Convert to insn.
21951         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
21952         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
21953         (frint_pattern): Likewise.
21954         (frint_suffix): Likewise.
21956 2013-04-29  Richard Biener  <rguenther@suse.de>
21958         PR tree-optimization/57081
21959         * loop-init.c: Include tree-flow.h.
21960         (loop_optimizer_finalize): Free number of iteration estimates.
21961         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
21963 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
21965         PR tree-optimization/57083
21966         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
21967         non-singleton shift count range, zero extend low_bound for uns case.
21969         * config/i386/predicates.md (general_vector_operand): New predicate.
21970         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
21971         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
21972         if they aren't nonimmediate operands.  If their original values
21973         satisfy const_vector_equal_evenodd_p, don't shift them.
21974         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
21975         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
21976         if not nonimmediate_operand.
21977         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
21978         instead of register_operand.
21979         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
21981 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
21983         * stor-layout.c (finalize_size_functions): Allocate a structure and
21984         reset cfun before dumping the functions.
21986 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
21988         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
21990         PR target/56866
21991         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
21992         use xop_pmacsdqh if uns_p.
21993         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
21994         the immediate rotate count.
21996 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
21998         * rtl.h (struct rtx_def): Add comment for field jump.
21999         (LRA_SUBREG_P): New macro.
22000         * recog.c (register_operand): Check LRA_SUBREG_P.
22001         * lra.c (lra): Add note at the end of RTL code. Align non-empty
22002         stack frame.
22003         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
22004         (lra_final_code_change): Skip subreg change for operators.
22005         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
22006         if there are no operand changes.
22007         * lra-constraints.c (curr_insn_set): New.
22008         (match_reload): Set LRA_SUBREG_P.
22009         (emit_spill_move): Ditto.
22010         (check_and_process_move): Use curr_insn_set. Process only single
22011         set insns.  Don't initialize sec_mem_p and change_p.
22012         (simplify_operand_subreg): Use LRA_SUBREG_P.
22013         (reg_in_class_p): New function.
22014         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
22015         of #ifdef.  Add code to remove cycling.
22016         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
22017         non-null disp.  Reload inner instead of disp when base and index
22018         are null.  Try to put lo_sum into register.
22019         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
22020         (check_and_process_move): Move code for move cost check to
22021         simple_move_p.  Remove equiv_substitution.
22022         (simple_move_p): New function.
22023         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
22024         curr_insn_set.  Call check_and_process_move only for single set
22025         insns.  Use the new function.  Move call of check_and_process_move
22026         after operand equiv substitution and address process.
22028 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
22030         PR go/57045
22031         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
22032         with nonlocal goto receivers or returns twice calls, ignore
22033         unininitialized values from abnormal edges to nl goto receiver
22034         or returns twice call.
22036 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
22038         PR tree-optimization/57051
22039         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
22040         and VEC_RSHIFT_EXPR if shift count is a multiple of element
22041         bitsize.
22043 2013-04-26  Richard Biener  <rguenther@suse.de>
22045         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
22046         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
22047         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
22048         (fixup_loop_arrays_after_move): New function.
22049         (move_sese_region_to_fn): Properly outline the loop tree parts
22050         of the SESE region.
22052 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
22054         * config/i386/i386.md (type, unit): Fix long lines.
22056 2013-04-26  Richard Biener  <rguenther@suse.de>
22058         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
22059         (lto-streamer-out.o): Likewise.
22060         * cfgloop.c (init_loops_structure): Export, add struct function
22061         argument and adjust.
22062         (flow_loops_find): Adjust.
22063         * cfgloop.h (enum loop_estimation): Add EST_LAST.
22064         (init_loops_structure): Declare.
22065         * lto-streamer-in.c: Include cfgloop.h.
22066         (input_cfg): Input the loop tree.
22067         * lto-streamer-out.c: Include cfgloop.h.
22068         (output_cfg): Output the loop tree.
22069         (output_struct_function_base): Do not drop PROP_loops.
22071 2013-03-26  Richard Biener  <rguenther@suse.de>
22073         * tree-cfg.c (execute_build_cfg): Build the loop tree.
22074         (pass_build_cfg): Provide PROP_loops.
22075         (move_sese_region_to_fn): Remove loops that are outlined into fn
22076         for now.
22077         * tree-inline.c: Include cfgloop.h.
22078         (initialize_cfun): Do not drop PROP_loops.
22079         (copy_loops): New function.
22080         (copy_cfg_body): Copy loop structure.
22081         (tree_function_versioning): Initialize destination loop tree.
22082         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
22083         (pass_parallelize_loops): Do IL verification.
22084         * loop-init.c (loop_optimizer_init): Fixup loops if required.
22085         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
22086         the CFG make sure we fixup loops as well.
22087         * tree-ssa-tail-merge.c: Include cfgloop.h.
22088         (replace_block_by): When merging loop latches mark loops for fixup.
22089         * lto-streamer-out.c (output_struct_function_base): Drop
22090         PROP_loops for now.
22091         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
22092         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
22093         * ipa-split.c: Include cfgloop.h.
22094         (split_function): Add the new return block to the loop tree root.
22095         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
22096         whether we have removed the forwarder block.
22097         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
22098         * cfgloop.h (place_new_loop): Declare.
22099         * cfgloopmanip.c (place_new_loop): Export.
22100         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
22101         (tree-switch-conversion.o): Likewise.
22102         (tree-complex.o): Likewise.
22103         (tree-inline.o): Likewise.
22104         (tree-ssa-tailmerge.o): Likewise.
22105         (ipa-split.o): Likewise.
22106         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
22107         (tree-ssa-copy.o): Likewise.
22108         * tree-switch-conversion.c: Include cfgloop.h
22109         (process_switch): If we emit a bit-test cascade, schedule loops
22110         for fixup.
22111         * tree-complex.c: Include cfgloop.h.
22112         (expand_complex_div_wide): Properly add new basic-blocks to loops.
22113         * asan.c: Include cfgloop.h.
22114         (create_cond_insert_point): Properly add new basic-blocks to
22115         loops, schedule loop fixup.
22116         * cfgloop.c (verify_loop_structure): Check that looks are not
22117         marked for fixup.
22118         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
22119         to loops.
22120         (expand_omp_for_generic): Likewise.
22121         (expand_omp_sections): Likewise.
22122         (expand_omp_atomic_pipeline): Schedule loops for fixup.
22123         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
22124         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
22125         is initialized, not when loops are present.
22126         * tree-parloops.c (parallelize_loops): Remove checking here.
22127         * passes.c (init_optimization_passes): Schedule a copy-propagation
22128         pass before complete unrolling of inner loops.
22130 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
22132         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
22133         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
22134         (colorize_init): Add argument to _WIN32 version.
22135         * toplev.c: Include diagnostic-color.h.
22136         (process_options): Default to -fdiagnostics-color=auto if
22137         GCC_COLORS env var is in the environment.
22138         * common.opt (fdiagnostics-color=): Add Var and Init.
22139         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
22140         env var is in the environment, the default is auto rather than never.
22142         * diagnostic.h (file_name_as_prefix): Add context argument.
22143         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
22144         the string as locus.
22145         * langhooks.c (lhd_print_error_function): Adjust caller.
22147 2013-04-25  Lawrence Crowl  <crowl@google.com>
22149         * var-tracking.c (shared_hash_def::htab):
22150         Change type to hash_table.  Update dependent calls and types.
22152 2013-04-25  Lawrence Crowl  <crowl@google.com>
22154         * Makefile.in: Update as needed below.
22156         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
22157         Move declaration to after the type's method definitons.
22159         * attribs.c (htab_t scoped_attributes::attribute_hash):
22160         Change type to hash_table.  Update dependent calls and types.
22162         * bitmap.c (htab_t bitmap_desc_hash):
22163         Change type to hash_table.  Update dependent calls and types.
22165         * cselib.c (htab_t cselib_hash_table):
22166         Change type to hash_table.  Update dependent calls and types.
22168         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
22169         (hash_string_slot_node): Move implementation into lto-streamer.h
22170         struct string_slot_hasher.
22171         (eq_string_slot_node): Likewise.
22173         * data-streamer-out.c: Update output_block::string_hash_table
22174         dependent calls and types.
22176         * dwarf2cfi.c (htab_t trace_index):
22177         Change type to hash_table.  Update dependent calls and types.
22179         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
22180         Change type to hash_table.  Update dependent calls and types.
22181         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
22182         (htab_t optimize_external_refs::map): Likewise.
22183         (htab_t output_comp_unit::extern_map): Likewise.
22184         (htab_t output_comdat_type_unit::extern_map): Likewise.
22185         (htab_t output_macinfo::macinfo_htab): Likewise.
22186         (htab_t optimize_location_lists::htab): Likewise.
22187         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
22189         * except.c (htab_t ehspec_hash_type):
22190         Change type to hash_table.  Update dependent calls and types.
22191         (assign_filter_values::ttypes): Likewise.
22192         (assign_filter_values::ehspec): Likewise.
22193         (sjlj_assign_call_site_values::ar_hash): Likewise.
22194         (convert_to_eh_region_ranges::ar_hash): Likewise.
22196         * gcse.c (htab_t pre_ldst_table):
22197         Change type to hash_table.  Update dependent calls and types.
22199         * ggc-common.c (htab_t saving_htab):
22200         Change type to hash_table.  Update dependent calls and types.
22201         (htab_t loc_hash): Likewise.
22202         (htab_t ptr_hash): Likewise.
22203         (call_count): Rename ggc_call_count.
22204         (call_alloc): Rename ggc_call_alloc.
22205         (loc_descriptor): Rename make_loc_descriptor.
22206         (add_statistics): Rename ggc_add_statistics.
22208         * ggc-common.c (saving_htab):
22209         Change type to hash_table.  Update dependent calls and types.
22211         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
22212         (push_gimplify_context): Likewise.
22213         (pop_gimplify_context): Likewise.
22214         (struct gimple_temp_hash_elt): Added.
22215         (struct gimplify_hasher): Likewise.
22216         (struct gimplify_ctx.temp_htab):
22217         Change type to hash_table.  Update dependent calls and types.
22219         * gimple-fold.c: Include gimplify-ctx.h.
22221         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
22222         Change type to hash_table.  Update dependent calls and types.
22223         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
22224         avoid potential global name collision.
22226         * gimplify.c: Include gimplify-ctx.h.
22227         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
22228         (htab_t gimplify_ctx::temp_htab):
22229         Update dependent calls and types for new type hash_table.
22230         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
22231         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
22233         * gimplify-ctx.h: New.
22234         (struct gimple_temp_hash_elt): Move from gimplify.c.
22235         (class gimplify_hasher): New.
22236         (struct gimplify_ctx): Move from gimple.h.
22237         (htab_t gimplify_ctx::temp_htab):
22238         Change type to hash_table.  Update dependent calls and types.
22240         * graphite-clast-to-gimple.c: Include graphite-htab.h.
22241         (htab_t ivs_params::newivs_index):
22242         Change type to hash_table.  Update dependent calls and types.
22243         (htab_t ivs_params::params_index): Likewise.
22244         (htab_t print_generated_program::params_index): Likewise.
22245         (htab_t gloog::newivs_index): Likewise.
22246         (htab_t gloog::params_index): Likewise.
22248         * graphite.c: Include graphite-htab.h.
22249         4htab_t graphite_transform_loops::bb_pbb_mapping):
22250         Change type to hash_table.  Update dependent calls and types.
22252         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
22253         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
22254         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
22256         * graphite-dependences.c: Include graphite-htab.h.
22257         (loop_is_parallel_p): Change hash table type of parameter.
22259         * graphite-htab.h: New.
22260         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
22261         (extern find_pbb_via_hash): Move from graphite-poly.h.
22262         (extern loop_is_parallel_p): Move from graphite-poly.h.
22263         (extern get_loop_body_pbbs): Move from graphite-poly.h.
22265         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
22266         (extern loop_is_parallel_p): Move to graphite-htab.h.
22267         (extern get_loop_body_pbbs): Move to graphite-htab.h.
22269         * haifa-sched.c (htab_t delay_htab):
22270         Change type to hash_table.  Update dependent calls and types.
22271         (htab_t delay_htab_i2): Likewise.
22273         * ira-color.c (htab_t allocno_hard_regs_htab):
22274         Change type to hash_table.  Update dependent calls and types.
22276         * ira-costs.c (htab_t cost_classes_htab):
22277         Change type to hash_table.  Update dependent calls and types.
22279         * loop-invariant.c (htab_t merge_identical_invariants::eq):
22280         Change type to hash_table.  Update dependent calls and types.
22282         * loop-iv.c (htab_t bivs):
22283         Change type to hash_table.  Update dependent calls and types.
22285         * loop-unroll.c (htab_t opt_info::insns_to_split):
22286         Change type to hash_table.  Update dependent calls and types.
22287         (htab_t opt_info::insns_with_var_to_expand): Likewise.
22289         * lto-streamer.h (struct string_slot): Move from data-streamer.h
22290         (struct string_slot_hasher): New.
22291         (htab_t output_block::string_hash_table):
22292         Change type to hash_table.  Update dependent calls and types.
22294         * lto-streamer-in.c (freeing_string_slot_hasher): New.
22295         (htab_t file_name_hash_table):
22296         Change type to hash_table.  Update dependent calls and types.
22298         * lto-streamer-out.c: Update output_block::string_hash_table dependent
22299         calls and types.
22301         * lto-streamer.c (htab_t tree_htab):
22302         Change type to hash_table.  Update dependent calls and types.
22304         * omp-low.c: Include gimplify-ctx.h.
22306         * passes.c (htab_t name_to_pass_map):
22307         Change type to hash_table.  Update dependent calls and types.
22308         (pass_traverse): Rename to passes_pass_traverse.
22310         * plugin.c (htab_t event_tab):
22311         Change type to hash_table.  Update dependent calls and types.
22313         * postreload-gcse.c (htab_t expr_table):
22314         Change type to hash_table.  Update dependent calls and types.
22315         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
22317         * sese.c (debug_rename_map_1): Make extern.
22318         (htab_t copy_bb_and_scalar_dependences::rename_map):
22319         Change type to hash_table.  Update dependent calls and types.
22321         * sese.h (extern debug_rename_map): Move to .c file.
22323         * store-motion.c (htab_t store_motion_mems_table):
22324         Change type to hash_table.  Update dependent calls and types.
22326         * trans-mem.c (htab_t tm_new_mem_hash):
22327         Change type to hash_table.  Update dependent calls and types.
22329         * tree-browser.c (htab_t TB_up_ht):
22330         Change type to hash_table.  Update dependent calls and types.
22332         * tree-cfg.c (htab_t discriminator_per_locus):
22333         Change type to hash_table.  Update dependent calls and types.
22335         * tree-complex.c: Include tree-hasher.h
22336         (htab_t complex_variable_components):
22337         Change type to hash_table.  Update dependent calls and types.
22339         * tree-eh.c (htab_t finally_tree):
22340         Change type to hash_table.  Update dependent calls and types.
22342         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
22343         struct int_tree_hasher.
22344         (extern int_tree_map_eq): Likewise.
22345         (uid_decl_map_hash): Removed.
22346         (extern decl_tree_map_eq): Likewise.
22348         * tree-hasher.h: New.
22349         (struct int_tree_hasher): New.
22350         (typedef int_tree_htab_type): New.
22352         * tree-inline.c: Include gimplify-ctx.h.
22354         * tree-mudflap.c: Include gimplify-ctx.h.
22356         * tree-parloops.c: Include tree-hasher.h.
22357         (htab_t eliminate_local_variables_stmt::decl_address):
22358         Change type to hash_table.  Update dependent calls and types.
22359         (htab_t separate_decls_in_region::decl_copies): Likewise.
22361         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
22362         Change type to hash_table.  Update dependent calls and types.
22364         * tree-sra.c (candidates):
22365         Change type to hash_table.  Update dependent calls and types.
22367         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
22368         in tree-flow.h.
22369         (int_tree_map_hash): Likewise.
22371         * tree-ssa-dom.c (htab_t avail_exprs):
22372         Change type to hash_table.  Update dependent calls and types.
22374         * tree-ssa-live.c (var_map_base_init::tree_to_index):
22375         Change type to hash_table.  Update dependent calls and types.
22377         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
22378         Change type to hash_table.  Update dependent calls and types.
22380         * tree-ssa-phiopt.c (seen_ssa_names):
22381         Change type to hash_table.  Update dependent calls and types.
22383         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
22384         Change type to hash_table.  Update dependent calls and types.
22386         * tree-ssa-uncprop.c (equiv):
22387         Change type to hash_table.  Update dependent calls and types.
22389 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
22391         PR rtl-optimization/57003
22392         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
22393         call note_stores with kill_clobbered_value callback again after
22394         killing regs_invalidated_by_call.
22396 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
22398         * config/aarch64/aarch64-simd.md
22399         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
22400         (aarch64_simd_bsl<mode>): Likewise.
22401         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
22403 2013-04-25  Marek Polacek  <polacek@redhat.com>
22405         PR tree-optimization/57066
22406         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
22408 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
22410         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
22412 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
22414         * config/aarch64/aarch64-builtins.c
22415         (aarch64_fold_builtin): New.
22416         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
22417         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
22418         * config/aarch64/aarch64-simd-builtins.def (abs): New.
22419         * config/aarch64/arm_neon.h
22420         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
22422 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
22423             Tejas Belagod  <tejas.belagod@arm.com>
22425         * config/aarch64/aarch64-builtins.c
22426         (aarch64_gimple_fold_builtin): New.
22427         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
22428         * config/aarch64/aarch64-simd-builtins.def (addv): New.
22429         * config/aarch64/aarch64-simd.md (addpv4sf): New.
22430         (addvv4sf): Update.
22431         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
22433 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22435         * config/aarch64/aarch64.md
22436         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
22438 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22440         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
22441         (*ngcsi_uxtw): New pattern.
22443 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22444             Julian Brown  <julian@codesourcery.com>
22446         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
22447         (TB_DREG): Add T_V4HF.
22448         (v4hf_UP): New macro.
22449         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
22450         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
22451         Handle initialisation of V4HF. Adjust initialisation of reinterpret
22452         built-ins.
22453         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
22454         (arm_vector_mode_supported_p): Handle V4HF.
22455         (arm_mangle_map): Handle V4HFmode.
22456         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
22457         * config/arm/arm_neon_builtins.def: Add entries for
22458         vcvtv4hfv4sf, vcvtv4sfv4hf.
22459         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
22460         (neon_vcvtv4hfv4sf): Likewise.
22461         * config/arm/neon-gen.ml: Handle half-precision floating point
22462         features.
22463         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
22464         * config/arm/arm_neon.h: Regenerate.
22465         * config/arm/neon.ml (type elts): Add F16.
22466         (type vectype): Add T_float16x4, T_floatHF.
22467         (type vecmode): Add V4HF.
22468         (type features): Add Requires_FP_bit feature.
22469         (elt_width): Handle F16.
22470         (elt_class): Likewise.
22471         (elt_of_class_width): Likewise.
22472         (mode_of_elt): Refactor.
22473         (type_for_elt): Handle F16, fix error messages.
22474         (vectype_size): Handle T_float16x4.
22475         (vcvt_sh): New function.
22476         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
22477         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
22478         (string_of_mode): Handle V4HF.
22479         * doc/arm-neon-intrinsics.texi: Regenerate.
22481 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
22483         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
22484         format specifier in 'X' case.
22486 2013-04-25  Alan Modra  <amodra@gmail.com>
22488         PR target/57052
22489         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
22490         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
22491         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
22492         Repeat for many other rotate/shift and mask patterns using subregs.
22493         Name lshiftrt insns.
22494         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
22495         on WORDS_BIG_ENDIAN.
22497 2013-04-25  Alan Modra  <amodra@gmail.com>
22499         * config.gcc: Support little-endian powerpc-linux targets.
22500         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
22501         (LINK_OS_LINUX_SPEC): Define.
22502         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
22503         Preserve MASK_LITTLE_ENDIAN.
22504         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
22505         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
22506         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
22507         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
22508         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
22509         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
22510         Correct fp word order for little-endian.  Don't shift toc entries
22511         smaller than a word for little-endian.
22512         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
22513         (bswapdi2 splits): Correct low-part subreg for little-endian.
22514         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
22515         low/high where such is correct only for be.
22516         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
22517         little-endian for -mcall-aixdesc.
22519 2013-04-25  Alan Modra  <amodra@gmail.com>
22521         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
22522         replace_equiv_address_nv.
22524 2013-04-25  Alan Modra  <amodra@gmail.com>
22526         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
22528 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
22530         Revert:
22531         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
22532         * rtl.h (struct rtx_def): ...
22534 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
22536         PR rtl-optimizations/57046
22537         * lra-constraints (split_reg): Set up lra_risky_transformations_p
22538         for multi-reg splits.
22540 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
22542         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
22544 2013-04-24  Sterling Augustine  <saugustine@google.com>
22546         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
22547         (comp_dir_string, debug_str_dwo_section): New.
22548         (DEBUG_STR_DWO_SECTION): Rename to ...
22549         (DEBUG_DWO_STR_SECTION): ... this.
22550         (DEBUG_NORM_STR_SECTION): Delete.
22551         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
22552         (DEBUG_STR_DWO_SECTION_FLAGS): New.
22553         (find_AT_string): Move most logic to ...
22554         (find_AT_string_in_table): ... here.  New.
22555         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
22556         add_skeleton_AT_string.  Delete logic.
22557         (output_skeleton_debug_sections): Remove call to
22558         add_top_level_skeleton_die_attrs.
22559         (add_comp_dir_attribute): Move logic to comp_dir_string.
22560         (dwarf2out_init): Initialize debug_str_dwo_section.
22561         (output_indirect_string): Call find_string_form.
22562         (output_indirect_strings): Rewrite.
22563         (prune_unused_types): Empty skeleton_debug_str_hash.
22564         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
22565         (dwarf2out_finish):  Call output_indirect_strings.
22567 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
22569         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
22571 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
22573         * rtl.h (struct rtx_def): Add comment for field jump.
22574         (LRA_SUBREG_P): New macro.
22575         * recog.c (register_operand): Check LRA_SUBREG_P.
22576         * lra.c (lra): Add note at the end of RTL code. Align non-empty
22577         stack frame.
22578         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
22579         (lra_final_code_change): Skip subreg change for operators.
22580         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
22581         if there are no operand changes.
22582         * lra-constraints.c (curr_insn_set): New.
22583         (match_reload): Set LRA_SUBREG_P.
22584         (emit_spill_move): Ditto.
22585         (check_and_process_move): Use curr_insn_set. Process only single
22586         set insns.  Don't initialize sec_mem_p and change_p.
22587         (simplify_operand_subreg): Use LRA_SUBREG_P.
22588         (reg_in_class_p): New function.
22589         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
22590         of #ifdef.  Add code to remove cycling.
22591         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
22592         non-null disp.  Reload inner instead of disp when base and index
22593         are null.  Try to put lo_sum into register.
22594         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
22595         (check_and_process_move): Move code for move cost check to
22596         simple_move_p.  Remove equiv_substitution.
22597         (simple_move_p): New function.
22598         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
22599         curr_insn_set.  Call check_and_process_move only for single set
22600         insns.  Use the new function.  Move call of check_and_process_move
22601         after operand equiv substitution and address process.
22603 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
22605         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
22606         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
22607         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
22609 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
22611         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
22613 2013-04-24  Marek Polacek  <polacek@redhat.com>
22615         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
22616         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
22617         (select_loops_exit_conditions): Likewise.
22618         (number_of_iterations_for_all_loops): Likewise.
22619         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
22620         (scev_analysis): Likewise.
22622 2013-04-02  Catherine Moore  <clm@codesourcery.com>
22623             Chao-ying Fu  <fu@mips.com>
22625         * config/mips/micromips.md (jraddiusp): New pattern.
22626         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
22627         instruction if possible.
22629 2013-04-24  Alan Modra  <amodra@gmail.com>
22631         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
22633 2013-04-24  Julian Brown  <julian@codesourcery.com>
22634             Chung-Lin Tang  <cltang@codesourcery.com>
22636         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
22637         dependency behavior in enumeration type DIE generation. Add TODO note
22638         to comments about future DW_FORM_sdata/udata re-work of related code.
22640 2013-04-23  Lawrence Crowl  <crowl@google.com>
22642         * Makefile.in: Update as needed below.
22644         * hash-table.h (class hash_table):
22645         Correct many methods with parameter types compare_type to the correct
22646         value_type.  (Correct code was unlikely to notice the change.)
22647         (hash_table::elements_with_deleted) New.
22648         (class hashtable::iterator): New.
22649         (hashtable::begin()): New.
22650         (hashtable::end()): New.
22651         (FOR_EACH_HASH_TABLE_ELEMENT): New.
22653         * statistics.c (statistics_hashes):
22654         Change type to hash_table.  Update dependent calls and types.
22656         * tree-into-ssa.c (var_infos):
22657         Change type to hash_table.  Update dependent calls and types.
22659         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
22660         Change type to hash_table.  Update dependent calls and types.
22662         * tree-ssa-loop-im.c (struct mem_ref.refs):
22663         Change type to hash_table.  Update dependent calls and types.
22665         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
22666         Change type to hash_table.  Update dependent calls and types.
22668         * tree-ssa-sccvn.c (vn_tables_s::nary):
22669         Change type to hash_table.  Update dependent calls and types.
22670         (vn_tables_s::phis): Likewise.
22671         (vn_tables_s::references): Likewise.
22673         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
22674         (vn_reference_eq): Update parameter and return types.
22676         * tree-ssa-structalias.c (pointer_equiv_class_table):
22677         Change type to hash_table.  Update dependent calls and types.
22678         (location_equiv_class_table): Likewise.
22680         * tree-vect-data-refs.c: Consequential changes for making
22681         peeling a hash_table.
22683         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
22684         (destroy_loop_vec_info): Dependent hash_table update.
22686         * tree-vectorizer.h (peeling_htab):
22687         Change type to hash_table.  Update dependent calls and types.
22689 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
22691         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
22692         to check the register content is equal or not.
22693         * lra-constraints.c (match_reload): Use lra_assign_reg_val
22694         to assign register content record.
22695         * lra-eliminations.c (update_reg_eliminate): Use
22696         lra_update_reg_val_offset to update register content offset.
22697         * lra-int.h (struct lra_reg): Add offset member.
22698         (lra_reg_val_equal_p): New static inline function.
22699         (lra_update_reg_val_offset): New static inline function.
22700         (lra_assign_reg_val): New static inline function.
22701         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
22702         to assign register content record.
22703         (initialize_lra_reg_info_element): Initial offset to zero.
22705 2013-04-23  Catherine Moore  <clm@codesourcery.com>
22707         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
22708         operands.  Record compression.
22710 2013-04-23  Xinliang David Li  <davidxl@google.com>
22712         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
22714 2013-04-23  Richard Biener  <rguenther@suse.de>
22716         PR middle-end/57036
22717         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
22718         parameter, only add abnormal goto edges from the copied body
22719         if the call could perform abnormal gotos.
22720         (copy_cfg_body): Adjust.
22722 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
22724         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
22726 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
22728         * coretypes.h (gimple_stmt_iterator): Add struct to make
22729         compatible with C.
22731 2013-04-23  Richard Biener  <rguenther@suse.de>
22733         PR tree-optimization/57026
22734         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
22735         from SSA names occuring in abnormal PHI nodes.
22737 2013-04-22  Andi Kleen  <ak@linux.intel.com>
22739         * lto/lto.c (print_lto_report_1): Fix LTO report names.
22741 2013-04-22  Andi Kleen  <ak@linux.intel.com>
22743         * lto/lto.c (print_lto_report_1): Declare early.
22744         (read_cgraph_and_symbols): Call print_lto_report_1 early.
22746 2013-04-22  Andi Kleen  <ak@linux.intel.com>
22748         * common.opt (-flto-report-wpa): Add.
22749         * doc/invoke.texi (-flto-report-wpa): Add.
22750         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
22751         (lto_main): dito.
22753 2013-04-22  Xinliang David Li  <davidxl@google.com>
22755         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
22756         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
22757         * Makefile.in: New dependency
22759             David Daney <ddaney.cavm@gmail.com>
22761         * configure.ac (gcc_cv_as_micromips_support):  Use the
22762         --fatal-warnings option.
22763         * configure: Regenerate.
22765 2013-04-22  Marek Polacek  <polacek@redhat.com>
22767         PR sanitizer/56990
22768         * tsan.c (instrument_expr): Don't instrument expression
22769         in case its size is zero.
22771 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
22773         PR target/57032
22774         Revert:
22775         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
22777         * config/alpha/alpha.c (TARGET_LRA_P): New define.
22779 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
22781         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
22782         (gimple_stmt_iterator): New typedef.
22783         * gimple.h (gimple_stmt_iterator): Rename to...
22784         (gimple_stmt_iterator_d): ... This.
22785         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
22786         trees be valid for GIMPLE and GENERIC.
22787         (TARGET_GIMPLE_FOLD_BUILTIN): New.
22788         * gimple-fold.c (gimple_fold_call): Call target hook
22789         gimple_fold_builtin.
22790         * hooks.c (hook_bool_gsiptr_false): New.
22791         * hooks.h (hook_bool_gsiptr_false): New.
22792         * target.def (fold_stmt): New.
22793         * doc/tm.texi: Regenerate.
22795 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
22797         PR target/57018
22798         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
22799         a set sp if no stack realignment.
22801 2013-04-22  Nick Clifton  <nickc@redhat.com>
22803         * config.gcc (tilegx-linux): Extend extra_objs rather than
22804         overwriting it.
22805         (tilepro-linux): Likewise.
22807 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
22809         * config/aarch64/aarch64-builtins.c
22810         (CF): Remove.
22811         (CF0, CF1, CF2, CF3, CF4, CF10): New.
22812         (VAR<1-12>): Add MAP parameter.
22813         (BUILTIN_*): Likewise.
22814         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
22815         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
22816         (aarch64_ushl_n<mode>): Likewise.
22817         (aarch64_sshr_n<mode>): Likewise.
22818         (aarch64_ushr_n<mode>): Likewise.
22819         (aarch64_<maxmin><mode>): Likewise.
22820         (aarch64_sqrt<mode>): Likewise.
22821         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
22822         (vshr<q>_n_*): Likewise.
22824 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
22826         * config/aarch64/aarch64-builtins.c
22827         (aarch64_simd_builtin_type_mode): Handle SF types.
22828         (sf_UP): Define.
22829         (BUILTIN_GPF): Define.
22830         (aarch64_init_simd_builtins): Handle SF types.
22831         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
22832         (frecps): Likewise.
22833         (frecpx): Likewise.
22834         * config/aarch64/aarch64-simd.md
22835         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
22836         (aarch64_frecpe<mode>): New.
22837         (aarch64_frecps<mode>): Likewise.
22838         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
22839         (v8type): Add frecp<esx>.
22840         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
22841         (aarch64_frecps<mode>): Likewise.
22842         * config/aarch64/iterators.md (FRECP): New.
22843         (frecp_suffix): Likewise.
22844         * config/aarch64/arm_neon.h
22845         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
22847 2013-04-22  Christian Bruel  <christian.bruel@st.com>
22849         PR target/56995
22850         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
22851         (REG_CLASS_NAMES): Idem.
22852         (REG_CLASS_CONTENTS): Idem.
22853         (REGCLASS_HAS_FP_REG): Idem.
22854         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
22855         (sh_conditional_register_usage): Idem.
22857 2013-04-21  Jeff Law  <law@redhat.com>
22859         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
22860         (ssa_forward_propagate_and_combine): Use it.
22862 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
22864         * lra.c: Update the flow chart diagram.
22866 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
22868         PR rtl-optimization/56847
22869         * lra-constraints.c (process_alt_operands): Discourage alternative
22870         with non-matche doffsettable memory constraint fro memory with
22871         known offset.
22873 2013-04-19  Richard Biener  <rguenther@suse.de>
22875         PR tree-optimization/56982
22876         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
22877         function.
22878         * gimplify.c (gimplify_call_expr): Notice special calls.
22879         (gimplify_modify_expr): Likewise.
22880         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
22881         abnormal control flow receivers.
22882         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
22883         in the same way as cfun->has_nonlocal_labels.
22884         (gimple_purge_dead_abnormal_call_edges): Likewise.
22885         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
22886         receivers start a basic-block.
22888 2013-04-19  Richard Biener  <rguenther@suse.de>
22890         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
22891         member ...
22892         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
22893         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
22894         (SLP_TREE_LOAD_PERMUTATION): Add.
22895         (vect_transform_slp_perm_load): Adjust prototype.
22896         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
22897         (vect_free_slp_instance): Likewise.
22898         (vect_create_new_slp_node): Likewise.
22899         (vect_supported_slp_permutation_p): Remove.
22900         (vect_slp_rearrange_stmts): Adjust.
22901         (vect_supported_load_permutation_p): Likewise.  Inline
22902         vect_supported_slp_permutation_p here.
22903         (vect_analyze_slp_instance): Compute load permutations per
22904         slp node instead of per instance.
22905         (vect_get_slp_defs): Adjust.
22906         (vect_transform_slp_perm_load): Likewise.
22907         (vect_schedule_slp_instance): Remove redundant code.
22908         (vect_schedule_slp): Remove hack for PR56270, add it ...
22909         * tree-vect-stmts.c (vectorizable_load): ... here, do not
22910         CSE loads for SLP.  Adjust.
22912 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
22914         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
22915         spelling in two comments.
22917 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
22919         PR target/56797
22920         * config/arm/arm.c (load_multiple_sequence): Require SP
22921         as base register for loads if SP is in the register list.
22923 2013-04-19  Martin Jambor  <mjambor@suse.cz>
22925         PR tree-optimization/56718
22926         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
22927         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
22928         and made public.  Adjusted all callers.
22929         (ipa_intraprocedural_devirtualization): New function.
22930         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
22931         (ipa_intraprocedural_devirtualization): Likewise.
22932         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
22934 2013-04-19  Richard Biener  <rguenther@suse.de>
22936         PR tree-optimization/57000
22937         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
22939 2013-04-19  Terry Guo  <terry.guo@arm.com>
22941         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
22942         Replace with ...
22943         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
22944         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
22945         (cortex_m4_fmacs): Use new reservations.
22946         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
22948 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
22950         PR rtl-optimization/56999
22951         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
22952         related code.
22953         (lra_coalesce): Remove split_origin_bitmap and related code.
22954         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
22955         ranges if necessary.
22957 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
22959         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
22960         New array.
22961         (ix86_expand_call): Remove clobbered_registers array and use
22962         x86_64_ms_sysv_extra_clobbered_registers instead.
22963         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
22964         Declare here.
22965         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
22966         predicate.
22967         * config/i386/i386.md (*call_rex64_ms_sysv): Use
22968         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
22969         (*call_value_rex64_ms_sysv): Ditto.
22971 2013-04-18  Cary Coutant  <ccoutant@google.com>
22973         * dwarf2out.c (output_pubnames): Check die_perennial_p of
22974         parent instead of die_mark.
22976 2013-04-18  Diego Novillo  <dnovillo@google.com>
22978         * gimple.c (create_gimple_tmp): New.
22979         (get_expr_type): New.
22980         (build_assign): New.
22981         (build_type_cast): New.
22982         * gimple.h (enum ssa_mode): Define.
22983         (gimple_seq_set_location): New.
22984         * asan.c (build_check_stmt): Change some gimple_build_* calls
22985         to use build_assign and build_type_cast.
22987 2013-04-18  Richard Biener  <rguenther@suse.de>
22989         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
22990         handle negative step.  Remove redundant checks.
22991         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
22992         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
22993         for negative step and grouped loads fail to vectorize.
22995 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
22997         * emit-rtl.c (reset_insn_used_flags): New function.
22998         (reset_all_used_flags): Use it.
22999         (verify_insn_sharing): New function.
23000         (verify_rtl_sharing): Fix verification for SEQUENCEs.
23002 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
23004         PR tree-optimization/56984
23005         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
23006         and (x >> M) >= N don't register any assertion if N << M is the
23007         minimum value.
23009 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
23011         * lower-subreg.c (resolve_simple_move): If called self-recursive,
23012         do not delete_insn insns that have not yet been emitted, only
23013         unlink them with remove_insn.
23014         * df-scan.c (df_insn_delete): Revert r197492.
23016 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
23018         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
23019         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
23021 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
23023         * config/arm/arm.md (movsicc_insn): Convert define_insn into
23024         define_insn_and_split.
23025         (and_scc,ior_scc,negscc): Likewise.
23026         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
23028 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
23030         * config/arm/arm.c (use_return_insn): Return 0 for targets that
23031         can benefit from using a sequence of LDRD instructions in epilogue
23032         instead of a single LDM instruction.
23034 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
23036         PR 45688
23037         * doc/extend.texi: Fix typo.
23039 2013-04-17  Richard Biener  <rguenther@suse.de>
23041         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
23042         (vect_build_slp_tree): ... here.
23043         (vect_build_slp_tree_1): Compute which stmts of the SLP group
23044         match.  Remove special-casing of mismatched complex loads.
23045         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
23046         re-try the match with swapped commutative operands.
23047         (vect_supported_load_permutation_p): Remove special-casing of
23048         mismatched complex loads.
23049         (vect_analyze_slp_instance): Adjust.
23051 2013-04-17  Richard Biener  <rguenther@suse.de>
23053         PR rtl-optimization/56921
23054         * cfgloop.h (struct loop): Add simple_loop_desc member.
23055         (struct niter_desc): Mark with GTY(()).
23056         (simple_loop_desc): Do not use aux field but simple_loop_desc.
23057         * loop-iv.c (get_simple_loop_desc): Likewise.
23058         (free_simple_loop_desc): Likewise.
23060         Revert
23061         2013-04-16  Richard Biener  <rguenther@suse.de>
23063         PR rtl-optimization/56921
23064         * loop-init.c (pass_rtl_move_loop_invariants): Add
23065         TODO_do_not_ggc_collect to todo_flags_finish.
23066         (pass_rtl_unswitch): Same.
23067         (pass_rtl_unroll_and_peel_loops): Same.
23068         (pass_rtl_doloop): Same.
23070 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
23072         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
23073         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
23074         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
23075         references.
23076         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
23077         * tree-streamer.c (record_common_node): Adjust reference in comment.
23079 2013-04-17  Terry Guo  <terry.guo@arm.com>
23081         * config/arm/cortex-m4.md: Add a new bypass.
23083 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
23085         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
23086         New pattern.
23087         (*subs_<optab><mode>_multp2): New pattern.
23088         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
23089         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
23091 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
23093         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
23094         (*subs_mul_imm_<mode>): New pattern.
23096 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
23098         PR target/56948
23099         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
23100         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
23101         (vsx_movti_32bit): Change j->wa to O->wa.
23103 2013-04-16  Richard Biener  <rguenther@suse.de>
23105         PR rtl-optimization/56921
23106         * loop-init.c (pass_rtl_move_loop_invariants): Add
23107         TODO_do_not_ggc_collect to todo_flags_finish.
23108         (pass_rtl_unswitch): Same.
23109         (pass_rtl_unroll_and_peel_loops): Same.
23110         (pass_rtl_doloop): Same.
23112 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
23114         * config/arm/arm.c (emit_multi_reg_push): New declaration
23115         for an existing function.
23116         (arm_emit_strd_push): New function.
23117         (arm_expand_prologue): Used here.
23118         (arm_emit_ldrd_pop): New function.
23119         (arm_expand_epilogue): Used here.
23120         (arm_get_frame_offsets): Update condition.
23121         (arm_emit_multi_reg_pop): Add a special case for load of a single
23122         register with writeback.
23124 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
23126         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
23127         description.
23129 2013-04-16  Richard Biener  <rguenther@suse.de>
23131         PR tree-optimization/56756
23132         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
23133         (first_mem_ref_loc): New.
23134         (execute_sm): Place the load temporarily before a previous
23135         access instead of in the latch edge to ensure its SSA dependencies
23136         are defined at points dominating the load.
23138 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
23140         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
23141         correct fix by moving header and footer insn to the footer of
23142         the merged basic block.  Clear BB_END of the merged-away block.
23144         PR middle-end/43631
23145         * emit-rtl.c (make_note_raw): New function.
23146         (link_insn_into_chain): New static inline function.
23147         (add_insn): Use it.
23148         (add_insn_before, add_insn_after): Factor insn chain linking code...
23149         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
23150         using link_insn_into_chain.
23151         (note_outside_basic_block_p): New helper function for emit_note_after
23152         and emit_note_before.
23153         (emit_note_after): Use nobb variant of add_insn_after if the note
23154         should not be contained in a basic block.
23155         (emit_note_before): Use nobb variant of add_insn_before if the note
23156         should not be contained in a basic block.
23157         (emit_note_copy): Use make_note_raw.
23158         (emit_note): Likewise.
23159         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
23160         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
23161         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
23162         the moved barrier the tail of the basic block it follows.
23163         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
23165 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
23167         PR tree-optimization/56962
23168         * gimple-ssa-strength-reduction.c (record_increment): Only set
23169         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
23170         either rhs1 or rhs2 is equal to c->base_expr.
23172 2013-04-15  Richard Biener  <rguenther@suse.de>
23174         PR tree-optimization/56933
23175         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
23176         member.
23177         (GROUP_READ_WRITE_DEPENDENCE): Remove.
23178         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
23179         * tree-vect-data-refs.c (vect_analyze_group_access): Move
23180         dependence check ...
23181         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
23182         ... here.
23183         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
23184         GROUP_READ_WRITE_DEPENDENCE.
23186 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
23188         * emit-rtl.c (reset_all_used_flags): New function.
23189         (verify_rtl_sharing): Call reset_all_used_flags before and after
23190         performing the checks.
23192 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23194         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
23195         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
23196         * config/arm/constraints.md (De): New constraint.
23197         * config/arm/neon.md (anddi3_neon): Delete.
23198         (neon_vand<mode>): Expand to standard anddi3 pattern.
23199         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
23200         Move earlier in the file.
23201         (neon_inv_logic_op2): Likewise.
23202         (arm_anddi_operand_neon): New predicate.
23204 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
23206         * configure.ac (gcc_cv_ld_as_needed): Set
23207         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
23208         Use -z ignore, -z record on *-*-solaris2*.
23209         (HAVE_LD_AS_NEEDED): Update comment.
23210         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
23211         * configure: Regenerate.
23212         * config.in: Regenerate.
23213         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
23214         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
23215         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
23216         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
23217         equivalents.  Fix markup.
23218         * doc/tm.texi: Regenerate.
23220 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
23222         * config/i386/i386.opt: New option mstack-protector-guard=.
23223         * config/i386/i386-opts.h: Add enum stack_protector_guard.
23224         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
23225         TARGET_SSP_TLS_GUARD.
23226         * config/i386/i386.c (ix86_option_override_internal): Set
23227         ix86_stack_protector_guard.
23228         * config/i386/i386.md (stack_protect_set): Enable for
23229         TARGET_SSP_TLS_GUARD only.
23230         (stack_protect_set_<mode>): Ditto.
23231         (stack_protect_test): Ditto.
23232         (stack_protect_test_<mode>): Ditto.
23233         * doc/invoke.texi (i386 Option): Document.
23235 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
23237         PR target/56890
23238         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
23239         (S_MODES): Set H_MODE bit.
23240         (SF_MODES): Set only S_MODE and SF_MODE bits.
23241         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
23242         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
23243         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
23244         <MODE_FLOAT>: Likewise.
23246 2013-04-15  Joey Ye  <joey.ye@arm.com>
23248         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
23250 2013-04-15  Joey Ye  <joey.ye@arm.com>
23252         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
23253         for real far jump.
23254         (thumb_far_jump_used_p): Count instruction size and set
23255         far_jump_used.
23257 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
23259         * reorg.c (fill_simple_delay_slots): Reindent block of code.
23260         * resource.c (mark_target_live_regs): Reformat conditional block.
23262 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
23264         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
23265         notes, they are emitted only just before final.
23266         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
23268 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
23270         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
23271         * cfgrtl.c (delete_insn): Call it here instead.
23272         * lra-spills.c (lra_final_code_change): Use delete_insn.
23273         * haifa-sched.c (sched_remove_insn): Likewise.
23274         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
23275         returning to the nop pool.
23276         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
23277         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
23279 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
23281         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
23282         * doc/tm.texi: Regenerated.
23284 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
23286         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
23287         QImode checks.
23289 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
23291         * df-core.c (df_find_def): Compare register numbers.
23292         (df_find_use): Likewise.
23294 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
23296         PR target/56903
23297         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
23298         lra_in_progress for return.
23300 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
23302         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
23303         define_insn into define_insn_and_split and emit movsicc patterns.
23305 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
23307         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
23309 2013-04-12  Richard Biener  <rguenther@suse.de>
23311         * tree-pass.h (TODO_do_not_ggc_collect): New.
23312         * passes.c (execute_one_ipa_transform_pass): Honor
23313         TODO_do_not_ggc_collect.
23314         (execute_one_pass): Likewise.
23316         Revert
23317         2013-04-10  Richard Biener  <rguenther@suse.de>
23319         * passes.c (init_optimization_passes): Remove reload pass.
23320         * ira.c (do_reload): Merge into ...
23321         (ira): ... this.
23322         (rest_of_handle_reload): Remove.
23323         (pass_reload): Likewise.
23324         * config/i386/i386.c (ix86_option_override): Refer to ira instead
23325         of reload for vzeroupper pass placement.
23327 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
23329         PR tree-optimization/56918
23330         PR tree-optimization/56920
23331         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
23332         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
23333         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
23334         use wide_mul_with_sign method.
23336 2013-04-12  Richard Biener  <rguenther@suse.de>
23338         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
23339         not be considered a gimple constant.
23341 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
23343         * fold-const.c (const_binop): Handle vector shifts by a scalar.
23344         (fold_binary_loc): Call const_binop also for mixed vector-scalar
23345         operations.
23347 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
23348             Jakub Jelinek  <jakub@redhat.com>
23350         * opts.c: Include diagnostic-color.h.
23351         (common_handle_option): Handle OPT_fdiagnostics_color_.
23352         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
23353         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
23354         (diagnostic-color.o): New.
23355         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
23356         (diagnostic_color_rule): New enum.
23357         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
23358         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
23359         the location string.
23360         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
23361         either NULL, or color kind.
23362         * diagnostic-color.c: New file.
23363         * diagnostic-color.h: New file.
23364         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
23365         arguments.
23366         * doc/invoke.texi (-fdiagnostics-color): Document.
23367         * pretty-print.h (pp_show_color): Define.
23368         (struct pretty_print_info): Add show_color field.
23369         * diagnostic.c: Include diagnostic-color.h.
23370         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
23371         macros.  Colorize error:, warning: etc. strings and also the location
23372         string.
23373         (diagnostic_show_locus): Colorize the caret line.
23374         * pretty-print.c: Include diagnostic-color.h.
23375         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
23376         inside of %< %> quotes or quoted through q format modifier.
23378 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
23380         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
23382 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
23384         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
23385         code in CC_NZ mode.
23386         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
23387         pattern.
23389 2013-04-11  Marek Polacek  <polacek@redhat.com>
23391         PR tree-optimization/48184
23392         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
23394 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
23396         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
23397         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
23398         (skip_simple_arithmetic): Tidy up.
23399         * tree.h (skip_simple_constant_arithmetic): Declare.
23401 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
23403         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
23405 2013-04-11  Richard Biener  <rguenther@suse.de>
23407         * tree-vect-loop.c (get_initial_def_for_induction): Properly
23408         generate vector constants.
23410 2013-04-11  Richard Biener  <rguenther@suse.de>
23412         PR tree-optimization/56878
23413         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
23414         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
23415         New function.
23416         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
23417         Prefer to align the DR with the most invariant base address.
23419 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
23421         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
23422         comment.
23424 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
23426         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
23427         floating-point vector comparisons against 0.
23429 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
23431         PR tree-optimization/56899
23432         * fold-const.c (extract_muldiv_1): Apply distributive law
23433         only if TYPE_OVERFLOW_WRAPS (ctype).
23435 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
23437         PR target/56124
23438         * ira-costs.c (scan_one_insn): Check whether the source rtx of
23439         loading has side effect.
23441 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
23443         * config/sparc/sparc.c: Include tree-pass.h.
23444         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
23445         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
23446         head of file.  Change return type.  Split off gate function.
23447         (sparc_gate_work_around_errata): New function.
23448         (pass_work_around_errata): New pass definition.
23449         (insert_pass_work_around_errata) New pass insert definition to
23450         insert pass_work_around_errata just after delayed-branch scheduling.
23451         (sparc_option_override): Insert the pass.
23452         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
23454 2013-04-10  David S. Miller  <davem@davemloft.net>
23456         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
23457         or -mcpu=hypersparc.
23459         * target.def (cstore_mode): New hook.
23460         * target.h: Include insn-codes.h
23461         * targhooks.c: Likewise.
23462         (default_cstore_mode): New function.
23463         * targhooks.h: Declare it.
23464         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
23465         * doc/tm.texi: Rebuild.
23466         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
23467         target hook, rather than inspecting the insn_data.
23468         * config/sparc/sparc.c (sparc_cstore_mode): New function.
23469         (TARGET_CSTORE_MODE): Redefine.
23470         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
23471         result patterns.
23472         * config/sparc/predicates.md (cstore_result_operand): New special
23473         predicate.
23474         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
23475         Use it for operand 0.
23476         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
23477         (*snesi_special): Likewise.
23478         (*snesi_zero): Likewise.
23479         (*seqsi_zero): Likewise.
23480         (*sltu_insn): Likewise.
23481         (*sgeu_insn): Likewise.
23482         (*seqdi_special): Make operand 0 and comparison operation be of
23483         DImode.
23484         (*snedi_special): Likewise.
23485         (*snedi_special_vis3): Likewise.
23486         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
23487         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
23488         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
23489         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
23490         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
23491         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
23492         (*sltu_extend_sp64): Likewise.
23493         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
23494         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
23495         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
23496         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
23497         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
23499 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
23501         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
23502         (aarch64_start_file): Use the new function.
23504 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
23505             Jason Merrill  <jason@redhat.com>
23507         * common.opt: Add -gdwarf.
23508         * opts.c (common_handle_option): Handle it.
23509         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
23511 2013-04-10  Richard Biener  <rguenther@suse.de>
23513         * passes.c (execute_todo): Do not call ggc_collect conditional here.
23514         (execute_one_ipa_transform_pass): But unconditionally here.
23515         (execute_one_pass): And here.
23516         (init_optimization_passes): Remove reload pass.
23517         * tree-pass.h (TODO_ggc_collect): Remove.
23518         (pass_reload): Likewise.
23519         * ira.c (do_reload): Merge into ...
23520         (ira): ... this.
23521         (rest_of_handle_reload): Remove.
23522         (pass_reload): Likewise.
23523         * config/i386/i386.c (ix86_option_override): Refer to ira instead
23524         of reload for vzeroupper pass placement.
23525         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
23526         and todo_flags_finish of all passes.
23528 2013-04-10  Richard Biener  <rguenther@suse.de>
23530         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
23531         first_const_oprnd field, rename first_def_type to first_op_type.
23532         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
23533         (vect_get_and_check_slp_defs): Always use the type of the
23534         operand.  Allow mixed vect_external_def, vect_constant_def types.
23535         (vect_get_constant_vectors): Handle mixed vect_external_def,
23536         vect_constant_def types.
23538 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
23540         PR tree-optimization/55524
23541         * tree-ssa-math-opts.c
23542         (convert_mult_to_fma): Don't use an fms construct
23543         when we don't have an fms operation, but fnma, and it looks
23544         likely that we'll be able to use the latter.
23546 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
23548         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
23549         function.
23550         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
23551         inline fail caused by overwritable functions.
23553 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
23555         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
23556         unnecessary bits in the constant power of two case.
23558 2013-04-10  Richard Biener  <rguenther@suse.de>
23560         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
23561         broken code swapping operands.
23562         (vect_build_slp_tree): Do not compute load permutations here.
23563         (vect_analyze_slp_instance): Compute load permutations here,
23564         after building the SLP tree.
23566 2013-04-09  Christian Bruel  <christian.bruel@st.com>
23568         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
23569         of next/prev_real_insn.
23571 2013-04-09  Jan Hubicka  <jh@suse.cz>
23573         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
23574         Drop aliased parameter.
23575         (function_and_variable_visibility): Do not handle alias pairs.
23576         * cgraph.c (varpool_externally_visible_p): Update prototype.
23577         * varpool.c (varpool_add_new_variable): Update.
23579 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23581         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
23583 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
23585         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
23587         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
23589 2013-04-09  Marek Polacek  <polacek@redhat.com>
23591         PR tree-optimization/48762
23592         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
23594 2013-04-09  Richard Biener  <rguenther@suse.de>
23596         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
23597         dealing with cost.
23598         (vect_build_slp_tree): Likewise.
23599         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
23600         calculating the cost of a SLP instance.
23601         (vect_analyze_slp_instance): Use it from here, after building
23602         the SLP tree.
23604 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
23606         PR middle-end/56883
23607         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
23608         expand_omp_for_static_chunk): Use simple_p = true in
23609         force_gimple_operand_gsi calls when assigning to addressable decls.
23611 2013-04-09  Jeff Law  <law@redhat.com>
23613         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
23614         when the boolean was created by converting a wider object which
23615         had a boolean range.
23617 2013-04-09  Richard Biener  <rguenther@suse.de>
23619         * tree-vectorizer.h (slp_void_p): Remove.
23620         (slp_tree): Typedef before _slp_tree declaration.
23621         (struct _slp_tree): Use a vector of slp_tree as children.
23622         (vect_get_place_in_interleaving_chain): Remove.
23623         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
23624         Move ...
23625         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
23626         and make static.
23627         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
23628         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
23629         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
23630         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
23631         Use slp_node instead of slp_void_p and adjust.
23633 2013-04-09  Richard Biener  <rguenther@suse.de>
23635         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
23636         work that is not necessary.
23638 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
23640         PR tree-optimization/56854
23641         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
23642         forward into clobber stmts if it would change MEM_REF lhs into
23643         non-MEM_REF.
23645 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
23647         * tree.c (type_hash_lookup, type_hash_add): Make static.
23648         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
23650 2013-04-09  Richard Biener  <rguenther@suse.de>
23652         * tree.h (unsave_expr_now): Remove.
23653         * tree-inline.c (mark_local_for_remap_r): Remove.
23654         (unsave_expr_1): Likewise.
23655         (unsave_r): Likewise.
23656         (unsave_expr_now): Likewise.
23657         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
23658         (propagate_tree_value): Likewise.
23660 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
23662         * doc/rtl.texi (sequence): Rewrite documentation to match the
23663         current use of SEQUENCE rtl objects.
23664         * rtl.def (SEQUENCE): Likewise.
23666         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
23667         Update documentation.
23668         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
23669         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
23671         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
23673 2013-04-08  Teresa Johnson  <tejohnson@google.com>
23675         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
23676         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
23677         methods.
23678         (estimate_edge_size_and_time): Add comment to suggest using rounding
23679         methods.
23680         (estimate_node_size_and_time): Ditto.
23681         (remap_edge_change_prob): Use helper rounding divide methods.
23682         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
23683         (gimple_mod_pow2_value_transform): Ditto.
23684         (gimple_mod_subtract_transform): Ditto.
23685         (gimple_ic_transform): Ditto.
23686         (gimple_stringops_transform): Ditto.
23687         * stmt.c (conditional_probability): Ditto.
23688         (emit_case_dispatch_table): Ditto.
23689         * lto-cgraph.c (merge_profile_summaries): Ditto.
23690         * tree-optimize.c (execute_fixup_cfg): Ditto.
23691         * cfgcleanup.c (try_forward_edges): Ditto.
23692         * cfgloopmanip.c (scale_loop_profile): Ditto.
23693         (loopify): Ditto.
23694         (duplicate_loop_to_header_edge): Ditto.
23695         (lv_adjust_loop_entry_edge): Ditto.
23696         * tree-vect-loop.c (vect_transform_loop): Ditto.
23697         * profile.c (compute_branch_probabilities): Ditto.
23698         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
23699         * lto-streamer-in.c (input_cfg): Ditto.
23700         * gimple-streamer-in.c (input_bb): Ditto.
23701         * ipa-cp.c (update_profiling_info): Ditto.
23702         (update_specialized_profile): Ditto.
23703         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
23704         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
23705         rounding methods.
23706         * sched-rgn.c (compute_dom_prob_ps): Ditto.
23707         (compute_trg_info): Ditto.
23708         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
23709         (purge_dead_edges): Ditto.
23710         * loop-unswitch.c (unswitch_loop): Ditto.
23711         * cgraphclones.c (cgraph_clone_edge): Ditto.
23712         (cgraph_clone_node): Ditto.
23713         * tree-inline.c (copy_bb): Ditto.
23714         (copy_edges_for_bb): Ditto.
23715         (initialize_cfun): Ditto.
23716         (copy_cfg_body): Ditto.
23717         (expand_call_inline): Ditto.
23719 2013-04-08  Kai Tietz  <ktietz@redhat.com>
23721         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
23722         TARGET_CYGWIN64 by TARGET_64BIT.
23724 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
23726         * config/epiphany/epiphany.md (GPR_1): New constant.
23727         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
23728         * config/epiphany/epiphany.c (gen_compare_reg):
23729         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
23730         is already in place.
23731         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
23732         Don't require being called during rtl expansion; If y operlaps r0,
23733         return 0.
23734         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
23735         (epiphany_expand_epilogue): Likewise.
23737         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
23738         Don't use CC_FPmode for ORDERED / UNORDERED.
23739         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
23741         * config/epiphany/constraints.md (CnL): New constraint.
23742         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
23743         * config/epiphany/predicates.md (add_operand): Allow 1024.
23745         * config/epiphany/epiphany.md (logical_op): New code iterator.
23746         (op_mnc): New code attribute.
23747         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
23748         (mov_f+1, mov_f+2): New peephole2 patterns.
23750         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
23751         (cstoresi4): Also allow re-use of zero result when doing a NE
23752         comparison to a non-zero operand.
23753         Use (clobber (scratch)) for first insn if the gpr output is not needed.
23755         * config/epiphany/epiphany.md (<insn_opname>v2si3):
23756         Use gen_addsi3_i / gen_subsi3_i.
23758 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
23760         PR c++/34949
23761         PR c++/50243
23762         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
23763         contain anything but clobbers, at most one __builtin_stack_restore,
23764         optionally debug stmts and final resx, and if it has at least one
23765         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
23766         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
23767         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
23768         which isn't defaut definition, remove them.
23769         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
23770         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
23771         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
23772         with MEM_REF LHS with SSA_NAME address.
23774 2013-04-08  Jeff Law  <law@redhat.com>
23776         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
23778 2013-04-08  Richard Biener  <rguenther@suse.de>
23780         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
23781         extra newline.
23782         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
23783         determined vector type.
23784         (vect_analyze_data_refs): Likewise.
23785         (vect_get_new_vect_var): Adjust.
23786         (vect_create_destination_var): Preserve SSA name versions.
23787         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
23788         not dump anything here.
23790 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
23792         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
23793         Add member lr_slot_known.
23794         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
23795         if necessary.
23796         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
23797         Remove code that sets lr_slot_offset according to what a previous
23798         version of epiphany_emit_save_restore used to do.
23799         (epiphany_emit_save_restore): When doing an lr save or restore,
23800         set/verify lr_slot_known and lr_slot_offset.
23802 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
23804         PR target/54338
23805         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
23806         in ALL_REGS.
23808 2013-04-08  Richard Biener  <rguenther@suse.de>
23810         * alias.c (find_base_term): Fix thinko in previous change.
23812 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
23814         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
23815         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
23816         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
23817         if possible to compute val.
23818         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
23819         For QImode integers don't require anything about precision.  Use
23820         const_with_all_bytes_same to find out if the constant doesn't have
23821         repeated bytes in it.
23823 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
23825         * config/s390/s390.c (s390_expand_insv): Only accept insertions
23826         within mode size.
23828 2013-04-08  Marek Polacek  <polacek@redhat.com>
23830         PR rtl-optimization/48182
23831         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
23832         value to 1.
23834 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
23836         PR target/55487
23837         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
23838         nuses, make sure we have a label.
23840 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
23842         PR target/56843
23843         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
23844         (rs6000_emit_swdiv_low_precision): Remove.
23845         (rs6000_emit_swdiv): Rewrite to handle between one and four
23846         iterations of Newton-Raphson generally; modify required number of
23847         iterations for some cases.
23848         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
23850 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
23852         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
23853         set-but-unused variable.
23855         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
23856         basic blocks of released function bodies garbage-collectable.
23858         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
23859         (struct rtl_opt_pass): Add TODO_df_finish.
23861         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
23863 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23865         * config/arm/constraints.md (q): New constraint.
23866         * config/arm/ldrdstrd.md: New file.
23867         * config/arm/arm.md (ldrdstrd.md) New include.
23868         (arm_movdi): Use "q" instead of "r" constraint
23869         for double-word memory access.
23870         (movdf_soft_insn): Likewise.
23871         * config/arm/vfp.md (movdi_vfp): Likewise.
23872         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
23873         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
23874         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
23875         (mem_ok_for_ldrd_strd): Likewise.
23876         (output_move_double): Update assertion.
23878 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23880         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
23882 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23884         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
23885         define_insn_and_split.
23886         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
23888 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23890         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
23891         define_insn_and_split.
23892         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
23893         (shiftsi3_compare): New pattern.
23894         (rrx): New pattern.
23895         * config/arm/unspecs.md (UNSPEC_RRX): New.
23897 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23899         * config/arm/arm.md (negdi_extendsidi): New pattern.
23900         (negdi_zero_extendsidi): Likewise.
23902 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23904         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
23905         define_insn_and_split.
23906         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
23907         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
23909 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23911         * config/arm/arm.md (arm_subdi3): Convert define_insn into
23912         define_insn_and_split.
23913         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
23914         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
23916 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23918         * config/arm/arm.md (subsi3_carryin): New pattern.
23919         (subsi3_carryin_const): Likewise.
23920         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
23921         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
23923 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23925         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
23927 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23929         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
23930         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
23932 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23934         * config/arm/arm.c (arm_expand_builtin): Change fcode
23935         type to unsigned int.
23937 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
23939         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
23941 2013-04-04  Ian Lance Taylor  <iant@google.com>
23943         * doc/standards.texi (Standards): The Go frontend supports the Go 1
23944         language standard.
23946 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
23948         PR middle-end/56729
23949         * df-scan.c (df_insn_delete): Disable failing assert.
23951 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23953         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
23954         New function prototype.
23955         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
23956         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
23957         (arm_builtin_vectorized_function): New function.
23959 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23961         * config/arm/arm_neon_builtins.def: New file.
23962         * config/arm/arm.c (neon_builtin_data): Move contents to
23963         arm_neon_builtins.def.
23964         (enum arm_builtins): Include neon builtin definitions.
23965         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
23966         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
23968 2013-04-04  Marek Polacek  <polacek@redhat.com>
23970         PR tree-optimization/48186
23971         * predict.c (maybe_hot_frequency_p): Return false if
23972         HOT_BB_FREQUENCY_FRACTION is 0.
23973         (cgraph_maybe_hot_edge_p): Likewise.
23975 2013-04-04  Richard Biener  <rguenther@suse.de>
23977         PR tree-optimization/56826
23978         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
23979         more accurately.
23981 2013-04-04  Richard Biener  <rguenther@suse.de>
23983         PR tree-optimization/56213
23984         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
23985         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
23987 2013-04-04  Richard Biener  <rguenther@suse.de>
23989         PR tree-optimization/56837
23990         * tree-loop-distribution.c (classify_partition): For non-zero
23991         values require that the value has the same precision as its
23992         mode to be useful as memset value.
23994 2013-04-03  Nick Clifton  <nickc@redhat.com>
23996         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
23997         (fmssf4): Use fmsf.s on E3V5 architectures.
23998         (fnmasf4): Use fnmaf.s on E3V5 architectures.
23999         (fnmssf4): Use fnmsf.s on E3V5 architectures.
24001 2013-04-03  Jeff Law  <law@redhat.com>
24003         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
24004         (lra-eliminations.o): Likewise.
24006 2013-04-03  Teresa Johnson  <tejohnson@google.com>
24008         * gcov-io.c (compute_working_sets): Moved most of body of old
24009         compute_working_sets here from profile.c.
24010         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
24011         (gcov_working_set_t): Moved typedef here from basic-block.h
24012         (compute_working_set): Declare.
24013         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
24014         (get_working_sets): Renamed from compute_working_set,
24015         replace most of body with call to new compute_working_sets.
24016         (get_exec_counts): Replace call to compute_working_sets
24017         to get_working_sets.
24018         * profile.h (get_working_sets): Renamed from compute_working_set.
24019         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
24020         to get_working_sets.
24021         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
24022         * gcov-dump.c (dump_working_sets): New function.
24024 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
24026         * hwint.c (sext_hwi, zext_hwi): New functions.
24027         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
24028         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
24029         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
24030         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
24031         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
24032         (sext_hwi, zext_hwi): New functions.
24034 2013-04-03  Jeff Law  <law@redhat.com>
24036         PR tree-optimization/56799
24037         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
24038         back test for widening conversion erroneously dropped in prior change.
24040 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
24042         PR target/56809
24043         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
24044         instead of next_real_insn.
24046 2013-04-03  Marek Polacek  <polacek@redhat.com>
24048         PR sanitizer/55702
24049         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
24051 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
24053         PR target/56809
24054         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
24055         next_real_insn.
24056         (thumb1_output_casesi): Likewise.
24057         (thumb2_output_casesi): Likewise.
24059 2013-04-03  Richard Biener  <rguenther@suse.de>
24061         PR tree-optimization/56817
24062         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
24063         Split out ...
24064         (tree_unroll_loops_completely_1): ... new function to manually
24065         walk the loop tree, properly defering outer loops of unrolled
24066         loops to later iterations.
24068 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
24070         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
24071         (vectorizable_load): Likewise.
24072         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
24073         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
24075 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
24077         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
24078         BIT_FIELD_REF.
24080 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
24082         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
24084 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
24086         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
24088 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
24090         PR tree-optimization/56790
24091         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
24092         folding.
24094 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
24096         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
24097         Handle VEC_MERGE.
24098         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
24099         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
24100         equal arguments.
24102 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
24104         PR c/19449
24105         * tree.h (force_folding_builtin_constant_p): New decl.
24106         * builtins.c (force_folding_builtin_constant_p): New variable.
24107         (fold_builtin_constant_p): Fold immediately also if
24108         force_folding_builtin_constant_p.
24110 2013-04-03  Richard Biener  <rguenther@suse.de>
24112         PR tree-optimization/56812
24113         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
24114         DRs of the same interleaving chain are independent.
24116 2013-04-02  Jason Merrill  <jason@redhat.com>
24118         * gdbinit.in (pbb): Use debug fn.
24120 2013-04-02  Lawrence Crowl  <crowl@google.com>
24122         * sese.h (struct ivtype_map_elt_s): Remove unused.
24123         (extern debug_ivtype_map): Remove unused.
24124         (extern eq_ivtype_map_elts): Remove unused.
24125         * sese.c (debug_ivtype_map): Removed unused.
24126         (debug_ivtype_map_1): Removed unused.
24127         (debug_ivtype_elt): Remove unused.
24128         (eq_ivtype_map_elts): Remove unused.
24130 2013-04-02  Kai Tietz  <ktietz@redhat.com>
24132         PR target/52790
24133         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
24134         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
24135         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
24136         function.
24137         (legitimize_pe_coff_symbol): Likewise.
24138         (is_imported_p): New helper-function.
24139         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
24140         for Windows x64 targets.
24141         (ix86_expand_prologue): Optimize for pe-coff targets.
24142         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
24143         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
24144         medium/large code-model.
24145         (legitimize_pic_address): Likewise.
24146         (legitimize_tls_address): Likewise.
24147         (ix86_expand_call): Likewise.
24148         (x86_output_mi_thunk): Likewise.
24149         (get_dllimport_decl): Add new beimport argument.
24150         (construct_plt_address): Don't assert for x64 pe-coff targets.
24151         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
24152         targets.
24153         (SYMBOL_FLAG_STUBVAR): New macro.
24154         (SYMBOL_REF_STUBVAR_P): Likewise.
24155         * config/i386/winnt.c (stub_list): New structure.
24156         (stub_head): New local variable.
24157         (i386_pe_record_stub): New function.
24158         (i386_pe_file_end): Emit refptr-stubs.
24160 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
24162         PR rtl-optimization/56745
24163         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
24164         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
24166         PR c++/34949
24167         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
24168         and both of them are MEM_REFs, just compare first argument for
24169         equality and attempt to deal even with differing offsets.
24171         PR c++/34949
24172         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
24173         of gimple_clobber_p to be MEM_REF.
24174         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
24175         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
24176         after gimplification.
24177         * asan.c (get_mem_ref_of_assignment): Don't instrument
24178         gimple_clobber_p stmts.
24179         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
24180         gimple_clobber_p stmt if they have MEM_REF lhs and
24181         are dead because of another gimple_clobber_p stmt.
24182         * tree-ssa-live.c (clear_unused_block_pointer): Treat
24183         gimple_clobber_p stmts like debug stmts.
24184         (remove_unused_locals): Remove clobbers with MEM_REF lhs
24185         that refer to unused VAR_DECLs or uninitialized values.
24186         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
24187         gimple_clobber_p stmts if they refer to removed parameters.
24188         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
24189         formatting.
24191 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
24193         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
24194         using SWI48 mode attribute.
24196 2013-04-02  Wei Mi  <wmi@google.com>
24198         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
24199         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
24200         *<rotate_insn><mode>3_mask in i386.md.
24202 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
24204         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
24206 2013-04-02  Richard Biener  <rguenther@suse.de>
24208         PR tree-optimization/56778
24209         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
24210         Runtime alias tests are not supported for gather loads.
24211         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
24212         stmts referenced from SSA operands before updating SSA form.
24214 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
24215             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
24217         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
24218         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
24219         * config/arm/cortex-a53.md: New file.
24220         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
24221         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
24222         * config/arm/arm.c (arm_issue_rate): Likewise.
24223         * config/arm/arm-tune.md: Regenerate
24224         * config/arm/arm-tables.opt: Regenerate.
24225         * config/arm/arm-cores.def: Add cortex-a53.
24227 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
24229         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
24230         non-static link.
24232 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
24234         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
24235         scalar load/store operations using B/H registers.
24236         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
24238 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
24240         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
24241         scalar move.
24242         * config/aarch64/aarch64.c
24243         (aarch64_simd_scalar_immediate_valid_for_move): New.
24244         * config/aarch64/aarch64-protos.h
24245         (aarch64_simd_scalar_immediate_valid_for_move): New.
24246         * config/aarch64/constraints.md (Dh, Dq): New.
24247         * config/aarch64/iterators.md (hq): New.
24249 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
24251         * reorg.c (get_branch_condition): Deal with conditional returns.
24252         (fill_simple_delay_slots): Remove dead code dealing with jumps.
24254 2013-04-01  Wei Mi  <wmi@google.com>
24256         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
24257         Truncate operand 2 using %b asm operand modifier.
24258         (*<shift_insn><mode>3_mask): Ditto.
24259         (*<rotate_insn><mode>3_mask): Ditto.
24261 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
24263         PR middle-end/56798
24264         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
24266 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
24268         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
24269         of next_real_insn.
24270         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
24272 2013-03-30  Lawrence Crowl  <crowl@google.com>
24274         * dse.c (clear_alias_sets): Remove never set.
24275         (disqualified_clear_alias_sets): Remove never set.
24276         (clear_alias_mode_pool): Remove never set.
24277         (dse_step0): Remove condition that is never true.
24278         (canon_address): Remove condition that is never true.
24279         (dse_step7): Remove condition that is never true.
24280         (rest_of_handle_dse): Remove condition that is never true.
24281         (rest_of_handle_dse::did_global): Remove never read from above.
24282         (dse_step2_spill): Remove never called from above.
24283         (dse_step5_spill): Remove never called from above.
24285 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
24287         * doc/md.texi (Standard Names) <casesi>: Update documentation for
24288         JUMP_TABLE_DATA changes.
24289         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
24290         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
24291         (Insns) <jump_table_data>: New entry.
24292         * doc/tm.texi: Regenerate.
24294         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
24296         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
24297         for table jump at the end of a basic block using tablejump_p.
24298         * targhooks.c (default_invalid_within_doloop): Likewise.
24299         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
24300         target hook implementation that is identical to the default hook.
24301         (rs6000_invalid_within_doloop): Remove.
24303         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
24304         unused variable from tablejump_p call.
24306         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
24307         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
24308         (INSN_DELETED_P): Likewise.
24309         (emit_jump_table_data): New prototype.
24310         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
24311         after 4th as unused.
24312         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
24313         * sched-vis.c (print_insn): Likewise.
24314         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
24315         insn for compatibility with back ends that use next_active_insn to
24316         identify jump table data.
24317         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
24318         (remove_insn): Likewise.
24319         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
24320         to be emitted.
24321         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
24322         (emit_jump_table_data): New function.
24324         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
24325         basic block, a JUMP_TABLE_DATA never is.
24326         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
24327         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
24328         off from code handling real insns.
24329         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
24330         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
24331         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
24332         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
24333         is not a NONDEBUG_INSN_P.
24334         * ira-costs.c (scan_one_insn): Likewise.
24335         * jump.c (mark_all_labels): Likewise.
24336         (mark_jump_label_1): Likewise.
24337         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
24338         * lra.c (get_insn_freq): Expect all insns reaching here to be in
24339         a basic block.
24340         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
24341         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
24342         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
24343         JUMP_TABLE_DATA_P insns.
24344         (calculate_elim_costs_all_insns): Likewise.
24345         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
24346         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
24347         (delete_output_reload): Code style fixups.
24348         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
24349         insn flags on this non-insn.
24350         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
24351         as scheduling barriers, for pre-change compatibility.
24352         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
24353         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
24355         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
24356         redundant JUMP_TABLE_DATA_P test.
24357         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
24358         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
24359         (frv_for_each_packet): Likewise.
24360         * config/i386/i386.c (min_insn_size): Likewise.
24361         (ix86_avoid_jump_mispredicts): Likewise.
24362         * config/m32r/m32r.c (m32r_is_insn): Likewise.
24363         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
24364         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
24365         (mips16_insn_length): Robustify.
24366         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
24367         (mips16_split_long_branches): Likewise.
24368         * config/pa/pa.c (pa_combine_instructions): Likewise.
24369         * config/rs6000/rs6000.c (get_next_active_insn): Treat
24370         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
24371         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
24372         as contributing to pool range lengths.
24373         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
24374         Remove redundant JUMP_TABLE_DATA_P test.
24375         (sh_loop_align): Likewise.
24376         (split_branches): Likewise.
24377         (sh_insn_length_adjustment): Likewise.
24378         * config/spu/spu.c (get_branch_target): Likewise.
24380 2013-03-29  Jan Hubicka  <jh@suse.cz>
24382         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
24383         gcov streaming; stream hot bb threshold to ltrans.
24384         * predict.c (get_hot_bb_threshold): Break out from ....
24385         (maybe_hot_count_p): ... here.
24386         (set_hot_bb_threshold): New function.
24387         * lto-section-in.c (lto_section_name): Add profile.
24388         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
24389         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
24390         and data-streamer.h
24391         (histogram_entry): New structure.
24392         (histogram, histogram_pool): New global vars.
24393         (histogram_hash): New structure.
24394         (histogram_hash::hash): New method.
24395         (histogram_hash::equal): Likewise.
24396         (account_time_size): New function.
24397         (cmp_counts): New function.
24398         (dump_histogram): New function.
24399         (ipa_profile_generate_summary): New function.
24400         (ipa_profile_write_summary): New function.
24401         (ipa_profile_read_summary): New function.
24402         (ipa_profile): Decide on threshold.
24403         (pass_ipa_profile): Add ipa_profile_write_summary and
24404         ipa_profile_read_summary.
24405         * Makefile.in (ipa.o): Update dependencies.
24406         * lto-streamer.h (LTO_section_ipa_profile): New section.
24408 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
24410         * tree.h (VAR_P): New.
24412 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
24414         PR lto/56777
24415         * doc/invoke.texi ([-fwhole-program]): Fix typo.
24417 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
24419         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
24420         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
24421         (control_flow_insn_p): Likewise.
24422         * cfgrtl.c (duplicate_insn_chain): Likewise.
24423         * final.c (get_attr_length_1): Likewise.
24424         (shorten_branches): Likewise.
24425         (final_scan_insn): Likewise.
24426         * function.c (instantiate_virtual_regs): Likewise.
24427         * gcse.c (insert_insn_end_basic_block): Likewise.
24428         * ira-costs.c (scan_one_insn): Likewise.
24429         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
24430         * lra.c (check_rtl): Likewise.
24431         * reload1.c (elimination_costs_in_insn): Likewise.
24432         * reorg.c (follow_jumps): Likewise.
24434         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
24435         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
24436         (thumb_far_jump_used_p): Likewise.
24437         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
24438         (workaround_speculation): Likewise.
24439         (add_sched_insns_for_speculation): Likewise.
24440         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
24441         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
24442         (frv_for_each_packet): Likewise.
24443         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
24444         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
24445         (final_emit_insn_group_barriers): Likewise.
24446         * config/m32r/m32r.c (m32r_is_insn): Likewise.
24447         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
24448         (mips16_insn_length): Likewise.
24449         * config/pa/pa.c (pa_reorg): Likewise.
24450         (pa_combine_instructions): Likewise.
24451         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
24452         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
24453         (sh_reorg): Likewise.
24454         (split_branches): Likewise.
24455         * config/spu/spu.c (get_branch_target): Likewise.
24457         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
24458         JUMP_TABLE_DATA_P.
24460 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
24462         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
24463         Fix declaration name.
24465 2013-03-28  Lawrence Crowl  <crowl@google.com>
24467         * graphds.h (struct graph.indicies): Remove unused.
24468         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
24469         (SCOP_ORIGINAL_PDDRS): Remove unused.
24470         * sese.h (extern insert_loop_close_phis): Removed unused.
24471         (extern insert_guard_phis): Removed unused.
24472         (extern ivtype_map_elt_info): Removed unused.
24473         (new_ivtype_map_elt): Removed unused.
24474         * sese.c (ivtype_map_elt_info): Removed unused.
24476 2013-03-28  Lawrence Crowl  <crowl@google.com>
24478         * Makefile.in: Add several missing include dependences.
24479         (DUMPFILE_H): New.
24480         (test-dump.o): New.  This object is not added to any executable,
24481         but is present for ad-hoc testing.
24482         * bitmap.c
24483         (debug (const bitmap_head_def &)): New.
24484         (debug (const bitmap_head_def *)): New.
24485         * bitmap.h
24486         (extern debug (const bitmap_head_def &)): New.
24487         (extern debug (const bitmap_head_def *)): New.
24488         * cfg.c
24489         (debug (edge_def &)): New.
24490         (debug (edge_def *)): New.
24491         * cfghooks.c
24492         (debug (basic_block_def &)): New.
24493         (debug (basic_block_def *)): New.
24494         * dumpfile.h
24495         (dump_node (const_tree, int, FILE *)): Correct source file.
24496         * dwarf2out.c
24497         (debug (die_struct &)): New.
24498         (debug (die_struct *)): New.
24499         * dwarf2out.h
24500         (extern debug (die_struct &)): New.
24501         (extern debug (die_struct *)): New.
24502         * gimple-pretty-print.c
24503         (debug (gimple_statement_d &)): New.
24504         (debug (gimple_statement_d *)): New.
24505         * gimple-pretty-print.h
24506         (extern debug (gimple_statement_d &)): New.
24507         (extern debug (gimple_statement_d *)): New.
24508         * ira-build.c
24509         (debug (ira_allocno_copy &)): New.
24510         (debug (ira_allocno_copy *)): New.
24511         (debug (ira_allocno &)): New.
24512         (debug (ira_allocno *)): New.
24513         * ira-int.h
24514         (extern debug (ira_allocno_copy &)): New.
24515         (extern debug (ira_allocno_copy *)): New.
24516         (extern debug (ira_allocno &)): New.
24517         (extern debug (ira_allocno *)): New.
24518         * ira-lives.c
24519         (debug (live_range &)): New.
24520         (debug (live_range *)): New.
24521         * lra-int.h
24522         (debug (lra_live_range &)): New.
24523         (debug (lra_live_range *)): New.
24524         * lra-lives.c
24525         (debug (lra_live_range &)): New.
24526         (debug (lra_live_range *)): New.
24527         * omega.c
24528         (debug (omega_pb_d &)): New.
24529         (debug (omega_pb_d *)): New.
24530         * omega.h
24531         (extern debug (omega_pb_d &)): New.
24532         (extern debug (omega_pb_d *)): New.
24533         * print-rtl.c
24534         (debug (const rtx_def &)): New.
24535         (debug (const rtx_def *)): New.
24536         * print-tree.c
24537         (debug_tree (tree): Move within file.
24538         (debug_raw (const tree_node &)): New.
24539         (debug_raw (const tree_node *)): New.
24540         (dump_tree_via_hooks (const tree_node *, int)): New.
24541         (debug (const tree_node &)): New.
24542         (debug (const tree_node *)): New.
24543         (debug_verbose (const tree_node &)): New.
24544         (debug_verbose (const tree_node *)): New.
24545         (debug_head (const tree_node &)): New.
24546         (debug_head (const tree_node *)): New.
24547         (debug_body (const tree_node &)): New.
24548         (debug_body (const tree_node *)): New.
24549         (debug_vec_tree (tree): Move and reimplement in terms of dump.
24550         (debug (vec<tree, va_gc> &)): New.
24551         (debug (vec<tree, va_gc> *)): New.
24552         * rtl.h
24553         (extern debug (const rtx_def &)): New.
24554         (extern debug (const rtx_def *)): New.
24555         * sbitmap.c
24556         (debug_raw (simple_bitmap_def &)): New.
24557         (debug_raw (simple_bitmap_def *)): New.
24558         (debug (simple_bitmap_def &)): New.
24559         (debug (simple_bitmap_def *)): New.
24560         * sbitmap.h
24561         (extern debug (simple_bitmap_def &)): New.
24562         (extern debug (simple_bitmap_def *)): New.
24563         (extern debug_raw (simple_bitmap_def &)): New.
24564         (extern debug_raw (simple_bitmap_def *)): New.
24565         * sel-sched-dump.c
24566         (debug (vinsn_def &)): New.
24567         (debug (vinsn_def *)): New.
24568         (debug_verbose (vinsn_def &)): New.
24569         (debug_verbose (vinsn_def *)): New.
24570         (debug (expr_def &)): New.
24571         (debug (expr_def *)): New.
24572         (debug_verbose (expr_def &)): New.
24573         (debug_verbose (expr_def *)): New.
24574         (debug (vec<rtx> &)): New.
24575         (debug (vec<rtx> *)): New.
24576         * sel-sched-dump.h
24577         (extern debug (vinsn_def &)): New.
24578         (extern debug (vinsn_def *)): New.
24579         (extern debug_verbose (vinsn_def &)): New.
24580         (extern debug_verbose (vinsn_def *)): New.
24581         (extern debug (expr_def &)): New.
24582         (extern debug (expr_def *)): New.
24583         (extern debug_verbose (expr_def &)): New.
24584         (extern debug_verbose (expr_def *)): New.
24585         (extern debug (vec<rtx> &)): New.
24586         (extern debug (vec<rtx> *)): New.
24587         * sel-sched-ir.h
24588         (_list_iter_cond_expr): Make inline instead of static.
24589         * sreal.c
24590         (debug (sreal &)): New.
24591         (debug (sreal *)): New.
24592         * sreal.h
24593         (extern debug (sreal &)): New.
24594         (extern debug (sreal *)): New.
24595         * tree.h
24596         (extern debug_raw (const tree_node &)): New.
24597         (extern debug_raw (const tree_node *)): New.
24598         (extern debug (const tree_node &)): New.
24599         (extern debug (const tree_node *)): New.
24600         (extern debug_verbose (const tree_node &)): New.
24601         (extern debug_verbose (const tree_node *)): New.
24602         (extern debug_head (const tree_node &)): New.
24603         (extern debug_head (const tree_node *)): New.
24604         (extern debug_body (const tree_node &)): New.
24605         (extern debug_body (const tree_node *)): New.
24606         (extern debug (vec<tree, va_gc> &)): New.
24607         (extern debug (vec<tree, va_gc> *)): New.
24608         * tree-cfg.c
24609         (debug (struct loop &)): New.
24610         (debug (struct loop *)): New.
24611         (debug_verbose (struct loop &)): New.
24612         (debug_verbose (struct loop *)): New.
24613         * tree-dump.c: Add header dependence.
24614         * tree-flow.h
24615         (extern debug (struct loop &)): New.
24616         (extern debug (struct loop *)): New.
24617         (extern debug_verbose (struct loop &)): New.
24618         (extern debug_verbose (struct loop *)): New.
24619         * tree-data-ref.c
24620         (debug (data_reference &)): New.
24621         (debug (data_reference *)): New.
24622         (debug (vec<data_reference_p> &)): New.
24623         (debug (vec<data_reference_p> *)): New.
24624         (debug (vec<ddr_p> &)): New.
24625         (debug (vec<ddr_p> *)): New.
24626         * tree-data-ref.h
24627         (extern debug (data_reference &)): New.
24628         (extern debug (data_reference *)): New.
24629         (extern debug (vec<data_reference_p> &)): New.
24630         (extern debug (vec<data_reference_p> *)): New.
24631         (extern debug (vec<ddr_p> &)): New.
24632         (extern debug (vec<ddr_p> *)): New.
24633         * tree-ssa-alias.c
24634         (debug (pt_solution &)): New.
24635         (debug (pt_solution *)): New.
24636         * tree-ssa-alias.h
24637         (extern debug (pt_solution &)): New.
24638         (extern debug (pt_solution *)): New.
24639         * tree-ssa-alias.c
24640         (debug (_var_map &)): New.
24641         (debug (_var_map *)): New.
24642         (debug (tree_live_info_d &)): New.
24643         (debug (tree_live_info_d *)): New.
24644         * tree-ssa-alias.h
24645         (extern debug (_var_map &)): New.
24646         (extern debug (_var_map *)): New.
24647         (extern debug (tree_live_info_d &)): New.
24648         (extern debug (tree_live_info_d *)): New.
24650 2013-03-28  Jan Hubicka  <jh@suse.cz>
24652         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
24654 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
24656         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
24657         record only when desired or required.
24659 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
24661         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
24662         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
24664 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
24666         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
24667         (*andsi3_compare0_uxtw): New pattern.
24668         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
24669         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
24671 2013-03-28  Jan Hubicka  <jh@suse.cz>
24673         * data-streamer-in.c (streamer_read_gcov_count): New function.
24674         * gimple-streamer-out.c: Include value-prof.h.
24675         (output_gimple_stmt): Output histogram.
24676         (output_bb): Use streamer_write_gcov_count.
24677         * value-prof.c: Include data-streamer.h
24678         (dump_histogram_value): Add HIST_TYPE_MAX.
24679         (stream_out_histogram_value): New function.
24680         (stream_in_histogram_value): New function.
24681         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
24682         (stream_out_histogram_value, stream_in_histogram_value): Declare.
24683         * data-streamer-out.c (streamer_write_gcov_count): New function.
24684         (streamer_write_gcov_count_stream): New function.
24685         * lto-cgraph.c (lto_output_edge): Update counter streaming.
24686         (lto_output_node): Likewise.
24687         (input_node, input_edge): Likewise.
24688         * lto-streamer-out.c (output_cfg): Update streaming.
24689         * lto-streamer-in.c (input_cfg): Likewise.
24690         * data-streamer.h (streamer_write_gcov_count,
24691         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
24692         * gimple-streamer-in.c: Include value-prof.h
24693         (input_gimple_stmt): Input histograms.
24694         (input_bb): Update profile streaming.
24696 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
24698         * genmodes.c (emit_max_int): New function.
24699         (emit_insn_modes_h): Added call to emit_max_function.
24700         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
24701         Added doc.
24702         * machmode.def: Fixed comment.
24704 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
24706         * combine.c (try_combine): Removed useless assert.
24707         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
24709 2013-03-28  Marek Polacek  <polacek@redhat.com>
24710             Richard Biener  <rguenther@suse.de>
24712         PR tree-optimization/56695
24713         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
24714         build signed result of a vector comparison.
24715         * tree-cfg.c (verify_gimple_comparison): Check that a result
24716         of a vector comparison has signed type.
24718 2013-03-28  Richard Biener  <rguenther@suse.de>
24720         PR tree-optimization/37021
24721         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
24722         do not restrict gaps between groups.
24723         * tree-vect-stmts.c (vectorizable_load): Properly account for
24724         a gap between groups.
24726 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
24728         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
24729         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
24730         is not enabled.
24732 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
24734         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
24735         * doc/extend.texi (Named Address Spaces): Ditto.
24736         (Variable Attributes): Ditto.
24738 2013-03-27  Kai Tietz  <ktietz@redhat.com>
24740         * config.build: Add support for cygwin x64 target.
24741         * config.gcc: Likewise.
24742         * config.host: Likewise.
24743         * configure.ac: Likewise
24744         * configure: Regenerated.
24746 2013-03-27  Kai Tietz  <ktietz@redhat.com>
24748         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
24749         * config/i386/t-cygwin-w64: New file.
24750         * config/i386/cygwin-w64.h: New file.
24751         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
24752         and add support for x64-cygwin target.
24753         (CPP_SPEC): Likewise.
24754         (CXX_WRAP_SPEC_LIST): Undefine before define.
24755         (LIBGCJ_SONAME): Use 15 as version.
24757 2013-03-27  Richard Biener  <rguenther@suse.de>
24759         PR tree-optimization/56716
24760         * tree-ssa-structalias.c (perform_var_substitution): Adjust
24761         dumping for ref nodes.
24763 2013-03-27  Martin Jambor  <mjambor@suse.cz>
24765         PR tree-optimization/55334
24766         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
24767         restricted pointers to arrays.
24769 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
24771         * Makefile.in (.SUFFIXES): Add .cc.
24772         (.c.o): Apply same recipe for implicit rule .cc.o.
24774 2013-03-27  Richard Biener  <rguenther@suse.de>
24776         PR tree-optimization/37021
24777         * tree-vect-data-refs.c (vect_check_strided_load): Allow
24778         REALPART/IMAGPART_EXPRs around the supported refs.
24779         * tree-ssa-structalias.c (find_func_aliases): Assume that
24780         floating-point values are not used to transfer pointers.
24782 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
24784         * target.def (TARGET_HAS_IFUNC_P): New target hook.
24785         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
24786         * doc/tm.texi: Regenerate.
24787         * targhooks.h (default_has_ifunc_p): New.
24788         * targhooks.c (default_has_ifunc_p): Ditto.
24789         * config/linux-protos.h: New file.
24790         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
24791         hook for linux which disables support of indirect functions in android.
24792         * config/linux-android.c: New file.
24793         * config/t-linux-android.c: Ditto.
24794         * config.gcc: Added new object file linux-android.o.
24795         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
24796         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
24797         * varasm.c (do_assemble_alias): Likewise.
24798         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
24799         doesn't support indirect functions.
24800         * configure: Regenerate.
24802 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
24804         PR target/56102
24805         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
24806         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
24807         mult-word mode.
24809 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
24811         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
24813 2013-03-27  Terry Guo  <terry.guo@arm.com>
24815         * config/arm/arm-cores.def: Added core cortex-r7.
24816         * config/arm/arm-tune.md: Regenerated.
24817         * config/arm/arm-tables.opt: Regenerated.
24818         * doc/invoke.texi: Added entry for core cortex-r7.
24820 2013-03-27  Walter Lee  <walt@tilera.com>
24822         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
24823         double-decrement of next_scratch_regno.
24825 2013-03-27  Walter Lee  <walt@tilera.com>
24827         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
24828         input operands.
24829         (insn_v1mulus): Ditto.
24830         (insn_v2muls): Ditto.
24832 2013-03-27  Walter Lee  <walt@tilera.com>
24834         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
24835         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
24837 2013-03-27  Walter Lee  <walt@tilera.com>
24839         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
24840         (*sibcall_value): Ditto.
24842 2013-03-27  Walter Lee  <walt@tilera.com>
24844         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
24845         (insn_mnz_v8qi): ... this ...
24846         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
24847         vector equivalent.
24848         (insn_v<n>mnz): Replaced by ...
24849         (insn_v1mnz): ... this ...
24850         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
24851         equivalent.
24852         (insn_mz_<mode>): Replaced by ...
24853         (insn_mz_v8qi): ... this ...
24854         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
24855         vector equivalent.
24856         (insn_v<n>mz): Replaced by ...
24857         (insn_v1mz): ... this ...
24858         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
24859         equivalent.
24861 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
24863         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
24865 2013-03-26  Roland McGrath  <mcgrathr@google.com>
24867         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
24868         than fprintf with a non-constant, non-format string.
24870 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
24872         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
24873         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
24874         operand 0 predicate.
24875         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
24876         attribute.  Use general_x64nomem_operand as operand 1 predicate.
24877         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
24878         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
24879         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
24880         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
24881         (mov<mode>_insv_1): Remove expander.  Merge insn with
24882         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
24883         Use general_x64nomem_operand as operand 1 predicate.
24884         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
24885         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
24886         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
24887         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
24888         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
24889         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
24890         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
24891         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
24892         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
24893         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
24894         (general_x64nomem_operand): Ditto.
24896 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
24898         * config/rtems.opt: Add -pthread option.
24900 2013-03-26  Richard Biener  <rguenther@suse.de>
24902         * alias.c (find_base_term): Avoid redundant and not used recursion.
24903         (base_alias_check): Get the initial base term from the caller.
24904         (true_dependence_1): Compute and pass base terms to base_alias_check.
24905         (write_dependence_p): Likewise.
24906         (may_alias_p): Likewise.
24908 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
24910         * config/aarch64/aarch64.c (aarch64_classify_address): Support
24911         PC-relative load in SI modes and above only.
24913 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
24915         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
24916         * config/arm/iwmmxt.md (WCGR0): Update.
24917         (WCGR1, WCGR2, WCGR3): Likewise.
24919 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
24921         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
24922         Use x64 and nox64 isa attributes.
24924 2013-03-26  Richard Biener  <rguenther@suse.de>
24926         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
24927         alignment computations and rely on get_object_alignment_1
24928         for the !TYPE_P case.
24929         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
24931 2013-03-26  Walter Lee  <walt@tilera.com>
24933         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
24934         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
24936 2013-03-25  Jeff Law  <law@redhat.com>
24938         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
24939         check for INTEGRAL_TYPE_P that was missing due to checking in
24940         wrong version of prior patch.
24942 2013-03-25  Walter Lee  <walt@tilera.com>
24944         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
24945         TILEGX_INSN_SHUFFLEBYTES1.
24946         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
24947         shufflebytes1.
24948         (tilegx_builtins): Ditto.
24949         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
24951 2013-03-25  Walter Lee  <walt@tilera.com>
24953         * config/tilegx/tilegx.md (floatsisf2): New pattern.
24954         (floatunssisf2): New pattern.
24955         (floatsidf2): New pattern.
24956         (floatunssidf2): New pattern.
24958 2013-03-25  Walter Lee  <walt@tilera.com>
24960         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
24961         tests for constraint J, K, N, P.
24963 2013-03-25  Walter Lee  <walt@tilera.com>
24965         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
24966         Use indirect/pcrel encoding.
24967         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
24968         Ditto.
24970 2013-03-25  Steve Ellcey  <sellcey@mips.com>
24972         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
24973         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
24974         * config/mips/mips.c (mips_option_override): Set IMADD default.
24975         * config/mips/mips.h (PTF_AVOID_IMADD): New.
24976         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
24977         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
24978         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
24979         * doc/invoke.texi (-mimadd/-mno-imadd): New.
24981 2013-03-25  Jeff Law  <law@redhat.com>
24983         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
24984         slightly to avoid creating and folding useless trees.  Simplify
24985         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
24987 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
24989         * config/i386/i386.md (*zero_extendsidi2): Merge with
24990         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
24991         * config/i386/predicates.md (x86_64_zext_operand): Rename from
24992         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
24993         targets.  Clarify comment.
24995 2013-03-25  Martin Jambor  <mjambor@suse.cz>
24997         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
24998         pass-through jump functions differently.
24999         (ipa_read_jump_function): Likewise.  Also use setter functions to set
25000         up jump functions.
25002 2013-03-25  Martin Jambor  <mjambor@suse.cz>
25004         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
25005         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
25006         process it.
25007         (ipa_get_indirect_edge_target): New function.
25008         (devirtualization_time_bonus): New parameter known_aggs, pass it to
25009         ipa_get_indirect_edge_target.  Update all callers.
25010         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
25011         ipa_get_indirect_edge_target_1 instead of calling
25012         ipa_get_indirect_edge_target.
25013         (create_specialized_node): Pass aggvlas to
25014         ipcp_discover_new_direct_edges.
25016 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
25018         * config/arm/arm.md (f_sels, f_seld): New types.
25019         (*cmov<mode>): New pattern.
25020         * config/arm/predicates.md (arm_vsel_comparison_operator): New
25021         predicate.
25023 2013-03-25  Kai Tietz  <ktietz@redhat.com>
25025         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
25026         POSIX-printf for mingw-hosted  builds.
25028 2013-03-25  Richard Biener  <rguenther@suse.de>
25030         PR middle-end/56694
25031         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
25032         must-not-throw stmt location.
25034 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
25036         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
25037         Emit load-acquire versions when acq is true.
25038         (arm_emit_store_exclusive): Add rel parameter.
25039         Emit store-release versions when rel is true.
25040         (arm_split_compare_and_swap): Use acquire-release instructions
25041         instead.
25042         of barriers when appropriate.
25043         (arm_split_atomic_op): Likewise.
25044         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
25045         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
25046         (VUNSPEC_SLX): Likewise.
25047         (VUNSPEC_LDA): Likewise.
25048         (VUNSPEC_STL): Likewise.
25049         * config/arm/sync.md (atomic_load<mode>): New pattern.
25050         (atomic_store<mode>): Likewise.
25051         (arm_load_acquire_exclusive<mode>): Likewise.
25052         (arm_load_acquire_exclusivesi): Likewise.
25053         (arm_load_acquire_exclusivedi): Likewise.
25054         (arm_store_release_exclusive<mode>): Likewise.
25056 2013-03-25  Catherine Moore  <clm@codesourcery.com>
25058         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
25059         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
25060         * config/mip/predicates.md (lwsp_swsp_operand,
25061         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
25062         sb16_operand, db4_operand, db7_operand, ib3_operand,
25063         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
25064         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
25065         andi16_operand): New predicates.
25066         * config/mips/mips.md (compression): New attribute.
25067         (enabled): New attribute.
25068         (length): Consider compression in computing length.
25069         (shift_compression): New code attribute.
25070         (*add<mode>3): New operands. Record compression.
25071         (sub<mode>3): Likewise.
25072         (one_cmpl<mode>2): Likewise.
25073         (*and<mode>3): Likewise.
25074         (*ior<mode>3): Likewise.
25075         (unnamed pattern for xor): Likewise.
25076         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
25077         (*<optab><mode>3): Likewise.
25078         (*mov<mode>_internal: Likewise.
25079         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
25080         (mips_unsigned_immediate_p): New.
25081         (umips_lwsp_swsp_address_p): New.
25082         (m16_based_address_p): New.
25083         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
25084         (mips_unsigned_immediate_p): New prototype.
25085         (lwsp_swsp_address_p): New prototype.
25086         (m16_based_address_p): New prototype.
25087         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
25088         (mips_signed_immediate_p): New function.
25089         (m16_based_address_p): New function.
25090         (lwsp_swsp_address_p): New function.
25091         (mips_print_operand_punctuation): Recognize short delay slot insns
25092         for microMIPS.add<mode>3"
25094 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
25096         PR target/56720
25097         * config/arm/iterators.md (v_cmp_result): New mode attribute.
25098         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
25100 2013-03-25  Richard Biener  <rguenther@suse.de>
25102         PR tree-optimization/56689
25103         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
25104         any edge.
25106 2013-03-25  Richard Biener  <rguenther@suse.de>
25108         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
25109         of bitmap.
25110         (memory_references): Likewise.
25111         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
25112         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
25113         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
25114         (gather_mem_refs_in_loops): Fold into ...
25115         (analyze_memory_references): ... this.  Move initialization
25116         to tree_ssa_lim_initialize.
25117         (fill_always_executed_in): Rename to ...
25118         (fill_always_executed_in_1): ... this.
25119         (fill_always_executed_in): Move contains_call computation to
25120         this new function from ...
25121         (tree_ssa_lim_initialize): ... here.
25122         (tree_ssa_lim): Call fill_always_executed_in.
25124 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
25126         * postreload.c (reload_combine): Fix code detecting returns.
25128 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
25130         * function.c (emit_use_return_register_into_block): On cc0 targets,
25131         do not emit the sequence between cc0 setter and user.
25133 2013-03-25  Kai Tietz  <ktietz@redhat.com>
25135         * config/i386/predicates.md (local_symbolic_operand): Interpret
25136         dll-imported symbols as none-local.
25138 2013-03-25  Richard Biener  <rguenther@suse.de>
25140         * tree-ssa-loop-im.c (struct depend): Remove.
25141         (struct lim_aux_data): Make depends a vec of gimples.
25142         (free_lim_aux_data): Adjust.
25143         (add_dependency): Likewise.
25144         (set_level): Likewise.
25146 2013-03-25  Richard Biener  <rguenther@suse.de>
25148         PR middle-end/56434
25149         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
25150         the pointer returned by calls with ECF_MALLOC set.
25152 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
25154         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
25156 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
25158         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
25159         using MMXMODE mode iterator.
25160         (*move<mode>_internal): Merge with *movv2sf_internal and
25161         *movv2sf_internal_rex64 using MMXMODE mode iterator.
25163 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
25165         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
25166         (record_last_mem_set_info): Likewise.
25168         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
25169         of XNEWVEC followed by memset.
25170         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
25172 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
25174         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
25175         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
25176         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
25177         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
25178         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
25179         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
25180         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
25181         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
25182         BARRIER_P instead of GET_CODE.
25184 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
25186         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
25187         inaccuracy in the probing code.
25189         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
25190         (ctrapdi4): Likewise.
25192 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
25194         * calls.c (expand_call): Add missing guard to code handling return
25195         of non-BLKmode structures in MSB.
25196         * function.c (expand_function_end): Likewise.
25198 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
25200         * combine.c (try_combine): Adjust comment.  Do not add the set of
25201         insn #0 if the destination indirectly is set or dies in insn #2.
25202         Tidy up code to distribute a new note.
25204 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
25206         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
25207         also for alternatives 16 and 17.
25209 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
25211         * config/i386/sse.md (*mov<mode>_internal): Merge with
25212         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
25213         Emit insn template depending on type attribute.  Use
25214         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
25215         movd instead of movq mnemonic for interunit moves.  Rewrite mode
25216         attribute calculation.  Remove unit attribute calculation.
25217         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
25218         Set prefix_data16 attribute for DImode ssemov types.
25219         Use Ym instead of y for SSE-MMX conversion alternatives.
25220         Reorder operand constraints.
25222 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
25224         * df.h (df_insn_delete): Adjust prototype.
25225         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
25226         and let it decide whether mark the basic block dirty.
25227         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
25228         * df-scan.c (df_insn_info_delete): New helper function, split
25229         off from df_insn_delete.
25230         (df_scan_free_bb_info): Use it.
25231         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
25232         Likewise.
25233         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
25234         that the insn is actually an insn and it has a non-NULL basic block.
25235         Do not mark basic block dirty if only deleting a DEBUG_INSN.
25237 2013-03-22  Richard Biener  <rguenther@suse.de>
25239         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
25240         dep_ref members.
25241         (mem_ref_alloc): Do not allocate them.
25242         (refs_independent_p): Do not query or maintain a cache.
25244 2013-03-22  Richard Biener  <rguenther@suse.de>
25246         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
25247         (gather_mem_refs_in_loops): Do not compute it.
25248         (analyze_memory_references): Do not allocate it.
25249         (tree_ssa_lim_finalize): Do not free it.
25250         (for_all_locs_in_loop): Do not query all_refs_in_loop.
25252 2013-03-22  Richard Biener  <rguenther@suse.de>
25254         * is-a.h (as_a): Use gcc_checking_assert.
25256 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
25258         * config/aarch64/aarch64.c (aarch64_print_operand): New
25259         format specifier for printing a constant in hex.
25260         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
25261         format specifier for printing second operand.
25263 2013-03-22  Richard Biener  <rguenther@suse.de>
25265         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
25266         bitmaps.
25267         (gather_mem_refs_in_loops): Perform store accumulation here.
25268         (create_vop_ref_mapping_loop): Remove.
25269         (create_vop_ref_mapping): Likewise.
25270         (analyze_memory_references): Initialize refs_stored_in_loop.
25271         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
25272         (record_indep_loop): Remove.
25273         (record_dep_loop): New function.
25274         (ref_indep_loop_p_1): Adjust to only walk over references
25275         in the loop, not its subloops.
25276         (ref_indep_loop_p): Rename to ...
25277         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
25278         maintaining a more fine-grained cache.
25279         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
25280         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
25282 2013-03-22  Richard Biener  <rguenther@suse.de>
25284         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
25285         (struct mem_ref): Make accesses_in_loop a vec of a vec of
25286         aggregate mem_ref_loc.
25287         (free_mem_ref_locs): Inline into ...
25288         (memref_free): ... this and adjust.
25289         (mem_ref_alloc): Adjust.
25290         (mem_ref_locs_alloc): Remove.
25291         (record_mem_ref_loc): Adjust.
25292         (get_all_locs_in_loop): Rewrite into ...
25293         (for_all_locs_in_loop): ... this iterator.
25294         (rewrite_mem_ref_loc): New functor.
25295         (rewrite_mem_refs): Use for_all_locs_in_loop.
25296         (sm_set_flag_if_changed): New functor.
25297         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
25298         (ref_always_accessed): New functor.
25299         (ref_always_accessed_p): Use for_all_locs_in_loop.
25301 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
25303         * tree-pass.h (PROP_gimple_lvec): New.
25304         * passes.c (dump_properties): Handle PROP_gimple_lvec.
25305         (init_optimization_passes): Move pass_lower_vector.
25306         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
25307         PROP_gimple_lvec.
25308         (pass_lower_vector): Provide PROP_gimple_lvec.
25309         (pass_lower_vector_ssa): Likewise.
25310         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
25312 2013-03-21  Mark Wielaard  <mjw@redhat.com>
25314         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
25316 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
25318         * config/i386/i386.md (*movdi_internal): Disparage slightly
25319         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
25320         conversion alternatives.
25322 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
25324         PR middle-end/48087
25325         * diagnostic.def (DK_WERROR): New kind.
25326         * diagnostic.h (werrorcount): Define.
25327         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
25328         promoted to DK_ERROR, increment DK_WERROR counter instead of
25329         DK_ERROR counter.
25330         * toplev.c (toplev_main): Call print_ignored_options even if
25331         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
25332         even if just werrorcount is non-zero.
25334         PR debug/55608
25335         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
25336         on failure.
25337         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
25338         (string_cst_pool_decl): New function.
25339         (optimize_one_addr_into_implicit_ptr): New function.
25340         (resolve_addr_in_expr): Optimize DWARF location expression
25341         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
25342         which doesn't live in memory, but has DW_AT_location or
25343         DW_AT_const_value, or refers to a string literal, into
25344         DW_OP_GNU_implicit_pointer.
25345         (optimize_location_into_implicit_ptr): New function.
25346         (resolve_addr): If removing DW_AT_location of a variable because
25347         it was DW_OP_addr of address of the variable, but the variable doesn't
25348         live in memory, try to emit const value attribute for the initializer.
25350 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
25352         * tree.h (VECTOR_TYPE_P): New macro.
25353         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
25354         TYPE_MODE): Use it.
25355         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
25356         VEC_COND_EXPR cannot be lvalues.
25357         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
25359 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
25361         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
25362         Restrict the transformation to equal modes.
25364 2013-03-21  Richard Biener  <rguenther@suse.de>
25366         PR tree-optimization/39326
25367         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
25368         (MEM_ANALYZABLE): Adjust.
25369         (record_mem_ref_loc): Move bitmap ops ...
25370         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
25371         unanalyzable refs, do not record locations for it.
25372         (analyze_memory_references): Allocate ref zero as shared
25373         unanalyzable ref.
25374         (refs_independent_p): Do not test for unanalyzed mems here.
25375         (ref_indep_loop_p_1): Special-case disambiguation against
25376         the unanalyzed ref.
25377         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
25379 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
25381         * config/arm/arm-protos.h (tune_params): Add
25382         prefer_neon_for_64bits field.
25383         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
25384         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
25385         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
25386         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
25387         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
25388         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
25389         (arm_option_override): Handle -mneon-for-64bits new option.
25390         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
25391         (prefer_neon_for_64bits): Declare new variable.
25392         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
25393         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
25394         (arch_enabled): Handle new arch types. Remove support for onlya8
25395         and nota8.
25396         (one_cmpldi2): Use new arch names.
25397         (zero_extend<mode>di2, extend<mode>di2): Ditto.
25398         * config/arm/arm.opt (mneon-for-64bits): Add option.
25399         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
25400         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
25401         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
25402         of onlya8.
25403         * doc/invoke.texi (-mneon-for-64bits): Document.
25405 2013-03-21  Richard Biener  <rguenther@suse.de>
25407         PR tree-optimization/39326
25408         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
25409         (sort_bbs_in_loop_postorder_cmp): New function.
25410         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
25412 2013-03-21  Richard Biener  <rguenther@suse.de>
25414         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
25415         (vect_insert_into_interleaving_chain): Likewise.
25416         (vect_drs_dependent_in_basic_block): Inline ...
25417         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
25418         split out from ...
25419         (vect_analyze_data_ref_dependence): ... here.  Simplify.
25420         (vect_check_interleaving): Simplify.
25421         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
25422         (vect_slp_analyze_data_ref_dependences): ... this new function.
25423         (dr_group_sort_cmp): New function.
25424         (vect_analyze_data_ref_accesses): Compute data-reference groups
25425         here instead of in vect_analyze_data_ref_dependence.  Use
25426         a more efficient algorithm.
25427         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
25428         vect_slp_analyze_data_ref_dependences.  Call
25429         vect_analyze_data_ref_accesses earlier.
25430         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
25431         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
25432         (vect_slp_analyze_data_ref_dependences): New prototype.
25434 2013-03-21  Richard Biener  <rguenther@suse.de>
25436         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
25437         ref is stored in the loop.
25438         (find_refs_for_sm): Walk only over all stores.
25439         (store_motion_loop): Allocate from lim_bitmap_obstack.
25440         (store_motion): Likewise.
25442 2013-03-21  Richard Biener  <rguenther@suse.de>
25444         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
25445         Update virtual SSA form.
25447 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
25449         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
25450         * configure: Regenerate.
25451         * config.in: Regenerate.
25452         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
25453         if !HAVE_LD_EH_FRAME_CIEV3.
25455 2013-03-21  Richard Biener  <rguenther@suse.de>
25457         * tree-cfg.c (verify_expr_no_block): New function.
25458         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
25459         nor DECL_VALUE_EXPR have locations with associated blocks.
25460         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
25461         (clear_unused_block_pointer): Remove code dealing with
25462         blocks in DECL_DEBUG_EXPR locations.
25464 2013-03-21  Richard Biener  <rguenther@suse.de>
25466         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
25467         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
25468         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
25469         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
25470         instead of DECL_DEBUG_EXPR_IS_FROM.
25471         * gimplify.c (gimplify_modify_expr): Likewise.
25472         * tree-cfg.c (verify_expr_location_1): Likewise.
25473         * tree-complex.c (create_one_component_var): Likewise.
25474         * tree-sra.c (create_access_replacement): Likewise.
25475         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
25476         (clear_unused_block_pointer): Likewise.
25477         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
25478         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
25479         * var-tracking.c (var_debug_decl): Likewise.
25480         (track_expr_p): Likewise.
25481         * tree-inline.c (add_local_variables): Likewise.  Set
25482         DECL_HAS_DEBUG_EXPR_P after copying it.
25483         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
25484         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
25486 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
25488         PR bootstrap/56656
25489         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
25490         * configure: Regenerate.
25491         * config.in: Regenerate.
25492         * config/i386/i386.md (*movdf_internal): Use
25493         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
25494         movd instead of movq mnemonic for interunit moves.
25495         (*movdi_internal): Ditto.
25497 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
25499         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
25500         (abd<mode>_3): New pattern.
25501         (aba<mode>_3): New pattern.
25502         (fabd<mode>_3): New pattern.
25504 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
25506         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
25507         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
25508         occurrence of REGISTER_PREFIX as its empty string.
25510 2013-03-20  Jeff Law  <law@redhat.com>
25512         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
25513         addititional equivalences for equality comparisons between an SSA_NAME
25514         and a constant where the SSA_NAME was set from a widening conversion.
25516 2013-03-20  Walter Lee  <walt@tilera.com>
25518         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
25520 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
25522         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
25523         depending on type attribute.
25524         (*movti_internal): Ditto.
25525         (*movtf_internal): Ditto.
25526         (*movxf_internal): Ditto.
25527         (*movdf_internal): Ditto.
25528         (*movsf_internal): Ditto.
25530 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
25532         * config/i386/i386.md (*movti_internal): Set prefix attribute to
25533         maybe_vex for sselog1 and ssemov types.
25534         (*movdi_internal): Reorder operand constraints.
25535         (*movsi_internal): Ditto.  Set prefix attribute to
25536         maybe_vex for sselog1 and ssemov types.
25537         (*movtf_internal): Set prefix attribute to maybe_vex
25538         for sselog1 and ssemov types.
25539         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
25540         DImode ssemov types.  Reorder operand constraints.
25541         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
25542         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
25543         attribute for SImode ssemov types.  Reorder operand constraints.
25545 2013-03-20  Martin Jambor  <mjambor@suse.cz>
25547         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
25548         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
25550 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
25552         * config/rs6000/predicates.md (indexed_address, update_address_mem
25553         update_indexed_address_mem): New predicates.
25554         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
25555         attribute for load/store instructions.
25556         * config/rs6000/dfp.md (movsd_store): Likewise.
25557         (movsd_load): Likewise.
25558         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
25559         (unnamed HI->DI extend define_insn): Likewise.
25560         (unnamed SI->DI extend define_insn): Likewise.
25561         (unnamed QI->SI extend define_insn): Likewise.
25562         (unnamed QI->HI extend define_insn): Likewise.
25563         (unnamed HI->SI extend define_insn): Likewise.
25564         (unnamed HI->SI extend define_insn): Likewise.
25565         (extendsfdf2_fpr): Likewise.
25566         (movsi_internal1): Likewise.
25567         (movsi_internal1_single): Likewise.
25568         (movhi_internal): Likewise.
25569         (movqi_internal): Likewise.
25570         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
25571         attribute for load/store instructions.
25572         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
25573         instructions.
25574         (mov<mode>_softfloat): Likewise.
25575         (mov<mode>_hardfloat32): Likewise.
25576         (mov<mode>_hardfloat64): Likewise.
25577         (mov<mode>_softfloat64): Likewise.
25578         (movdi_internal32): Likewise.
25579         (movdi_internal64): Likewise.
25580         (probe_stack_<mode>): Likewise.
25582 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
25584         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
25585         floating point, and decimal floating point to reload iterator.
25587         * config/rs6000/constraints.md (wl constraint): New constraints to
25588         return FLOAT_REGS if certain options are used to reduce the number
25589         of separate patterns that exist in the file.
25590         (wx constraint): Likewise.
25591         (wz constraint): Likewise.
25593         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
25594         -mdebug=reg, print wg, wl, wx, and wz constraints.
25595         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
25596         Initialize the reload functions for 64-bit binary/decimal floating
25597         point types.
25598         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
25599         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
25600         create the buffer on the stack to overcome not having a 32-bit
25601         load and store.
25602         (rs6000_emit_move): Likewise.
25603         (rs6000_secondary_memory_needed_rtx): Likewise.
25604         (rs6000_alloc_sdmode_stack_slot): Likewise.
25605         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
25606         via xxlxor, just like DFmode 0.0.
25608         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
25609         define as 1 if we are running on a power7 or newer.
25610         (enum r6000_reg_class_enum): Add new constraints.
25612         * config/rs6000/dfp.md (movsd): Delete, combine with binary
25613         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
25614         with other moves by using conditional constraits (wg).  Use LFIWZX
25615         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
25616         (movsd splitter): Likewise.
25617         (movsd_hardfloat): Likewise.
25618         (movsd_softfloat): Likewise.
25620         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
25621         binary and decimal floating point moves.
25622         (fmove_ok): New attributes to combine binary and decimal floating
25623         point moves, and to combine power6x (mfpgpr) moves along normal
25624         floating moves.
25625         (real_value_to_target): Likewise.
25626         (f32_lr): Likewise.
25627         (f32_lm): Likewise.
25628         (f32_li): Likewise.
25629         (f32_sr): Likewise.
25630         (f32_sm): Likewise.
25631         (f32_si): Likewise.
25632         (movsf): Combine binary and decimal floating point moves.  Combine
25633         power6x (mfpgpr) moves with other moves by using conditional
25634         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
25635         (mov<mode> for SFmode/SDmode); Likewise.
25636         (SFmode/SDmode splitters): Likewise.
25637         (movsf_hardfloat): Likewise.
25638         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
25639         (movsf_softfloat): Likewise.
25640         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
25642         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
25643         wx and wz constraints.
25645         * config/rs6000/constraints.md (wg constraint): New constraint to
25646         return FLOAT_REGS if -mmfpgpr (power6x) was used.
25648         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
25649         constraint.
25651         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
25652         -mdebug=reg, print wg, wl, wx, and wz constraints.
25653         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
25654         Initialize the reload functions for 64-bit binary/decimal floating
25655         point types.
25656         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
25657         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
25658         create the buffer on the stack to overcome not having a 32-bit
25659         load and store.
25660         (rs6000_emit_move): Likewise.
25661         (rs6000_secondary_memory_needed_rtx): Likewise.
25662         (rs6000_alloc_sdmode_stack_slot): Likewise.
25663         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
25664         via xxlxor, just like DFmode 0.0.
25666         * config/rs6000/dfp.md (movdd): Delete, combine with binary
25667         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
25668         with other moves by using conditional constraits (wg).  Use LFIWZX
25669         and STFIWX for loading SDmode on power7.
25670         (movdd splitters): Likewise.
25671         (movdd_hardfloat32): Likewise.
25672         (movdd_softfloat32): Likewise.
25673         (movdd_hardfloat64_mfpgpr): Likewise.
25674         (movdd_hardfloat64): Likewise.
25675         (movdd_softfloat64): Likewise.
25677         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
25678         64-bit binary and decimal floating point moves.
25679         (FMOVE64X): Likewise.
25680         (movdf): Combine 64-bit binary and decimal floating point moves.
25681         Combine power6x (mfpgpr) moves with other moves by using
25682         conditional constraits (wg).
25683         (mov<mode> for DFmode/DDmode): Likewise.
25684         (DFmode/DDmode splitters): Likewise.
25685         (movdf_hardfloat32): Likewise.
25686         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
25687         (movdf_softfloat32): Likewise.
25688         (movdf_hardfloat64_mfpgpr): Likewise.
25689         (movdf_hardfloat64): Likewise.
25690         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
25691         (movdf_softfloat64): Likewise.
25692         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
25693         (reload_<mode>_load): Move to later in the file so they aren't in
25694         the middle of the floating point move insns.
25695         (reload_<mode>_store): Likewise.
25697         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
25698         constraint.
25700         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
25701         constraint if -mdebug=reg.
25702         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
25703         Enable using dd reload support if needed.
25705         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
25706         binary and decimal floating point moves in rs6000.md.
25707         (movtd_internal): Likewise.
25709         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
25710         decimal floating point moves.
25711         (movtf): Likewise.
25712         (movtf_internal): Likewise.
25713         (mov<mode>_internal, TDmode/TFmode): Likewise.
25714         (movtf_softfloat): Likewise.
25715         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
25717         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
25718         movdi_internal64, using wg constraint for move direct operations.
25719         (movdi_internal64): Likewise.
25721         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
25722         MODES_TIEABLE_P for selected modes.  Print the numerical value of
25723         the various virtual registers. Use GPR/FPR first/last values,
25724         instead of hard coding the register numbers.  Print which modes
25725         have reload functions registered.
25726         (rs6000_option_override_internal): If -mdebug=reg, trace the options
25727         settings before/after setting cpu, target and subtarget settings.
25728         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
25729         and for secondary reload failures in rs6000_secondary_reload_inner.
25730         (rs6000_secondary_reload_fail): Likewise.
25731         (rs6000_secondary_reload_inner): Likewise.
25733         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
25734         macros for first/last GPR and FPR registers.
25735         (LAST_GPR_REGNO): Likewise.
25736         (FIRST_FPR_REGNO): Likewise.
25737         (LAST_FPR_REGNO): Likewise.
25739         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
25740         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
25741         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
25742         (vcond<mode><mode>): Likewise.
25743         (vcondu<mode><mode>): Likewise.
25744         (vector_gtu<mode>): Likewise.
25745         (vector_gte<mode>): Likewise.
25746         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
25747         to prevent the compiler from converting DImode operations to TImode.
25748         (ior<mode>3): Likewise.
25749         (and<mode>3): Likewise.
25750         (one_cmpl<mode>2): Likewise.
25751         (nor<mode>3): Likewise.
25752         (andc<mode>3): Likewise.
25754         * config/rs6000/constraints.md (wt constraint): New constraint
25755         that returns VSX_REGS if TImode is allowed in VSX registers.
25757         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
25758         constant under VSX.
25760         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
25761         similar to TImode, but it is restricted to being in the GPRs.
25763         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
25764         TImode to occupy a single VSX register.
25766         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
25767         -mvsx-timode for power7/power8.
25768         (power7 cpu): Likewise.
25769         (power8 cpu): Likewise.
25771         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
25772         sure that TFmode/TDmode take up two registers if they are ever
25773         allowed in the upper VSX registers.
25774         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
25775         registers.
25776         (rs6000_init_hard_regno_mode_ok): Likewise.
25777         (rs6000_debug_reg_global): Add debugging for PTImode and wt
25778         constraint.  Print if LRA is turned on.
25779         (rs6000_option_override_internal): Give an error if -mvsx-timode
25780         and VSX is not enabled.
25781         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
25782         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
25783         to reg+offset addressing.  Use PTImode when checking offset
25784         addresses for validity.
25785         (reg_offset_addressing_ok_p): Likewise.
25786         (rs6000_legitimate_offset_address_p): Likewise.
25787         (rs6000_legitimize_address): Likewise.
25788         (rs6000_legitimize_reload_address): Likewise.
25789         (rs6000_legitimate_address_p): Likewise.
25790         (rs6000_eliminate_indexed_memrefs): Likewise.
25791         (rs6000_emit_move): Likewise.
25792         (rs6000_secondary_reload): Likewise.
25793         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
25794         reloads to fpr registers to continue to use reg+offset addressing,
25795         but 64-bit reloads to altivec registers need reg+reg addressing.
25796         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
25797         it.  Treat LO_SUM like a PLUS operation.
25798         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
25799         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
25800         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
25801         registers to share a register with a smaller sized type, since VSX
25802         puts scalars in the upper 64-bits.
25803         (print_operand): Add support for PTImode.
25804         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
25805         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
25806         registers, but don't have arithmetic support.
25807         (rs6000_memory_move_cost): Add test for VSX.
25808         (rs6000_opt_masks): Add -mvsx-timode.
25810         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
25811         for TImode.
25812         (VSs): Likewise.
25813         (VSr): Use wt constraint for TImode.
25814         (VSv): Drop TImode support.
25815         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
25816         (vsx_movti_64bit): Likewise.
25817         (vsx_movti_32bit): Likewise.
25818         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
25819         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
25820         one '?' on the appropriate output constraint.  Do not allow TImode
25821         logical operations on 32-bit systems.
25822         (vsx_ior<mode>3): Likewise.
25823         (vsx_xor<mode>3): Likewise.
25824         (vsx_one_cmpl<mode>2): Likewise.
25825         (vsx_nor<mode>3): Likewise.
25826         (vsx_andc<mode>3): Likewise.
25827         (vsx_concat_<mode>): Likewise.
25828         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
25830         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
25831         OPTION_MASK_VSX_TIMODE.
25832         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
25833         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
25835         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
25836         (TI2 iterator): New iterator for TImode, PTImode.
25837         (wd mode attribute): Add values for vector types.
25838         (movti_string): Replace TI move operations with operations for TImode
25839         and PTImode.  Add support for TImode being allowed in VSX registers.
25840         (mov<mode>_string, TImode/PTImode): Likewise.
25841         (movti_ppc64): Likewise.
25842         (mov<mode>_ppc64, TImode/PTImode): Likewise.
25843         (TI mode splitters): Likewise.
25845         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
25846         constraint.
25848 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
25850         PR tree-optimization/56355
25851         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
25852         Also handle integers with undefined overflow.
25854 2013-03-20  Catherine Moore  <clm@codesourcery.com>
25855             Maciej W. Rozycki  <macro@codesourcery.com>
25856             Tom de Vries  <tom@codesourcery.com>
25857             Nathan Sidwell  <nathan@codesourcery.com>
25858             Iain Sandoe  <iain@codesourcery.com>
25859             Nathan Froyd  <froydnj@codesourcery.com>
25860             Chao-ying Fu  <fu@mips.com>
25862         * doc/extend.texi: (micromips, nomicromips, nocompression):
25863         Document new function attributes.
25864         * doc/invoke.texi (minterlink-compressed, mmicromips,
25865         m14k, m14ke, m14kec): Document new options.
25866         (minterlink-mips16): Update documentation.
25867         * doc/md.texi (ZC, ZD): Document new constraints.
25868         * configure.ac (gcc_cv_as_micromips): Check if linker
25869         supports the .set micromips directive.
25870         * configure: Regenerate.
25871         * config.in: Regenerate.
25872         * config/mips/mips-tables.opt: Regenerate.
25873         * config/mips/micromips.md: New file.
25874         * constraints.md (ZC, ZD): New constraints.
25875         * config/mips/predicates.md (movep_src_register): New predicate.
25876         (movep_src_operand): New predicate.
25877         (non_volatile_mem_operand): New predicate.
25878         * config/mips/mips.md (multimem): New type.
25879         (length): Differentiate between 17-bit and 18-bit branch offsets.
25880         (MOVEP1, MOVEP2): New mode iterator.
25881         (mov_<load>l): Use ZC constraint.
25882         (mov_<load>r): Likewise.
25883         (mov_<store>l): Likewise.
25884         (mov_<store>r): Likewise.
25885         (*branch_equality<mode>_inverted): Add microMIPS support.
25886         (*branch_equality<mode>): Likewise.
25887         (*jump_absolute): Likewise.
25888         (indirect_jump_<mode>): Likewise.
25889         (tablejump_<mode>): Likewise.
25890         (<optab>_internal): Likewise.
25891         (sibcall_internal): Likewise.
25892         (sibcall_value_internal): Likewise.
25893         (prefetch): Use constraint ZD.
25894         * config/mips/mips.opt (minterlink-compressed): New option.
25895         (minterlink-mips16): Now an alias for minterlink-compressed.
25896         (mmicromips): New option.
25897         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
25898         (compare_and_swap_12): Likewise.
25899         (sync_add<mode>): Likewise.
25900         (sync_<optab>_12): Likewise.
25901         (sync_old_<optab>_12): Likewise.
25902         (sync_new_<optab>_12): Likewise.
25903         (sync_nand_12): Likewise.
25904         (sync_old_nand_12): Likewise.
25905         (sync_new_nand_12): Likewise.
25906         (sync_sub<mode>): Likewise.
25907         (sync_old_add<mode>): Likewise.
25908         (sync_old_sub<mode>): Likewise.
25909         (sync_new_add<mode>): Likewise.
25910         (sync_new_sub<mode>): Likewise.
25911         (sync_<optab><mode>): Likewise.
25912         (sync_old_<optab><mode>): Likewise.
25913         (sync_new_<optab><mode>): Likewise.
25914         (sync_nand<mode>): Likewise.
25915         (sync_old_nand<mode>): Likewise.
25916         (sync_new_nand<mode>): Likewise.
25917         (sync_lock_test_and_set<mode>): Likewise.
25918         (test_and_set_12): Likewise.
25919         (atomic_compare_and_swap<mode>): Likewise.
25920         (atomic_exchange<mode>_llsc): Likewise.
25921         (atomic_fetch_add<mode>_llsc): Likewise.
25922         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
25923         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
25924         (umips_save_restore_pattern_p): Likewise.
25925         (umips_load_store_pair_p): Likewise.
25926         (umips_output_load_store_pair): Likewise.
25927         (umips_movep_target_p): Likewise.
25928         (umips_12bit_offset_address_p): Likewise.
25929         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
25930         (mips_base_mips16): Rename this...
25931         (mips_base_compression_flags): ...to this. Update all uses.
25932         (mips_attribute_table): Add micromips, nomicromips and nocompression.
25933         (mips_mips16_decl_p): Delete.
25934         (mips_nomips16_decl_p): Delete.
25935         (mips_get_compress_on_flags): New function.
25936         (mips_get_compress_off_flags): New function.
25937         (mips_get_compress_mode): New function.
25938         (mips_get_compress_on_name): New function.
25939         (mips_get_compress_off_name): New function.
25940         (mips_insert_attributes): Support multiple compression types.
25941         (mips_merge_decl_attributes): Likewise.
25942         (umips_12bit_offset_address_p): New function.
25943         (mips_start_function_definition): Emit .set micromips directive.
25944         (mips_call_may_need_jalx_p): New function.
25945         (mips_function_ok_for_sibcall): Add microMIPS support.
25946         (mips_print_operand_punctuation): Support short delay slots and
25947         compact jumps.
25948         (umips_swm_mask, umips_swm_encoding): New.
25949         (umips_build_save_restore): New function.
25950         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
25951         (was_mips16_p): Remove.
25952         (old_compression_mode): New.
25953         (mips_set_compression_mode): New function.
25954         (mips_set_current_function): Add microMIPS support.
25955         (mips_option_override): Likewise.
25956         (umips_save_restore_pattern_p): New function.
25957         (umips_output_save_restore): New function.
25958         (umips_load_store_pair_p_1): New function.
25959         (umips_load_store_pair_p): New function.
25960         (umips_output_load_store_pair_1): New function.
25961         (umips_output_load_store_pair): New function.
25962         (umips_movep_target_p) New function.
25963         (mips_prepare_pch_save): Add microMIPS support.
25964         * config/mips/mips.h (TARGET_COMPRESSION): New.
25965         (TARGET_CPU_CPP_BUILTINS): Update macro
25966         to use new compression flags and to support microMIPS.
25967         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
25968         (MIPS_ARCH_FLOAT_SPEC): Likewise.
25969         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
25970         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
25971         (ASM_SPEC): Support mmicromips and mno-micromips.
25972         (M16STORE_REG_P): New macro.
25973         (MIPS_CALL): Support TARGET_MICROMIPS.
25974         (MICROMIPS_J): New macro.
25975         (mips_base_mips16): Rename this...
25976         (mips_base_compression_flags): ...to this.
25977         (UMIPS_12BIT_OFFSET_P): New macro.
25978         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
25979         (MULTILIB_DIRNAMES): Likewise.
25980 2013-03-20  Richard Biener  <rguenther@suse.de>
25982         PR tree-optimization/56661
25983         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
25984         the result does not have to be distinct.
25986 2013-03-20  Richard Biener  <rguenther@suse.de>
25988         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
25989         remap_gimple_op_r.
25991 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
25992             Steven Bosscher  <steven@gcc.gnu.org>
25994         PR rtl-optimization/56605
25995         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
25997 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
25999         PR bootstrap/56656
26000         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
26001         that require movd instead of movq.
26003 2013-03-20  Richard Biener  <rguenther@suse.de>
26005         * tree-ssa-structalias.c (struct variable_info): Add pointer
26006         to the first field of an aggregate with sub-vars.  Make
26007         this and the pointer to the next subfield its ID.
26008         (vi_next): New function.
26009         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
26010         storedanything_id, integer_id): Increment by one.
26011         (new_var_info, get_call_vi, lookup_call_clobber_vi,
26012         get_call_clobber_vi): Adjust.
26013         (solution_set_expand): Simplify and speedup.
26014         (solution_set_add): Inline into ...
26015         (set_union_with_increment): ... this.  Adjust accordingly.
26016         (do_sd_constraint): Likewise.
26017         (do_ds_constraint): Likewise.
26018         (do_complex_constraint): Simplify.
26019         (build_pred_graph): Adjust.
26020         (solve_graph): Likewise.  Simplify and speedup.
26021         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
26022         get_constraint_for_component_ref, get_constraint_for_1,
26023         first_vi_for_offset, first_or_preceding_vi_for_offset,
26024         create_function_info_for, create_variable_info_for_1,
26025         create_variable_info_for, intra_create_variable_infos): Adjust.
26026         (init_base_vars): Push NULL for ID zero.
26027         (compute_points_to_sets): Adjust.
26029 2013-03-20  Richard Biener  <rguenther@suse.de>
26031         * cfgloop.c (verify_loop_structure): Streamline and avoid
26032         ICEing on corrupt loop tree.
26033         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
26034         loop tree.
26036 2013-03-20  Richard Biener  <rguenther@suse.de>
26038         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
26039         check whether an SSA update is needed.
26041 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
26043         * config/mips/constraints.md (T): Rename to...
26044         (Yf): ...this.
26045         (U): Rename to...
26046         (Yd): ...this.
26047         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
26048         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
26050 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
26052         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
26053         (*subsi3_carryin_uxtw): Likewise.
26055 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
26057         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
26058         (*rorsi3_insn_uxtw): Likewise.
26060 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
26062         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
26063         (*extrsi5_insn_uxtw): Likewise.
26065 2013-03-19  Richard Biener  <rguenther@suse.de>
26067         PR tree-optimization/56273
26068         * passes.c (init_optimization_passes): Move second VRP after DOM.
26070 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
26072         * config/i386/i386.md (*movti_internal): Merge from
26073         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
26074         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
26075         nox64 isa attributes.
26077 2013-03-18  Richard Biener  <rguenther@suse.de>
26079         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
26080         (unite): Likewise.
26081         (merge_node_constraints): Likewise.
26082         (build_succ_graph): Likewise.
26083         (valid_graph_edge): Inline into single caller.
26084         (unify_nodes): Likewise.  Use bitmap_set_bit return value
26085         and cache varinfo.
26086         (scc_visit): Fix formatting and variable use.
26087         (do_sd_constraint): Use gcc_checking_assert.
26088         (do_ds_constraint): Likewise.
26089         (do_complex_constraint): Likewise.
26090         (condense_visit): Likewise.  Cleanup.
26091         (dump_pred_graph): New function.
26092         (perform_var_substitution): Dump the pred-graph before
26093         variable substitution.
26094         (find_equivalent_node): Use gcc_checking_assert.
26095         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
26097 2013-03-18  Richard Biener  <rguenther@suse.de>
26099         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
26100         Remove cond_expr_stmt_list argument and do not gimplify the
26101         built expression.
26102         (vect_loop_versioning): Adjust.
26103         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
26104         Cleanup to use less temporaries.
26105         (vect_create_data_ref_ptr): Cleanup.
26107 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
26109         PR tree-optimization/56635
26110         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
26111         require types_compatible_p types.
26113 2013-03-18  Nick Clifton  <nickc@redhat.com>
26115         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
26116         spurious backslash.
26118         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
26119         Add missing line to comment describing function.
26121 2013-03-18  Richard Biener  <rguenther@suse.de>
26123         PR tree-optimization/56210
26124         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26125         Handle string / character search functions.
26126         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
26128 2013-03-18  Richard Biener  <rguenther@suse.de>
26130         PR middle-end/56483
26131         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
26132         and implement properly.
26133         * gimple.h (gimple_cond_single_var_p): Remove.
26135 2013-03-18  Richard Biener  <rguenther@suse.de>
26137         * tree-data-ref.h (find_data_references_in_loop): Declare.
26138         * tree-data-ref.c (get_references_in_stmt): Use a stack
26139         vector pre-allocated in the callers.
26140         (find_data_references_in_stmt): Adjust.
26141         (graphite_find_data_references_in_stmt): Likewise.
26142         (create_rdg_vertices): Likewise.
26143         (find_data_references_in_loop): Export.
26144         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
26145         Compute dependences here...
26146         (vect_analyze_data_refs): ...not here.  When we encounter
26147         a non-vectorizable data reference in basic-block vectorization
26148         truncate the data reference vector.  Do not bother to
26149         fixup data-dependence information for gather loads.
26150         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
26151         of data references, as reported.
26153 2013-03-18  Richard Biener  <rguenther@suse.de>
26155         PR tree-optimization/3713
26156         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
26157         has_constants and expr.
26158         (stmt_has_constants): Properly valueize SSA names when deciding
26159         whether the stmt has constants.
26161 2013-03-18  Richard Biener  <rguenther@suse.de>
26163         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
26164         whole function when there is nothing to do.
26165         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
26166         * tree-vectorizer.c (vectorize_loops): Update virtual and
26167         loop-closed SSA once.
26168         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
26170 2013-03-18  Richard Biener  <rguenther@suse.de>
26172         PR middle-end/56113
26173         * domwalk.c (bb_postorder): New global static.
26174         (cmp_bb_postorder): New function.
26175         (walk_dominator_tree): Replace scheme imposing an order for
26176         visiting dominator sons by one sorting them at the time they
26177         are pushed on the stack.
26179 2013-03-18  Richard Biener  <rguenther@suse.de>
26181         PR tree-optimization/39326
26182         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
26183         (struct mem_ref): Replace mem member with ao_ref typed member.
26184         (MEM_ANALYZABLE): Adjust.
26185         (memref_eq): Likewise.
26186         (mem_ref_alloc): Likewise.
26187         (gather_mem_refs_stmt): Likewise.
26188         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
26189         (execute_sm_if_changed_flag_set): Adjust.
26190         (execute_sm): Likewise.
26191         (ref_always_accessed_p): Likewise.
26192         (refs_independent_p): Likewise.
26193         (can_sm_ref_p): Likewise.
26195 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
26197         PR c/56566
26198         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
26199         return 1 even for !unsignedp.
26201 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
26203         * config/i386/i386.md (isa): Add x64 and nox64.
26204         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
26205         (*pushtf): Enable *roF alternative for x64 isa only.
26206         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
26207         mode attribute of integer alternatives to DImode for TARGET_64BIT.
26208         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
26209         (*movtf_internal): Merge from *movtf_internal_rex64 and
26210         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
26211         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
26212         nox64 isa attributes.
26213         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
26214         nox64 isa attributes.
26215         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
26217 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
26219         * config/alpha/alpha.c (TARGET_LRA_P): New define.
26221 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
26223         PR target/56640
26224         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
26225         class names.  Remove trailing comma after "ALL_REGS".
26227 2013-03-16  Jan Hubicka  <jh@suse.cz>
26229         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
26230         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
26231         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
26232         of cgraph_get_create_node.
26233         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
26235 2013-03-16  Jason Merrill  <jason@redhat.com>
26237         PR debug/49090
26238         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
26239         with DW_AT_default_value.
26241 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
26243         * BASE-VER: Set to 4.9.0.
26245 2013-03-14  Andi Kleen  <ak@linux.intel.com>
26247         PR target/56619
26248         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
26249         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
26250         Document _x* TSX intrinsics.
26252 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
26253             David Holsgrove  <david.holsgrove@xilinx.com>
26255         * configure.ac: Add MicroBlaze TLS support detection.
26256         * configure: Regenerate.
26257         * config/microblaze/microblaze-protos.h
26258         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
26259         symbol_mentioned_p, label_mentioned_p): Add prototypes.
26260         * config/microblaze/microblaze.c (microblaze_address_type): Add
26261         ADDRESS_TLS and tls_reloc address types.
26262         (microblaze_address_info): Add tls_reloc.
26263         (TARGET_HAVE_TLS): Define.
26264         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
26265          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
26266          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
26267         load_tls_operand,  microblaze_call_tls_get_addr,
26268         microblaze_legitimize_tls_address): New functions.
26269         (microblaze_classify_unspec): Handle UNSPEC_TLS.
26270         (get_base_reg): Use microblaze_tls_symbol_p.
26271         (microblaze_classify_address): Handle TLS.
26272         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
26273         label_mentioned_p and microblaze_tls_referenced_p.
26274         (microblaze_legitimize_address): Handle TLS.
26275         (microblaze_address_insns): Handle ADDRESS_TLS.
26276         (pic_address_needs_scratch): Handle TLS.
26277         (print_operand_address): Handle TLS.
26278         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
26279         (microblaze_expand_move): Handle TLS.
26280         (microblaze_legitimate_constant_p): Check
26281         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
26282         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
26283         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
26284         (PIC_OFFSET_TABLE_REGNUM): Set.
26285         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
26286         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
26287         (addsi3, movsi_internal2, movdf_internal): Update constraints
26288         * config/microblaze/predicates.md (arith_plus_operand): Define
26289         (move_operand): Redefine as move_src_operand,
26290         check microblaze_tls_referenced_p.
26292 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
26294         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
26295         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
26297 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
26299         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
26300         CC mode for AND.
26302 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
26304         PR tree-optimization/53265
26305         * common.opt (Waggressive-loop-optimizations): New option.
26306         * tree-ssa-loop-niter.c: Include tree-pass.h.
26307         (do_warn_aggressive_loop_optimizations): New function.
26308         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
26309         if number_of_latch_executions returned constant.
26310         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
26311         early.  If number_of_latch_executions returned constant, set
26312         nb_iterations_upper_bound back to it.
26313         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
26314         field.
26315         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
26316         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
26318         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
26319         (MULTILIB_OSDIRNAMES): Set.
26320         * genmultilib: If defaultosdirname doesn't start with :: , set
26321         defaultosdirname2 instead, clear it and emit two . multilib_raw
26322         entries instead of just one.
26324 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
26326         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
26327         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
26328         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
26329         (SUBTARGET_OVERRIDE_OPTIONS): New.
26331 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
26333         PR target/49880
26334         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
26335         (musermode): Convert to Var(TARGET_USERMODE).
26336         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
26337         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
26338         * config/sh/sh.c (sh_option_override): Use
26339         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
26340         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
26341         condition.
26342         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
26343         TARGET_SH4.
26344         (udivsi3_i4_single, divsi3_i4_single): Use
26345         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
26347 2013-03-13  Dave Korn  <dave.korn.cygwin@gmail.com>
26349         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
26350         default setting.
26352 2013-03-13  Richard Biener  <rguenther@suse.de>
26354         PR tree-optimization/56608
26355         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
26356         calls when vectorizing basic-blocks.
26358 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
26360         PR plugins/45078
26361         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
26362         tm_file.
26364 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
26366         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
26368 2013-03-11  Jan Hubicka  <jh@suse.cz>
26370         PR lto/56557
26371         * lto-streamer-out.c (output_symbol_p): Skip references from
26372         constructors of external variables.
26374 2013-03-11  Jan Hubicka  <jh@suse.cz>
26376         PR middle-end/56571
26377         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
26378         from pseudos.
26379         * emit-rtl.c (verify_rtx_sharing): Likewise.
26380         (copy_insn_1): Likewise.
26381         * rtl.c (copy_rtx): Likewise.
26383 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
26385         PR target/56591
26386         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
26387         output_operand_lossage message.
26389 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
26391         PR target/56470
26392         * arm.c (shift_op): Validate RTL pattern on the fly.
26393         (arm_print_operand, case 'S'): Don't use shift_operator to validate
26394         the RTL.
26396 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26398         PR target/56347
26399         * config/pa/pa.md (call_value): Check for calls to powf and direct to
26400         new call patterns that clobber %fr12.
26401         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
26402         split and postreload patterns.
26403         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
26404         registers %fr12 and %fr12R as call used.
26406 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
26408         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
26409         (canon_address, record_store, replace_read, check_mem_read_rtx,
26410         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
26411         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
26412         rest_of_handle_dse): Likewise.
26414 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
26416         PR middle-end/56524
26417         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
26418         Add base_optabs.
26419         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
26420         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
26421         (save_optabs_if_changed): Replace with...
26422         (init_tree_optimization_optabs): ...this.
26423         * optabs.c (save_optabs_if_changed): Rename to...
26424         (init_tree_optimization_optabs): ...this.  Take the optimization node
26425         as argument.  Do nothing if the base optabs are already correct.
26426         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
26427         to recompute optabs.
26428         * function.h (function): Remove optabs field.
26429         * function.c (invoke_set_current_function_hook): Call
26430         init_tree_optimization_optabs.  Use the result to initialize
26431         this_fn_optabs.
26433 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
26435         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
26436         if GTMA_HAS_NO_INSTRUMENTATION.
26437         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
26438         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
26439         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
26440         * gimple-pretty-print.c (dump_gimple_transaction): Handle
26441         GTMA_HAS_NO_INSTRUMENTATION.
26443 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
26445         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
26446         libasan_preinit.o.
26448 2013-03-08  Marek Polacek  <polacek@redhat.com>
26449             Jakub Jelinek  <jakub@redhat.com>
26451         PR tree-optimization/56478
26452         * predict.c (is_comparison_with_loop_invariant_p): Change the
26453         type of loop_step to tree.
26454         (predict_loops): Adjust.
26455         (predict_iv_comparison): Perform the computations on double_ints.
26457 2013-03-08  Richard Biener  <rguenther@suse.de>
26459         PR tree-optimization/56570
26460         * tree-cfg.c (verify_expr_location_1): Verify locations for
26461         DECL_DEBUG_EXPR.
26462         * tree-sra.c (create_access_replacement): Strip locations
26463         from DECL_DEBUG_EXPRs.
26465 2013-03-08  Richard Biener  <rguenther@suse.de>
26467         * tree-inline.c (expand_call_inline): Do not associate
26468         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
26469         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
26471 2013-03-08  Richard Biener  <rguenther@suse.de>
26473         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
26474         or block changes with -Og.  Fix for location / block encoding
26475         changes and PHI arguments with locations.
26477 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
26479         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
26480         for all counters.
26481         (struct output_info): Likewise.
26482         (register_overhead): Remove bad gcc_assert.
26483         (bitmap_find_bit): If there is only a single bitmap element, do not
26484         count a miss as a search.
26485         (print_statistics): Update for counter type changes.
26486         (dump_bitmap_statistics): Likewise.  Print headers such that they
26487         are properly lined up with the printed counters.
26489 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
26491         PR tree-optimization/56559
26492         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
26493         check that it has only a single use.
26495 2013-03-07  Richard Biener  <rguenther@suse.de>
26497         * doc/invoke.texi (fwhole-program): Discourage use in combination
26498         with -flto.
26500 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
26502         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
26504         PR tree-optimization/56539
26505         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
26506         instead of GSI_CONTINUE_LINKING as last argument to
26507         force_gimple_operand_gsi.  Adjust function comment.
26509         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
26510         aarch64-cores.def.
26512         PR middle-end/56548
26513         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
26514         promoted mode, convert the result back to the original mode.
26516 2013-03-06  Richard Biener  <rguenther@suse.de>
26518         PR middle-end/56294
26519         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
26520         (insert_updated_phi_nodes_compare_uids): New function.
26521         (update_ssa): Sort symbols_to_rename after UID before
26522         traversing it to insert PHI nodes.
26524 2013-03-06  Richard Biener  <rguenther@suse.de>
26526         PR middle-end/50494
26527         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
26528         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
26530         Revert
26531         2013-02-13  Richard Biener  <rguenther@suse.de>
26533         PR lto/50494
26534         * varasm.c (output_constant_def_1): Get the decl representing
26535         the constant as argument.
26536         (output_constant_def): Wrap output_constant_def_1.
26537         (make_decl_rtl): Use output_constant_def_1 with the decl
26538         representing the constant.
26539         (build_constant_desc): Optionally re-use a decl already
26540         representing the constant.
26541         (tree_output_constant_def): Adjust.
26543 2013-03-06  Joey Ye  <joey.ye@arm.com>
26545         PR lto/50293
26546         * gcc.c (convert_white_space): New function.
26547         (main): Handles white space in function name.
26549 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
26551         PR target/56529
26552         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
26553         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
26554         to SH_DIV_CALL_TABLE for TARGET_SH2.
26555         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
26556         list.
26557         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
26558         call-table options.
26560 2013-03-05  Sterling Augustine  <saugustine@google.com>
26561             Cary Coutant  <ccoutant@google.com>
26563         PR debug/55364
26564         * dwarf2out.c (resolve_addr): Don't call
26565         remove_loc_list_addr_table_entries a second time for the same
26566         expression.
26568 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
26570         PR debug/56510
26571         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
26572         (avoid_complex_debug_insns): New function.
26573         (expand_debug_locations): Call it.
26575         PR rtl-optimization/56484
26576         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
26577         lifetimes of hard registers on small register class machines.
26579 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
26581         * config/microblaze/microblaze-protos.h: Rename
26582         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
26583         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
26584         fast_interrupt.
26585         (microblaze_fast_interrupt_function_p): New function.
26586         (microblaze_is_interrupt_handler): Rename to
26587         microblaze_is_interrupt_variant and add fast_interrupt check.
26588         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
26589         (save_restore_insns): Likewise.
26590         (compute_frame_size): Likewise.
26591         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
26592         (microblaze_globalize_label): Likewise.
26593         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
26594         * config/microblaze/microblaze.md: Use wrapper
26595         microblaze_is_interrupt_variant.
26597 2013-03-05  Kai Tietz  <ktietz@redhat.com>
26599         * sdbout.c (sdbout_one_type): Switch to current function's section
26600         supporting cold/hot.
26602 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
26604         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
26605         -mxl-reorder.
26607 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
26609         PR middle-end/56461
26610         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
26611         if VALGRIND_GET_VBITS is defined, temporarily make object
26612         memory all defined, and restore previous valgrind addressability
26613         and definability afterwards.  Free this_object at the end.
26615         PR middle-end/56461
26616         * lra.c (lra): Call lra_clear_live_ranges if live_p,
26617         right before calling lra_create_live_ranges, also call it
26618         when clearing live_p.  Only call lra_clear_live_ranges
26619         at the end if live_p.
26621         PR middle-end/56461
26622         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
26624 2013-03-05  Richard Biener  <rguenther@suse.de>
26626         PR tree-optimization/56521
26627         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
26628         value-id.
26630 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
26632         PR c++/55135
26633         * except.h (remove_unreachable_eh_regions): New prototype.
26634         * except.c (remove_eh_handler_splicer): New function, split out
26635         of remove_eh_handler.
26636         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
26637         warning about running it on many EH regions one at a time.
26638         (remove_unreachable_eh_regions_worker): New function, walk the
26639         EH tree in depth-first order and remove non-marked regions.
26640         (remove_unreachable_eh_regions): New function.
26641         * tree-eh.c (mark_reachable_handlers): New function, split out
26642         from remove_unreachable_handlers.
26643         (remove_unreachable_handlers): Use mark_reachable_handlers and
26644         remove_unreachable_eh_regions.
26645         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
26646         and remove_unreachable_eh_regions.
26648 2013-03-05  Richard Biener  <rguenther@suse.de>
26650         PR middle-end/56525
26651         * loop-init.c (fix_loop_structure): Remove loops in two stages,
26652         not freeing them until the end.
26654 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26656         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
26658 2013-03-05  Richard Biener  <rguenther@suse.de>
26660         PR tree-optimization/56270
26661         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
26662         of loads after scheduling an SLP instance.
26664 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
26666         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
26667         tic6x.exp.
26668         (check_gcc_parallelize): Run guality.exp as a separate job from
26669         vect.exp with unsorted.exp and $(dg_target_exps) separately from
26670         struct-layout-1.exp with stackalign.exp.
26672         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
26674         PR middle-end/56461
26675         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
26676         load_index sbitmap even if some bit in it isn't set.
26678         PR middle-end/56461
26679         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
26680         (discover_iteration_bound_by_body_walk): Change queues to
26681         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
26682         spelling in comment.  Call safe_push on queues[bound_index] directly.
26683         Release queues[queue_index] in every iteration unconditionally.
26684         Release bounds vector.
26686         PR middle-end/56461
26687         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
26688         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
26689         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
26690         inner_phis vector.
26692 2013-03-05  Richard Biener  <rguenther@suse.de>
26694         PR lto/56515
26695         * tree-inline.c (remap_blocks_to_null): New function.
26696         (expand_call_inline): When expanding a call stmt without
26697         an associated block inline remap all callee blocks to NULL.
26699 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
26701         PR rtl-optimization/56494
26702         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
26703         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
26704         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
26706         PR middle-end/56461
26707         * sel-sched-ir.c (free_sched_pools): Release
26708         succs_info_pool.stack[succs_info_pool.max_top] vectors too
26709         if succs_info_pool.max_top isn't -1.
26711         PR bootstrap/56509
26712         * opts.c (opts_obstack, opts_concat): Moved to...
26713         * opts-common.c (opts_obstack, opts_concat): ... here.
26715 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
26717         PR middle-end/56461
26718         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
26720 2013-03-04  Martin Jambor  <mjambor@suse.cz>
26722         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
26723         all appropriate places.
26725 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
26727         PR tree-optimization/56424
26728         * ipa-split.c (split_function): Do not set the RSO flag if result is
26729         not by reference and its type is a register type.
26731 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
26733         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
26734         (microblaze_legitimate_pic_operand): Likewise
26735         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
26736         new function microblaze_legitimate_pic_operand
26737         * config/microblaze/microblaze-protos.h
26738         (microblaze_legitimate_pic_operand): Declare.
26740 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
26742         * config/microblaze/predicates.md (call_insn_simple_operand):
26743         New predicate for supported rtx code types.
26744         * config/microblaze/microblaze.md (call_internal1): Use
26745         call_insn_simple_operand predicate.
26747 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
26749         PR middle-end/56461
26750         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
26751         partitions.ordered_remove.
26753         PR middle-end/56461
26754         * tree-vect-stmts.c (vectorizable_conversion): Don't call
26755         vec_oprnds0.create (1) for modifier == NONE.
26757         PR middle-end/56461
26758         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
26759         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
26760         vec_oprnds1 right before pushing anything to it for
26761         scalar_shift_arg.
26763         PR middle-end/56461
26764         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
26765         set nbbs to 0 instead of having separate code path.
26766         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
26767         instead of false as last argument if returning NULL.
26769 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
26771         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
26772         the attribute is now called "target" instead of "option".
26773         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
26774         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
26775         attribute/pragma name for TARGET_OPTION_VALID_P and
26776         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
26777         * doc/tm.texi: Regenerated.
26779 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
26781         * config/microblaze/microblaze.c:
26782         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
26783         * config/microblaze/microblaze.h: Add -mxl-reorder to
26784         DRIVER_SELF_SPECS.
26785         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
26786         instructions emitted if TARGET_REORDER.
26787         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
26788         or 0 for -m/-mno case, but initialises as 2 to detect default use case
26789         separately.
26791 2013-03-01  Xinliang David Li  <davidxl@google.com>
26793         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
26794         walk length.
26796 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
26798         PR middle-end/56461
26799         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
26800         vector even when returning true.  Fix up function comment formatting.
26802         PR middle-end/56461
26803         * ira-build.c (ira_loop_nodes_count): New variable.
26804         (create_loop_tree_nodes): Initialize it.
26805         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
26807         PR middle-end/56461
26808         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
26809         method on dr_chain and result_chain.
26810         * tree-vect-stmts.c (vectorizable_store): Only call
26811         result_chain.create if j == 0.
26813         PR middle-end/56461
26814         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
26815         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
26816         before overwriting it.
26818 2013-03-01  Tobias Burnus  <burnus@net-b.de>
26820         * doc/extended.texi (C Extensions): Change order in @menu
26821         to match @node.
26822         (Other MIPS Built-in Functions): Move last MIPS entry before
26823         "picoChip Built-in Functions".
26824         (SH Built-in Functions): Move after RX Built-in Functions.
26825         * doc/gcc.texi (Introduction): Change order in @menu
26826         to match @node.
26827         * doc/md.texi (Constraints): Ditto.
26828         * gty.texi (Type Information): Ditto.
26829         (User-provided marking routines for template types): Make
26830         subsection.
26831         * doc/invoke.texi (AArch64 Options): Move before
26832         "Adapteva Epiphany Options".
26834 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
26835             Jakub Jelinek  <jakub@redhat.com>
26837         PR sanitizer/56454
26838         * asan.c (gate_asan): Lookup no_sanitize_address instead of
26839         no_address_safety_analysis attribute.
26840         * doc/extend.texi (no_address_safety_attribute): Rename to
26841         no_sanitize_address attribute, mention no_address_safety_analysis
26842         attribute as deprecated alias.
26844 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
26846         PR middle-end/56461
26847         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
26848         type to vec<vec<tree> > *.
26849         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
26850         to be vec<tree> instead of vec<tree> *, set vec_defs
26851         to vNULL and call vec_defs.create (number_of_vects), adjust other
26852         uses of vec_defs.
26853         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
26854         vectorizable_condition): Adjust vect_get_slp_defs callers.
26856 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
26858         * config/aarch64/aarch64.c
26859         (aarch64_float_const_representable): Remove unused variable.
26861 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
26863         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
26865 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
26867         * config/aarch64/aarch64-builtins.c
26868         (aarch64_init_simd_builtins): Make static.
26870 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
26872         * config/aarch64/aarch64.c
26873         (aarch64_simd_make_constant): Make static.
26875 2013-02-28  Martin Jambor  <mjambor@suse.cz>
26877         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
26878         with no initialization to the RHS of debug statements.
26880 2013-02-28  Martin Jambor  <mjambor@suse.cz>
26882         PR tree-optimization/56294
26883         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
26884         Adjust dumping.
26885         (get_access_replacement): Do not call create_access_replacement.
26886         Assert a replacement exists.
26887         (get_repl_default_def_ssa_name): Create the replacement declaration
26888         itself.
26890 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
26892         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
26893         final_end_function.
26895 2013-02-28  Marek Polacek  <polacek@redhat.com>
26897         PR rtl-optimization/56466
26898         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
26899         if we're changing a loop.
26900         (peel_loops_completely): Likewise.
26902 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
26904         PR c++/55813
26905         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
26907 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
26909         PR target/56445
26910         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
26911         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
26912         INTX_FTYPE_FX, FX_FTYPE_INTX.
26913         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
26915 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
26917         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
26918         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
26919         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
26920         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
26921         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
26922         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
26923         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
26924         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
26925         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
26926         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
26927         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
26928         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
26929         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
26930         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
26931         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
26932         (avrxmega6): Increase max flash segments from 5 to 6.
26933         * config/avr/t-multilib: Regenerate.
26934         * config/avr/avr-tables.opt: Regenerate.
26935         * doc/avr-mmcu.texi: Regenerate.
26937 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
26939         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
26940         (avr_device_to_arch): Rename to avr_device_to_ld.
26941         (avr_device_to_as): New prototype.
26942         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
26943         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
26944         * config/avr/driver-avr.c (avr_device_to_as): New.
26945         (avr_device_to_arch): Rename to avr_device_to_ld.
26947 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
26949         PR middle-end/56461
26950         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
26951         method on dr_chain and result_chain.
26953         PR middle-end/56461
26954         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
26955         pointer_set_destroy on not_executed_last_iteration.
26957         PR middle-end/56461
26958         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
26960         PR middle-end/56461
26961         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
26962         FOR_EACH_DEFINED_FUNCTION when freeing state.
26964         PR middle-end/56461
26965         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
26966         pool_free.
26967         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
26968         overwriting it.
26970         PR middle-end/56461
26971         * ipa-cp.c (decide_whether_version_node): Call vec_free on
26972         known_aggs[i].items and release known_aggs vector.
26974         PR middle-end/56461
26975         * ipa-reference.c (propagate): Free node_info even for alias nodes.
26977 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
26979         * config/microblaze/microblaze.c (microblaze_emit_compare):
26980         Use xor for EQ/NE comparisions.
26981         * config/microblaze/microblaze.md (cstoresf4): Add constraints
26982         (cbranchsf4): Adjust operator to comparison_operator.
26984 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
26986         PR middle-end/56461
26987         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
26988         vector.
26989         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
26990         vec_safe_push, always update *slot.
26991         (redirect_edge_var_map_clear): Use vec_free.
26992         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
26993         (free_var_map_entry): Use vec_free.
26994         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
26995         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
26997 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
26999         PR middle-end/45472
27000         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
27001         when the may_trap_p bit of the exprs being merged differs.
27002         Reorder tests for speculativeness in the logical and operator.
27004 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
27006         * incpath.c (add_standard_paths): Use reconcat instead of concat
27007         where appropriate and avoid leaking memory.
27009         * opts.h: Include obstack.h.
27010         (opts_concat): New prototype.
27011         (opts_obstack): New declaration.
27012         * opts.c (opts_concat): New function.
27013         (opts_obstack): New variable.
27014         (init_options_struct): Call gcc_init_obstack on opts_obstack.
27015         (finish_options): Use opts_concat instead of concat
27016         and XOBNEWVEC instead of XNEWVEC.
27017         * opts-common.c (generate_canonical_option, decode_cmdline_option,
27018         generate_option): Likewise.
27019         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
27020         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
27022         PR target/56455
27023         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
27024         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
27026 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
27028         PR middle-end/56461
27029         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
27031 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
27033         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
27034         (arm_block_move_unaligned_straight): Likewise.
27035         (arm_adjust_block_mem): Likewise.
27037 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
27039         PR target/48901
27040         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
27041         temp, cond and label.
27042         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
27044         PR target/52500
27045         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
27046         * config/c6x/c6x.h (dbx_register_map): Update declaration.
27048         PR target/52501
27049         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
27050         of prologue/epilogue functions.
27052         PR target/52550
27053         * config/tilegx/tilegx.c (tilegx_expand_prologue):
27054         Remove unused variable cfa_offset.
27055         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
27057         PR target/54639
27058         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
27059         type promotion to unsigned.
27061         PR target/54640
27062         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
27063         for HOST_WIDE_INT of 32 bit / same size as int.
27064         (arm_block_move_unaligned_straight): Likewise.
27065         (arm_adjust_block_mem): Likewise.
27067         PR target/54662
27068         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
27069         ALL_CFLAGS.
27071 2013-02-26  Marek Polacek  <polacek@redhat.com>
27073         PR tree-optimization/56426
27074         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
27076 2013-02-26  Richard Biener  <rguenther@suse.de>
27078         PR target/56444
27079         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
27080         unused variable loops.
27082 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
27084         PR tree-optimization/56448
27085         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
27086         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
27087         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
27088         later operands of the references, or even first operand for
27089         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
27091         PR tree-optimization/56443
27092         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
27093         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
27094         to type_for_mode langhook.
27096 2013-02-25  Matt Turner  <mattst88@gmail.com>
27098         * doc/invoke.texi: Document r4700.
27100 2013-02-25  Richard Biener  <rguenther@suse.de>
27102         PR tree-optimization/56175
27103         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
27104         split out from ...
27105         (simplify_bitwise_binary): ... here.  Also guard the conversion
27106         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
27108 2013-02-25  Catherine Moore  <clm@codesourcery.com>
27110         Revert:
27111         2013-02-24  Catherine Moore  <clm@codesourcery.com>
27112             Maciej W. Rozycki  <macro@codesourcery.com>
27113             Tom de Vries  <tom@codesourcery.com>
27114             Nathan Sidwell  <nathan@codesourcery.com>
27115             Iain Sandoe  <iain@codesourcery.com>
27116             Nathan Froyd  <froydnj@codesourcery.com>
27117             Chao-ying Fu  <fu@mips.com>
27119         * doc/extend.texi: (micromips, nomicromips, nocompression):
27120         Document new function attributes.
27121         * doc/invoke.texi (minterlink-compressed, mmicromips,
27122         m14k, m14ke, m14kec): Document new options.
27123         (minterlink-mips16): Update documentation.
27124         * doc/md.texi (ZC, ZD): Document new constraints.
27125         * configure.ac (gcc_cv_as_micromips): Check if linker
27126         supports the .set micromips directive.
27127         * configure: Regenerate.
27128         * config.in: Regenerate.
27129         * config/mips/mips-tables.opt: Regenerate.
27130         * config/mips/micromips.md: New file.
27131         * constraints.md (ZC, AD): New constraints.
27132         * config/mips/predicates.md (movep_src_register): New predicate.
27133         (movep_src_operand): New predicate.
27134         (non_volatile_mem_operand): New predicate.
27135         * config/mips/mips.md (multimem): New type.
27136         (length): Differentiate between 17-bit and 18-bit branch offsets.
27137         (MOVEP1, MOVEP2): New mode iterator.
27138         (mov_<load>l): Use ZC constraint.
27139         (mov_<load>r): Likewise.
27140         (mov_<store>l): Likewise.
27141         (mov_<store>r): Likewise.
27142         (*branch_equality<mode>_inverted): Add microMIPS support.
27143         (*branch_equality<mode>): Likewise.
27144         (*jump_absolute): Likewise.
27145         (indirect_jump_<mode>): Likewise.
27146         (tablejump_<mode>): Likewise.
27147         (<optab>_internal): Likewise.
27148         (sibcall_internal): Likewise.
27149         (sibcall_value_internal): Likewise.
27150         (prefetch): Use constraint ZD.
27151         * config/mips/mips.opt (minterlink-compressed): New option.
27152         (minterlink-mips16): Now an alias for minterlink-compressed.
27153         (mmicromips): New option.
27154         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
27155         (compare_and_swap_12): Likewise.
27156         (sync_add<mode>): Likewise.
27157         (sync_<optab>_12): Likewise.
27158         (sync_old_<optab>_12): Likewise.
27159         (sync_new_<optab>_12): Likewise.
27160         (sync_nand_12): Likewise.
27161         (sync_old_nand_12): Likewise.
27162         (sync_new_nand_12): Likewise.
27163         (sync_sub<mode>): Likewise.
27164         (sync_old_add<mode>): Likewise.
27165         (sync_old_sub<mode>): Likewise.
27166         (sync_new_add<mode>): Likewise.
27167         (sync_new_sub<mode>): Likewise.
27168         (sync_<optab><mode>): Likewise.
27169         (sync_old_<optab><mode>): Likewise.
27170         (sync_new_<optab><mode>): Likewise.
27171         (sync_nand<mode>): Likewise.
27172         (sync_old_nand<mode>): Likewise.
27173         (sync_new_nand<mode>): Likewise.
27174         (sync_lock_test_and_set<mode>): Likewise.
27175         (test_and_set_12): Likewise.
27176         (atomic_compare_and_swap<mode>): Likewise.
27177         (atomic_exchange<mode>_llsc): Likewise.
27178         (atomic_fetch_add<mode>_llsc): Likewise.
27179         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
27180         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
27181         (umips_save_restore_pattern_p): Likewise.
27182         (umips_load_store_pair_p): Likewise.
27183         (umips_output_load_store_pair): Likewise.
27184         (umips_movep_target_p): Likewise.
27185         (umips_12bit_offset_address_p): Likewise.
27186         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
27187         (mips_base_mips16): Rename this...
27188         (mips_base_compression_flags): ...to this. Update all uses.
27189         (mips_attribute_table): Add micromips, nomicromips and nocompression.
27190         (mips_mips16_decl_p): Delete.
27191         (mips_nomips16_decl_p): Delete.
27192         (mips_get_compress_on_flags): New function.
27193         (mips_get_compress_off_flags): New function.
27194         (mips_get_compress_mode): New function.
27195         (mips_get_compress_on_name): New function.
27196         (mips_get_compress_off_name): New function.
27197         (mips_insert_attributes): Support multiple compression types.
27198         (mips_merge_decl_attributes): Likewise.
27199         (umips_12bit_offset_address_p): New function.
27200         (mips_start_function_definition): Emit .set micromips directive.
27201         (mips_call_may_need_jalx_p): New function.
27202         (mips_function_ok_for_sibcall): Add microMIPS support.
27203         (mips_print_operand_punctuation): Support short delay slots and
27204         compact jumps.
27205         (umips_swm_mask, umips_swm_encoding): New.
27206         (umips_build_save_restore): New function.
27207         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
27208         (was_mips16_p): Remove.
27209         (old_compression_mode): New.
27210         (mips_set_compression_mode): New function.
27211         (mips_set_current_function): Add microMIPS support.
27212         (mips_option_override): Likewise.
27213         (umips_save_restore_pattern_p): New function.
27214         (umips_output_save_restore): New function.
27215         (umips_load_store_pair_p_1): New function.
27216         (umips_load_store_pair_p): New function.
27217         (umips_output_load_store_pair_1): New function.
27218         (umips_output_load_store_pair): New function.
27219         (umips_movep_target_p) New function.
27220         (mips_prepare_pch_save): Add microMIPS support.
27221         * config/mips/mips.h (TARGET_COMPRESSION): New.
27222         (TARGET_CPU_CPP_BUILTINS): Update macro
27223         to use new compression flags and to support microMIPS.
27224         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
27225         (MIPS_ARCH_FLOAT_SPEC): Likewise.
27226         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
27227         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
27228         (ASM_SPEC): Support mmicromips and mno-micromips.
27229         (M16STORE_REG_P): New macro.
27230         (MIPS_CALL): Support TARGET_MICROMIPS.
27231         (MICROMIPS_J): New macro.
27232         (mips_base_mips16): Rename this...
27233         (mips_base_compression_flags): ...to this.
27234         (UMIPS_12BIT_OFFSET_P): New macro.
27235         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
27236         (MULTILIB_DIRNAMES): Likewise.
27238 2013-02-25  Tom de Vries  <tom@codesourcery.com>
27240         PR rtl-optimization/56131
27241         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
27242         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
27243         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
27245 2013-02-25  Tobias Burnus  <burnus@net-b.de>
27247         * doc/invoke.texi (-fsanitize=): Move from optimization
27248         to debugging options.
27250 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
27252         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
27254 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
27255             Alexander Monakov  <amonakov@ispras.ru>
27257         PR middle-end/56077
27258         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
27259         flush pending lists also on non-jumps.  Adjust comment.
27261 2013-02-24  Catherine Moore  <clm@codesourcery.com>
27262             Maciej W. Rozycki  <macro@codesourcery.com>
27263             Tom de Vries  <tom@codesourcery.com>
27264             Nathan Sidwell  <nathan@codesourcery.com>
27265             Iain Sandoe  <iain@codesourcery.com>
27266             Nathan Froyd  <froydnj@codesourcery.com>
27267             Chao-ying Fu  <fu@mips.com>
27269         * doc/extend.texi: (micromips, nomicromips, nocompression):
27270         Document new function attributes.
27271         * doc/invoke.texi (minterlink-compressed, mmicromips,
27272         m14k, m14ke, m14kec): Document new options.
27273         (minterlink-mips16): Update documentation.
27274         * doc/md.texi (ZC, ZD): Document new constraints.
27275         * configure.ac (gcc_cv_as_micromips): Check if linker
27276         supports the .set micromips directive.
27277         * configure: Regenerate.
27278         * config.in: Regenerate.
27279         * config/mips/mips-tables.opt: Regenerate.
27280         * config/mips/micromips.md: New file.
27281         * constraints.md (ZC, AD): New constraints.
27282         * config/mips/predicates.md (movep_src_register): New predicate.
27283         (movep_src_operand): New predicate.
27284         (non_volatile_mem_operand): New predicate.
27285         * config/mips/mips.md (multimem): New type.
27286         (length): Differentiate between 17-bit and 18-bit branch offsets.
27287         (MOVEP1, MOVEP2): New mode iterator.
27288         (mov_<load>l): Use ZC constraint.
27289         (mov_<load>r): Likewise.
27290         (mov_<store>l): Likewise.
27291         (mov_<store>r): Likewise.
27292         (*branch_equality<mode>_inverted): Add microMIPS support.
27293         (*branch_equality<mode>): Likewise.
27294         (*jump_absolute): Likewise.
27295         (indirect_jump_<mode>): Likewise.
27296         (tablejump_<mode>): Likewise.
27297         (<optab>_internal): Likewise.
27298         (sibcall_internal): Likewise.
27299         (sibcall_value_internal): Likewise.
27300         (prefetch): Use constraint ZD.
27301         * config/mips/mips.opt (minterlink-compressed): New option.
27302         (minterlink-mips16): Now an alias for minterlink-compressed.
27303         (mmicromips): New option.
27304         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
27305         (compare_and_swap_12): Likewise.
27306         (sync_add<mode>): Likewise.
27307         (sync_<optab>_12): Likewise.
27308         (sync_old_<optab>_12): Likewise.
27309         (sync_new_<optab>_12): Likewise.
27310         (sync_nand_12): Likewise.
27311         (sync_old_nand_12): Likewise.
27312         (sync_new_nand_12): Likewise.
27313         (sync_sub<mode>): Likewise.
27314         (sync_old_add<mode>): Likewise.
27315         (sync_old_sub<mode>): Likewise.
27316         (sync_new_add<mode>): Likewise.
27317         (sync_new_sub<mode>): Likewise.
27318         (sync_<optab><mode>): Likewise.
27319         (sync_old_<optab><mode>): Likewise.
27320         (sync_new_<optab><mode>): Likewise.
27321         (sync_nand<mode>): Likewise.
27322         (sync_old_nand<mode>): Likewise.
27323         (sync_new_nand<mode>): Likewise.
27324         (sync_lock_test_and_set<mode>): Likewise.
27325         (test_and_set_12): Likewise.
27326         (atomic_compare_and_swap<mode>): Likewise.
27327         (atomic_exchange<mode>_llsc): Likewise.
27328         (atomic_fetch_add<mode>_llsc): Likewise.
27329         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
27330         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
27331         (umips_save_restore_pattern_p): Likewise.
27332         (umips_load_store_pair_p): Likewise.
27333         (umips_output_load_store_pair): Likewise.
27334         (umips_movep_target_p): Likewise.
27335         (umips_12bit_offset_address_p): Likewise.
27336         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
27337         (mips_base_mips16): Rename this...
27338         (mips_base_compression_flags): ...to this. Update all uses.
27339         (mips_attribute_table): Add micromips, nomicromips and nocompression.
27340         (mips_mips16_decl_p): Delete.
27341         (mips_nomips16_decl_p): Delete.
27342         (mips_get_compress_on_flags): New function.
27343         (mips_get_compress_off_flags): New function.
27344         (mips_get_compress_mode): New function.
27345         (mips_get_compress_on_name): New function.
27346         (mips_get_compress_off_name): New function.
27347         (mips_insert_attributes): Support multiple compression types.
27348         (mips_merge_decl_attributes): Likewise.
27349         (umips_12bit_offset_address_p): New function.
27350         (mips_start_function_definition): Emit .set micromips directive.
27351         (mips_call_may_need_jalx_p): New function.
27352         (mips_function_ok_for_sibcall): Add microMIPS support.
27353         (mips_print_operand_punctuation): Support short delay slots and
27354         compact jumps.
27355         (umips_swm_mask, umips_swm_encoding): New.
27356         (umips_build_save_restore): New function.
27357         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
27358         (was_mips16_p): Remove.
27359         (old_compression_mode): New.
27360         (mips_set_compression_mode): New function.
27361         (mips_set_current_function): Add microMIPS support.
27362         (mips_option_override): Likewise.
27363         (umips_save_restore_pattern_p): New function.
27364         (umips_output_save_restore): New function.
27365         (umips_load_store_pair_p_1): New function.
27366         (umips_load_store_pair_p): New function.
27367         (umips_output_load_store_pair_1): New function.
27368         (umips_output_load_store_pair): New function.
27369         (umips_movep_target_p) New function.
27370         (mips_prepare_pch_save): Add microMIPS support.
27371         * config/mips/mips.h (TARGET_COMPRESSION): New.
27372         (TARGET_CPU_CPP_BUILTINS): Update macro
27373         to use new compression flags and to support microMIPS.
27374         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
27375         (MIPS_ARCH_FLOAT_SPEC): Likewise.
27376         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
27377         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
27378         (ASM_SPEC): Support mmicromips and mno-micromips.
27379         (M16STORE_REG_P): New macro.
27380         (MIPS_CALL): Support TARGET_MICROMIPS.
27381         (MICROMIPS_J): New macro.
27382         (mips_base_mips16): Rename this...
27383         (mips_base_compression_flags): ...to this.
27384         (UMIPS_12BIT_OFFSET_P): New macro.
27385         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
27386         (MULTILIB_DIRNAMES): Likewise.
27388 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
27390         PR target/52555
27391         * target-globals.c (save_target_globals): For init_reg_sets and
27392         target_reinit remporarily set this_fn_optabs to this_target_optabs.
27394 2013-02-22  James Greenhalgh  <james.greenhalgh@arm.com>
27396         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
27397         * config/aarch64/t-aarch64
27398         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
27400 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
27402         PR inline-asm/56148
27403         * lra-constraints.c (process_alt_operands): Reload operand
27404         conflicting with earlier clobber only if no more other conflicting
27405         operands.
27407 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
27409         PR sanitizer/56393
27410         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
27411         if not linking a shared library.
27413 2013-02-22  Seth LaForge  <sethml@google.com>
27415         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
27417 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
27419         * config/arm/arm.md (split for extendsidi): Update condition.
27420         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
27421         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
27422         (qhs_zextenddi_cstr): Likewise.
27424 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
27426         PR middle-end/56420
27427         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
27428         avoid signed wrapping.
27429         (expand_mult): Handle properly multiplication by
27430         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
27431         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
27432         in the compiler if coeff is HOST_WIDE_INT_MIN.
27433         (expand_divmod): Don't make ext_op1 static, change it's type to
27434         uhwi.  Avoid undefined behavior in -INTVAL (op1).
27436         PR rtl-optimization/50339
27437         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
27438         field.
27439         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
27440         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
27441         into splitting_ashiftrt field.
27442         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
27443         ASHIFTRT.
27444         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
27445         choices.
27447 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
27449         PR middle-end/56108
27450         * trans-mem.c (execute_tm_mark): Do not expand transactions that
27451         are sure to go irrevocable.
27453 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
27455         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
27456         scalars are valid operands.
27458 2013-02-21  Martin Jambor  <mjambor@suse.cz>
27460         PR tree-optimization/56310
27461         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
27462         only matching indices and non-negative final offsets.
27463         (intersect_aggregates_with_edge): Pass src_idx to
27464         agg_replacements_to_vector.  Pass src_idx insstead of index to
27465         intersect_with_agg_replacements.
27467 2013-02-21  Martin Jambor  <mjambor@suse.cz>
27469         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
27470         instead of hard-wired defaults.
27472 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
27474         * doc/invoke.texi (MIPS Options): Update documentation of the
27475         floating-point multiply-accumulate instruction restrictions.
27477 2013-02-21  Kostya Serebryany  <kcc@google.com>
27479         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
27480         asan_shadow_offset on x86_64 linux.
27482 2013-02-21  Richard Biener  <rguenther@suse.de>
27484         PR tree-optimization/56415
27485         Revert
27486         2013-02-11  Richard Biener  <rguenther@suse.de>
27488         PR tree-optimization/56273
27489         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
27490         first VRP run.
27492 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
27494         PR bootstrap/56258
27495         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
27496         instead of @itemx.
27498         PR inline-asm/56405
27499         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
27500         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
27502 2013-02-20  Jan Hubicka  <jh@suse.cz>
27504         PR tree-optimization/56265
27505         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
27506         when target is referenced for first time.
27508 2013-02-20  Richard Biener  <rguenther@suse.de>
27510         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
27511         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
27512         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
27513         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
27514         not return anything.
27515         (rename_ssa_copies): Do not remove unused locals.
27516         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
27517         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
27518         * passes.c (execute_function_todo): Do not schedule unused locals
27519         removal if cleanup_tree_cfg did something.
27520         * tree-ssa-live.c (remove_unused_locals): Dump statistics
27521         about the number of removed locals.
27523 2013-02-20  Richard Biener  <rguenther@suse.de>
27525         PR tree-optimization/56398
27526         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
27528 2013-02-20  Martin Jambor  <mjambor@suse.cz>
27530         PR tree-optimization/55334
27531         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
27532         restricted pointers to arrays.
27534 2013-02-20  Richard Biener  <rguenther@suse.de>
27535             Jakub Jelinek  <jakub@redhat.com>
27537         PR tree-optimization/56396
27538         * tree-ssa-ccp.c (n_const_val): New static variable.
27539         (get_value): Return NULL for SSA names we don't have a lattice
27540         entry for.
27541         (ccp_initialize): Initialize n_const_val.
27542         * tree-ssa-copy.c (n_copy_of): New static variable.
27543         (init_copy_prop): Initialize n_copy_of.
27544         (get_value): Return NULL_TREE for SSA names we don't have a
27545         lattice entry for.
27547 2013-02-20  Martin Jambor  <mjambor@suse.cz>
27549         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
27551 2013-02-20  Richard Biener  <rguenther@suse.de>
27553         * genpreds.c (write_lookup_constraint): Do not compare first
27554         letter of the constraint again.
27556 2013-02-20  Richard Biener  <rguenther@suse.de>
27558         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
27559         and ceil_log2.
27560         (get_use_iv_cost): Terminate hashtable walk when coming across
27561         an empty entry.
27563 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
27565         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
27566         reassociation for avx2 targets.
27568 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27570         * config/microblaze/microblaze.c: microblaze_has_clz = 0
27571         Add version check for v8.10.a to enable microblaze_has_clz
27572         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
27573         version and TARGET_PATTERN_COMPARE check
27574         * config/microblaze/microblaze.md: New clzsi2 instruction
27576 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27578         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
27579         function before branching.
27581 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
27583         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
27584         DUMP_INSN_RTX_UID.
27585         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
27587 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
27589         PR middle-end/55889
27590         * sel-sched.c: Include ira.h.
27591         (implicit_clobber_conflict_p): New function.
27592         (moveup_expr): Use it.
27593         * Makefile.in (sel-sched.o): Depend on ira.h.
27595 2013-02-19  Richard Biener  <rguenther@suse.de>
27597         PR tree-optimization/56384
27598         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
27599         (vn_hash_type): Split out from ...
27600         (vn_hash_constant_with_type): ... here.
27601         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
27602         (vn_phi_eq): Compare types from vn_phi_s structure.
27603         (vn_phi_lookup): Populate vn_phi_s type.
27604         (vn_phi_insert): Likewise.
27606 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
27608         PR tree-optimization/56350
27609         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
27610         if haven't found reduction or nested cycle operand, rather than
27611         asserting we must find it.
27613         PR tree-optimization/56381
27614         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
27615         to fold_build3.
27617 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
27618             Jakub Jelinek  <jakub@redhat.com>
27620         PR target/52555
27621         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
27622         (swap_optab_enable): Same.
27623         (init_all_optabs): Use argument instead of global.
27624         * tree.h (struct tree_optimization_option): New field target_optabs.
27625         * expr.h (init_all_optabs): Add argument to prototype.
27626         (TREE_OPTIMIZATION_OPTABS): New.
27627         (save_optabs_if_changed): Protoize.
27628         * optabs.h: Declare this_fn_optabs.
27629         * optabs.c (save_optabs_if_changed): New.
27630         Declare this_fn_optabs.
27631         (init_optabs): Add argument to init_all_optabs() call.
27632         * function.c (invoke_set_current_function_hook): Handle per
27633         function optabs.
27634         * function.h (struct function): New field optabs.
27635         * config/mips/mips.c (mips_set_mips16_mode): Handle when
27636         optimization_current_node has changed.
27637         * target-globals.h (save_target_globals_default_opts): Protoize.
27638         * target-globals.c (save_target_globals_default_opts): New.
27640 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27642         PR target/56347
27643         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
27644         registers %fr12 and %fr12R as call used.
27646         PR target/56214
27647         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
27648         and HImode, require all displacements to be an integer multiple of
27649         their mode size.
27650         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
27651         only allow QImode and HImode when reload is in progress and strict is
27652         true.  Likewise for symbolic addresses.  Use base14_operand to check
27653         displacements in REG+BASE addresses.
27655 2013-02-18  Richard Biener  <rguenther@suse.de>
27657         PR tree-optimization/56366
27658         * tree-vect-loop.c (get_initial_def_for_induction): Properly
27659         handle sign-conversion of outer-loop initial induction value.
27661 2013-02-18  Richard Biener  <rguenther@suse.de>
27663         PR middle-end/56349
27664         * cfghooks.c (merge_blocks): If we merge a latch into another
27665         block adjust references to it.
27666         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
27667         (verify_loop_structure): Verify that a recorded latch is in fact
27668         a latch.
27670 2013-02-18  Richard Biener  <rguenther@suse.de>
27672         PR tree-optimization/56321
27673         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
27674         order SSA name release and virtual operand unlinking.
27676 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27678         * config/microblaze/microblaze.md (save_stack_block): Define.
27679         (restore_stack_block): Likewise.
27681 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27683         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
27684         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
27685         * config/microblaze/microblaze.c (microblaze_option_override):
27686         Bail out early for PIC modes when target does not support PIC.
27688 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27690         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
27691         Replace with a microblaze version.
27692         (microblaze_trampoline_init): Adapt for microblaze.
27693         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
27694         microblaze.
27696 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
27697             Dodji Seketeli  <dodji@redhat.com>
27699         PR asan/56330
27700         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
27701         (instrument_mem_region_access): Do not forget to always put
27702         instrumentation of the of 'base' and 'base + len' in a "if (len !=
27703         0) statement, even for cases where either 'base' or 'base + len'
27704         are not instrumented -- because they have been previously
27705         instrumented.  Simplify the logic by putting all the statements
27706         instrument 'base + len' inside a sequence, and then insert that
27707         sequence right before the current insertion point.  Then, to
27708         instrument 'base + len', just get an iterator on that statement.
27709         And do not forget to update the pointer to iterator the function
27710         received as argument.
27712 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
27714         PR rtl-optimization/56348
27715         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
27717 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
27719         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
27720         (clean_graph_dump_file): Pass base to start_graph_dump.
27722 2013-02-14  Richard Henderson  <rth@redhat.com>
27724         PR target/55941
27725         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
27727 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
27729         * collect2-aix.h: Define F_LOADONLY.
27731 2013-02-14  Richard Biener  <rguenther@suse.de>
27733         PR lto/50494
27734         * varasm.c (output_constant_def_1): Get the decl representing
27735         the constant as argument.
27736         (output_constant_def): Wrap output_constant_def_1.
27737         (make_decl_rtl): Use output_constant_def_1 with the decl
27738         representing the constant.
27739         (build_constant_desc): Optionally re-use a decl already
27740         representing the constant.
27741         (tree_output_constant_def): Adjust.
27743 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
27745         Fix an asan crash
27746         * asan.c (instrument_builtin_call):  Really put the length of the
27747         second source argument into src1_len.
27749 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
27751         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
27752         argument.  If it is false, don't create edge from then_bb to
27753         fallthru_bb.
27754         (insert_if_then_before_iter): Pass true to it.
27755         (build_check_stmt): Pass false to it.
27756         (transform_statements): Flush hash table only on extended basic
27757         block boundaries, rather than at the beginning of every bb.
27758         Don't flush hash table on nonfreeing_call_p calls.
27759         * tree-flow.h (nonfreeing_call_p): New prototype.
27760         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
27762 2013-02-13  David S. Miller  <davem@davemloft.net>
27764         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
27766 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
27768         PR target/56184
27769         * ira.c (max_regno_before_ira): Move from ...
27770         (ira): ... here.
27771         (fix_reg_equiv_init): Use max_regno_before_ira instead of
27772         vec_safe_length.
27774 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
27776         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
27778 2013-02-13  Richard Biener  <rguenther@suse.de>
27780         PR lto/56295
27781         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
27782         globals in MEM_REFs.
27784 2013-02-13  Richard Biener  <rguenther@suse.de>
27786         * loop-init.c (loop_optimizer_init): Clear loop state when
27787         re-initializing preserved loops.
27788         * loop-unswitch.c (unswitch_single_loop): Return whether
27789         we unswitched the loop.  Do not verify loop state here.
27790         (unswitch_loops): When we unswitched a loop discover new loops.
27792 2013-02-13  Kostya Serebryany  <kcc@google.com>
27794         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
27795         on x86_64 linux.
27796         * sanitizer.def: Rename __asan_init to __asan_init_v1.
27798 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
27800         Avoid instrumenting duplicated memory access in the same basic block
27801         * Makefile.in (asan.o): Add new dependency on hash-table.h
27802         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
27803         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
27804         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
27805         (free_mem_ref_resources, has_mem_ref_been_instrumented)
27806         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
27807         (get_mem_ref_of_assignment): New functions.
27808         (get_mem_refs_of_builtin_call): Extract from
27809         instrument_builtin_call and tweak a little bit to make it fit with
27810         the new signature.
27811         (instrument_builtin_call): Use the new
27812         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
27813         of is_gimple_builtin_call.
27814         (instrument_derefs, instrument_mem_region_access): Insert the
27815         instrumented memory reference into the hash table.
27816         (maybe_instrument_assignment): Renamed instrument_assignment into
27817         this, and change it to advance the iterator when instrumentation
27818         actually happened and return true in that case.  This makes it
27819         homogeneous with maybe_instrument_assignment, and thus give a
27820         chance to callers to be more 'regular'.
27821         (transform_statements): Clear the memory reference hash table
27822         whenever we enter a new BB, when we cross a function call, or when
27823         we are done transforming statements.  Use
27824         maybe_instrument_assignment instead of instrumentation.  No more
27825         need to special case maybe_instrument_assignment and advance the
27826         iterator after calling it; it's now handled just like
27827         maybe_instrument_call.  Update comment.
27829 2013-02-13  Richard Biener  <rguenther@suse.de>
27831         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
27832         Fix loop discovery code.
27834 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
27836         PR inline-asm/56148
27837         * lra-constraints.c (process_alt_operands): Match early clobber
27838         operand with itself.  Check conflicts with earlyclobber only if
27839         the operand is not reloaded.  Prefer to reload conflicting operand
27840         if earlyclobber and matching operands are the same.
27842 2013-02-12  Richard Biener  <rguenther@suse.de>
27844         PR lto/56297
27845         * lto-streamer-out.c (write_symbol): Do not output symbols
27846         for hard register variables.
27848 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
27850         PR target/54222
27851         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
27852         (umulsidi3_insn, mulsidi3_insn): New insns.
27854 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
27856         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
27857         (struct tune_params): Add vec_costs field.
27858         * config/arm/arm.c (arm_builtin_vectorization_cost)
27859         (arm_add_stmt_cost): New functions.
27860         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
27861         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
27862         (arm_default_vec_cost): New struct of type cpu_vec_costs.
27863         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
27864         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
27865         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
27866         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
27868 2013-02-12  Richard Biener  <rguenther@suse.de>
27870         PR lto/56295
27871         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
27872         decls again if possible.
27874 2013-02-12  Richard Biener  <rguenther@suse.de>
27876         PR middle-end/56288
27877         * tree-ssa.c (verify_ssa_name): Fix check, move
27878         SSA_NAME_IN_FREE_LIST check up.
27880 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
27881             Steven Bosscher   <steven@gcc.gnu.org>
27883         PR rtl-optimization/56151
27884         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
27885         equal to op0 or op1, and last_insn pattern is CODE operation
27886         with MEM dest and one of the operands matches that MEM.
27888 2013-02-11  Sriraman Tallam  <tmsriram@google.com>
27890         * doc/extend.texi: Document Function Multiversioning and "default"
27891         parameter string to target attribute.
27892         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
27893         target attribute parameter is "default".
27894         (ix86_compare_version_priority): Remove checks for target attribute.
27895         (ix86_mangle_function_version_assembler_name): Change error to sorry.
27896         Remove check for target attribute equal to NULL. Add assert.
27897         (ix86_generate_version_dispatcher_body): Change error to sorry.
27899 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
27900             Jack Howarth  <howarth@bromo.med.uc.edu>
27901             Patrick Marlier  <patrick.marlier@gmail.com>
27903         PR libitm/55693
27904         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
27905         define ENDFILE_SPEC as TM_DESTRUCTOR.
27906         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
27908 2013-02-11  Alexander Potapenko  <glider@google.com>
27909             Jack Howarth  <howarth@bromo.med.uc.edu>
27910             Jakub Jelinek  <jakub@redhat.com>
27912         PR sanitizer/55617
27913         * config/darwin.c (cdtor_record): Rename ctor_record.
27914         (sort_cdtor_records): Rename sort_ctor_records.
27915         (finalize_dtors): New routine to sort destructors by
27916         priority before use in assemble_integer.
27917         (machopic_asm_out_destructor): Use finalize_dtors if needed.
27919 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
27921         PR rtl-optimization/56275
27922         * simplify-rtx.c (avoid_constant_pool_reference): Check that
27923         offset is non-negative and less than cmode size before
27924         calling simplify_subreg.
27926 2013-02-11  Richard Biener  <rguenther@suse.de>
27928         PR tree-optimization/56264
27929         * cfgloop.h (fix_loop_structure): Adjust prototype.
27930         * loop-init.c (fix_loop_structure): Return the number of
27931         newly discovered loops.
27932         * tree-cfgcleanup.c (repair_loop_structures): When new loops
27933         are discovered, do a full loop-closed SSA rewrite.
27935 2013-02-11  Richard Biener  <rguenther@suse.de>
27937         PR tree-optimization/56273
27938         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
27939         first VRP run.
27940         (check_array_ref): Fix missing newline in dumps.
27941         (search_for_addr_array): Likewise.
27943 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
27945         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
27947 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
27949         PR target/56256
27950         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
27952 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
27954         PR rtl-optimization/56246
27955         * lra-constraints.c (simplify_operand_subreg): Try to reuse
27956         reload pseudo.
27957         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
27958         constraints are satisfied.
27960 2013-02-08  Jeff Law  <law@redhat.com>
27962         PR debug/53948
27963         * emit-rtl.c (reg_is_parm_p): New function.
27964         * regs.h (reg_is_parm_p): New prototype.
27965         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
27966         callee-clobbered registers.
27968 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
27970         PR target/56043
27971         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
27972         If there is no implicit builtin declaration, just return NULL.
27974 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
27976         * config/i386/sse.md (FMAMODEM): New mode iterator.
27977         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
27978         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
27980 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
27982         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
27983         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
27984         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
27986 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27988         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
27989         (microblaze*-*-elf): Likewise.
27990         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
27991         LINK_SPEC.
27992         * config/microblaze/microblaze-c.c: Add builtin defines for
27993         _LITTLE_ENDIAN and _BIG_ENDIAN.
27994         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
27995         add to TARGET_DEFAULT flags.
27996         Expand ASM_SPEC and LINK_SPEC.
27997         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
27998         * config/microblaze/microblaze.md: Update extendsidi2 and
27999         movdi_internal instructions to use low-order / high-order reg
28000         print_operands.
28001         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
28002         options and inversemask / mask of LITTLE_ENDIAN.
28003         * config/microblaze/t-microblaze: Expand multilib options to
28004         include mlittle-endian (le) and update exceptions patterns.
28006 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
28008         PR rtl-optimization/56195
28009         * lra-constraints.c (get_reload_reg): Don't reuse regs
28010         if they have smaller mode than requested, if they have
28011         wider mode than requested, try to return a SUBREG.
28013         PR tree-optimization/56250
28014         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
28015         if type is unsigned and code isn't MULT_EXPR.
28017 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
28019         PR tree-optimization/56064
28020         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
28021         bits according to mode.
28022         * fixed-value.h (fixed_from_double_int)
28023         (const_fixed_from_double_int): Adjust comments.
28025 2013-02-08  Richard Biener  <rguenther@suse.de>
28027         PR lto/56231
28028         * lto-streamer.h (struct data_in): Remove current_file, current_line
28029         and current_col members.
28030         * lto-streamer-out.c (lto_output_location): Stream changed bits
28031         en-block for efficiency.
28032         * lto-streamer-in.c (clear_line_info): Remove.
28033         (lto_input_location): Cache current file, line and column
28034         globally via local statics.  Read changed bits en-block.
28035         (input_function): Do not call clear_line_info.
28036         (lto_read_body): Likewise.
28037         (lto_input_toplevel_asms): Likewise.
28039 2013-02-08  Michael Matz  <matz@suse.de>
28041         PR tree-optimization/52448
28042         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
28043         (nt_call_phase): New static.
28044         (add_or_mark_expr): Only mark accesses with newer phase than any
28045         call seen.
28046         (nonfreeing_call_p): New.
28047         (nt_init_block): Update nt_call_phase, mark blocks as visited.
28048         (nt_fini_block): Keep blocks marked as visited.
28049         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
28051 2013-02-08  Richard Biener  <rguenther@suse.de>
28053         * ira.c (ira): Free broken dominator information.
28055 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
28057         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
28059 2013-02-08  Marek Polacek  <polacek@redhat.com>
28061         * cfgloop.c (verify_loop_structure): Add more checking of headers.
28063 2013-02-08  Richard Biener  <rguenther@suse.de>
28065         PR middle-end/56181
28066         * cfgloop.h (flow_loops_find): Adjust.
28067         (bb_loop_header_p): Declare.
28068         * cfgloop.c (bb_loop_header_p): New function split out from ...
28069         (flow_loops_find): ... here.  Adjust function signature,
28070         support incremental loop structure update.
28071         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
28072         * cfgloopmanip.c (fix_loop_structure): Move ...
28073         * loop-init.c (fix_loop_structure): ... here.
28074         (apply_loop_flags): Split out from ...
28075         (loop_optimizer_init): ... here.
28076         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
28077         in incremental mode, only remove dead loops here.
28079 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
28081         PR target/54222
28082         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
28083         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
28084         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
28085         (*round<mode>3.libgcc): New insns for fixed-modes.
28086         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
28087         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
28088         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
28089         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
28090         implementations.  Define to __builtin_avr_absFX,
28091         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
28092         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
28093         __builtin_avr_countlsFX, respectively.
28094         * config/avr/avr-c.c (target.h): Include it.
28095         (enum avr_builtin_id): New enum.
28096         (avr_resolve_overloaded_builtin): New static function.
28097         (avr_register_target_pragmas): Use it to set
28098         targetm.resolve_overloaded_builtin.
28099         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
28100         tree nodes used by DEF_BUILTIN.
28101         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
28102         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
28103         <AVR_BUILTIN_xxBITS>: Same.
28105 2013-02-08  Richard Biener  <rguenther@suse.de>
28107         * cfgloop.c (verify_loop_structure): Properly handle
28108         a loop exiting to another loop header.
28109         * ira-int.h (ira_loops): Remove.
28110         * ira.c (ira_loops): Remove.
28111         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
28112         (do_reload): Use loop_optimizer_finalize.
28113         * ira-build.c (create_loop_tree_nodes): Use get_loops and
28114         number_of_loops to access the loop tree.
28115         (more_one_region_p): Likewise.
28116         (finish_loop_tree_nodes): Likewise.
28117         (rebuild_regno_allocno_maps): Likewise.
28118         (mark_loops_for_removal): Likewise.
28119         (mark_all_loops_for_removal): Likewise.
28120         (remove_unnecessary_regions): Likewise.
28121         (ira_build): Likewise.
28122         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
28124 2013-02-08  Richard Biener  <rguenther@suse.de>
28126         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
28127         * ipa-pure-const.c (analyze_function): Avoid calling
28128         mark_irreducible_loops twice.
28129         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
28131 2013-02-07  David S. Miller  <davem@davemloft.net>
28133         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
28134         on 'reg'.
28135         * var-tracking.c (vt_add_function_parameter): Test the presence of
28136         HAVE_window_save properly and do not remap argument registers when
28137         we have a leaf function.
28139 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
28141         PR bootstrap/56227
28142         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
28143         instead of "ll".
28144         * config/i386/i386.c (ix86_print_operand): Ditto.
28146 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
28148         * lra-constraints.c (process_alt_operands): Fix recently added comment.
28150 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
28152         PR rtl-optimization/56225
28153         * lra-constraints.c (process_alt_operands): Check that reload hard
28154         reg can hold value for strict_low_part.
28156 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
28158         PR debug/56154
28159         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
28160         dwarf2out_end_function.
28161         (in_first_function_p, maybe_at_text_label_p,
28162         first_loclabel_num_not_at_text_label): New variables.
28163         (dwarf2out_var_location): In the first function find out
28164         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
28165         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
28166         functions.
28168 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
28170         PR rtl-optimization/56178
28171         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
28172         SUBREG of a register.  Tidy up related block of code.
28173         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
28174         note if the source is a register or a SUBREG of a register.
28176 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
28178         PR target/56228
28179         * config/rs6000/rs6000.md (ptrm): New mode attr.
28180         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
28181         call_value_indirect_aix<pttrsize>,
28182         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
28183         m in constraints.
28185 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
28187         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
28188         if -bnortl. Convert to strcmp and strncmp.
28190 2013-02-07  Alan Modra  <amodra@gmail.com>
28192         PR target/54009
28193         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
28194         addresses won't wrap when offsetting.
28195         (rs6000_secondary_reload): Provide secondary reloads needed for
28196         wrapping LO_SUM addresses.
28198 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
28200         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
28201         MACH, just __MACH__.
28203 2013-02-06  Richard Biener  <rguenther@suse.de>
28205         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
28206         instead of calling fix_loop_structure.
28208 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
28210         PR middle-end/56217
28211         * omp-low.c (use_pointer_for_field): Return false if
28212         lower_send_shared_vars doesn't generate any copy-out code.
28214 2013-02-06  Tom de Vries  <tom@codesourcery.com>
28216         PR rtl-optimization/56131
28217         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
28218         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
28219         of the label is NULL.  Add comment.
28221 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
28223         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
28225         PR sanitizer/55374
28226         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
28227         (STATIC_LIBTSAN_LIBS): Likewise.
28228         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
28229         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
28230         is defined, don't add anything else beyond that.
28231         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
28232         (LINK_COMMAND_SPEC): Use them.
28234         PR tree-optimization/56205
28235         * tree-stdarg.c (check_all_va_list_escapes): Return true if
28236         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
28237         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
28239 2013-02-05  Richard Biener  <rguenther@suse.de>
28241         PR tree-optimization/53342
28242         PR tree-optimization/53185
28243         * tree-vectorizer.h (vect_check_strided_load): Remove.
28244         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
28245         not disallow peeling for vectorized strided loads.
28246         (vect_check_strided_load): Make static and simplify.
28247         (vect_analyze_data_refs): Adjust.
28248         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
28249         correctly when vectorizing strided loads.
28251 2013-02-05  Richard Biener  <rguenther@suse.de>
28253         * doc/install.texi: Refer to ISL, not PPL.
28255 2013-02-05  Jan Hubicka  <jh@suse.cz>
28257         PR tree-optimization/55789
28258         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
28260 2013-02-05  Jan Hubicka  <jh@suse.cz>
28262         PR tree-optimization/55789
28263         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
28264         the dead call anyway.
28266 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
28268         PR sanitizer/55374
28269         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
28271 2013-02-04  Alexander Potapenko  <glider@google.com>
28272             Jack Howarth  <howarth@bromo.med.uc.edu>
28273             Jakub Jelinek  <jakub@redhat.com>
28275         PR sanitizer/55617
28276         * config/darwin.c (sort_ctor_records): Stabilized qsort
28277         on constructor priority by using original position.
28278         (finalize_ctors): New routine to sort constructors by
28279         priority before use in assemble_integer.
28280         (machopic_asm_out_constructor): Use finalize_ctors if needed.
28282 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
28284         PR libstdc++/54314
28285         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
28286         about visibility on artificial decls.
28287         * config/sol2.c (solaris_assemble_visibility): Likewise.
28289 2013-02-04  Kai Tietz  <ktietz@redhat.com>
28291         PR target/56186
28292         * config/i386/i386.c (function_value_ms_64): Add additional valtype
28293         argument and improve checking of return-argument types for 16-byte
28294         modes.
28295         (ix86_function_value_1): Add additional valtype argument on call
28296         of function_value_64.
28297         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
28298         handling infunction_value_64 function.
28300 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
28302         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
28304 2013-02-04  Richard Biener  <rguenther@suse.de>
28306         PR tree-optimization/56188
28307         * tree-ssa-structalias.c (label_visit): Consider case with
28308         initially non-empty points-to set.
28309         (perform_var_substitution): Dump node mapping and clean up.
28311 2013-02-04  Richard Guenther  <rguenther@suse.de>
28313         PR lto/56168
28314         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
28315         node prevail as last resort.
28316         (lto_symtab_merge_decls): Remove guard on LTRANS here.
28317         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
28319 2013-02-04  Richard Biener  <rguenther@suse.de>
28321         PR tree-optimization/56113
28322         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
28323         Merge into ...
28324         (equiv_class_lookup_or_add): ... this.
28325         (label_visit): Adjust and fix error in previous patch.
28326         (perform_var_substitution): Adjust.
28328 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
28330         * config/sh/divtab.c: Fix formatting and comments throughout the file.
28331         * config/sh/sh4-300.md: Likewise.
28332         * config/sh/sh4a.md: Likewise.
28333         * config/sh/constraints.md: Likewise.
28334         * config/sh/sh.md: Likewise.
28335         * config/sh/netbsd-elf.h: Likewise.
28336         * config/sh/predicates.md: Likewise.
28337         * config/sh/sh-protos.h: Likewise.
28338         * config/sh/ushmedia.h: Likewise.
28339         * config/sh/linux.h: Likewise.
28340         * config/sh/sh.c: Likewise.
28341         * config/sh/superh.h: Likewise.
28342         * config/sh/elf.h: Likewise.
28343         * config/sh/sh4.md: Likewise.
28344         * config/sh/sh.h: Likewise.
28346 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
28348         * config/pa/constraints.md: Adjust unused letters.  Change "T"
28349         constraint to match_test floating_point_store_memory_operand().
28350         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
28351         (base14_operand): New.
28352         (floating_point_store_memory_operand): New.
28353         (integer_store_memory_operand): Revise to use base14_operand and
28354         reg_plus_base_memory_operand.
28355         (move_dest_operand): Allow symbolic_memory_operands.
28356         (symbolic_memory_operand): Check for LO_SOM.
28357         (symbolic_operand): Change default case to break.
28358         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
28359         CONST_DOUBLE values to be reloaded by putting them into memory when
28360         the destination is a floating point register.
28361         (movdf): Remove code to handle CONST_DOUBLE.
28362         (movsf): Likewise.
28363         (reload_indf_r1): New.
28364         (reload_insf_r1): New.
28365         Consistently use "Q" and "T" constraints with integer and floating
28366         point move instructions, respectively.
28367         (movdi): Remove FAIL.
28368         Change predicate for source operand unamed DImode move from
28369         general_operand to move_src_operand.
28370         (umulsidi3): Change predicate for destination operand to
28371         register_operand.
28372         Likewise for similar unamed patterns.
28373         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
28374         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
28375         (hppa_legitimize_address): Simplify mask calculation.
28376         (pa_emit_move_sequence): Revised handling of secondary reloads from
28377         REG+D addresses for floating point loads and stores.  Directly handle
28378         loading CONST0_RTX (mode) to a floating point register.
28379         (pa_secondary_reload): Handle reloading DF and SFmode constant values
28380         to floating point registers.  Don't restrict secondary reloads to
28381         floating point registers to integer modes.  Revise some comments and
28382         cleanup some code.
28383         (TARGET_LEGITIMATE_ADDRESS_P): Define.
28384         (pa_legitimate_address_p): New.
28385         (pa_legitimize_reload_address): New.
28386         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
28387         (STRICT_REG_OK_FOR_BASE_P): New.
28388         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
28389         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
28391 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
28392             Andrew Dixie  <andrewd@gentrack.com>
28394         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
28395         flag set.
28397 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
28399         * expmed.c (extract_bit_field_1): Pass the full width of the
28400         structure to get_best_reg_extraction_insn.
28402 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
28404         PR target/54601
28405         * configure.ac (use_cxa_atexit): Add AIX.
28406         * configure: Regenerate.
28408         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
28410 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
28412         PR debug/54793
28413         * final.c (need_profile_function): New variable.
28414         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
28415         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
28416         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
28417         notes, targetm.asm_out.function_prologue doesn't emit anything,
28418         HAVE_prologue and profiler should be emitted before prologue,
28419         set need_profile_function instead of emitting it.
28420         (final_scan_insn): If need_profile_function, emit
28421         profile_function on the first NOTE_INSN_BASIC_BLOCK or
28422         NOTE_INSN_FUNCTION_BEG note.
28424 2013-02-01  Richard Henderson  <rth@redhat.com>
28426         * config/rs6000/rs6000.md (smulditi3): New.
28427         (umulditi3): New.
28429         * config/alpha/alpha.md (umulditi3): New.
28431 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
28433         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
28434         (ASM_OUTPUT_ALIGNED_LOCAL): New.
28436 2013-02-01  Richard Biener  <rguenther@suse.de>
28438         PR tree-optimization/56113
28439         * tree-ssa-structalias.c (label_visit): Reduce work for
28440         single-predecessor nodes.
28442 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
28444         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
28445         range isn't testing for zero.
28447 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
28449         PR middle-end/56113
28450         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
28452 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
28453             Nick Clifton  <nickc@redhat.com>
28455         * config/v850/constraints.md (Q): Define as a memory constraint.
28456         * config/v850/predicates.md (label_ref_operand): New predicate.
28457         (e3v5_shift_operand): New predicate.
28458         (ior_operator): New predicate.
28459         * config/v850/t-v850: Add e3v5 multilib.
28460         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
28461         (v850_gen_movdi): Prototype.
28462         * config/v850/v850.c: Add support for e3v5 architecture.
28463         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
28464         TARGET_V850E_UP.
28465         (construct_save_jarl): Add e3v5 long JARL support.
28466         (v850_adjust_insn_length): New function.  Adjust length of call
28467         insns when using e3v5 instructions.
28468         (v850_gen_movdi): New function: Generate instructions to move a
28469         DImode value.
28470         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
28471         (CPP_SPEC): Define __v850e3v5__ as appropriate.
28472         (TARGET_USE_FPU): Enable for e3v5.
28473         (CONST_OK_FOR_W): New macro.
28474         (ADJUST_INSN_LENGTH): Define.
28475         * config/v850/v850.md (UNSPEC_LOOP): Define.
28476         (attr cpu): Add v850e3v5.
28477         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
28478         (movdi): New pattern.
28479         (movdi_internal): New pattern.
28480         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
28481         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
28482         (cstoresf4): Likewise.
28483         (cstoredf4): Likewise.
28484         (insv): New pattern.
28485         (rotlso3_a): New pattern.
28486         (rotlsi3_b): New pattern
28487         (rotlsi3_v850e3v5): New pattern.
28488         (doloop_begin): New pattern.
28489         (fix_loop_counter): New pattern.
28490         (doloop_end): New pattern.
28491         (branch_normal): Add e3v5 long branch support.
28492         (branch_invert): Likewise.
28493         (branch_z_normal): Likewise.
28494         (branch_z_invert): Likewise.
28495         (branch_nz_normal): Likewise.
28496         (branch_nz_invert): Likewise.
28497         (call_internal_short): Add e3v5 register-indirect JARL support.
28498         (call_internal_long): Likewise.
28499         (call_value_internal_short): Likewise.
28500         (call_value_internal_long): Likewise.
28501         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
28502         (mloop): New option.
28503         * config.gcc: Add support for configuring v840e3v5 target.
28504         * doc/invoke.texi: Document new v850 specific command line options.
28506 2013-01-31  Paul Koning  <ni1d@arrl.net>
28508         PR debug/55059
28509         PR debug/54508
28510         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
28511         children if parent is a class.
28512         (prune_unused_types_prune): Don't add DW_AT_declaration.
28514 2013-01-31  Richard Biener  <rguenther@suse.de>
28516         PR tree-optimization/56157
28517         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
28518         match up operand with SLP child.
28520 2013-01-31  Jason Merrill  <jason@redhat.com>
28522         PR debug/54410
28523         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
28524         parameters the first time.
28525         (gen_scheduled_generic_parms_dies): Check completeness here.
28527 2013-01-31  Richard Biener  <rguenther@suse.de>
28529         PR middle-end/53073
28530         * common.opt (faggressive-loop-optimizations): New flag,
28531         enabled by default.
28532         * doc/invoke.texi (faggressive-loop-optimizations): Document.
28533         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
28534         infer_loop_bounds_from_undefined by it.
28536 2013-01-31  Richard Biener  <rguenther@suse.de>
28538         PR tree-optimization/56150
28539         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
28540         visit virtual operands.
28541         (find_uses_to_rename_bb): Likewise.
28543 2013-01-31  Richard Biener  <rguenther@suse.de>
28545         PR tree-optimization/56150
28546         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
28547         mixed store non-store stmts.
28549 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
28551         PR sanitizer/55374
28552         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
28553         LIBASAN_EARLY_SPEC is defined.
28554         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
28555         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
28556         before %o.
28557         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
28559         PR c++/55742
28560         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
28561         invalid args instead of ICEing on it.
28562         (ix86_valid_target_attribute_tree): Return error_mark_node if
28563         ix86_valid_target_attribute_inner_p failed.
28564         (ix86_valid_target_attribute_p): Return false only if
28565         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
28566         target("default") attribute.
28567         (sorted_attr_string): Change argument from const char * to tree,
28568         merge in all target attribute arguments rather than just one.
28569         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
28570         instead of free.  Avoid using strcat.
28571         (ix86_mangle_function_version_assembler_name): Mangle
28572         target("default") as if no target attribute is present.  Adjust
28573         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
28574         instead of xmalloc and XDELETEVEC instead of free.
28575         (ix86_function_versions): Don't return true if one of the decls
28576         doesn't have target attribute.  If they don't and one of the decls
28577         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
28578         sorted_attr_string caller.  Use XDELETEVEC instead of free.
28579         (ix86_supports_function_versions): Remove.
28580         (make_name): Fix up formatting.
28581         (make_dispatcher_decl): Remove resolver_name and its initialization.
28582         Avoid leaking memory.
28583         (is_function_default_version): Return true if there is
28584         target("default") attribute rather than no target attribute at all.
28585         (make_resolver_func): Avoid leaking memory.
28586         (ix86_generate_version_dispatcher_body): Likewise.
28587         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
28588         * target.def (supports_function_versions): Remove.
28589         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
28590         * doc/tm.texi: Regenerated.
28592 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
28594         PR rtl-optimization/56144
28595         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
28596         for values with side effects.
28598 2013-01-30  Richard Biener  <rguenther@suse.de>
28600         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
28601         (sparseset_pop): Likewise.
28602         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
28603         to be able to use quick_push in the worker loop.
28605 2013-01-30  Marek Polacek  <polacek@redhat.com>
28607         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
28609 2013-01-30  Richard Biener  <rguenther@suse.de>
28611         PR lto/56147
28612         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
28614 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
28616         PR tree-optimization/56064
28617         * fixed-value.c (fixed_from_double_int): New function.
28618         * fixed-value.h (fixed_from_double_int): New prototype.
28619         (const_fixed_from_double_int): New static inline function.
28620         * fold-const.c (native_interpret_fixed): New static function.
28621         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
28622         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
28623         (native_encode_fixed): New static function.
28624         (native_encode_expr) <FIXED_CST>: Use it.
28625         (native_interpret_int): Move double_int worker code to...
28626         * double-int.c (double_int::from_buffer): ...this new static method.
28627         * double-int.h (double_int::from_buffer): Prototype it.
28629 2013-01-30  Richard Biener  <rguenther@suse.de>
28631         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
28632         New pointer-map and obstack.
28633         (init_alias_vars): Allocate pointer-map and obstack.
28634         (delete_points_to_sets): Free them.
28635         (find_what_var_points_to): Cache result.
28636         (find_what_p_points_to): Adjust for changed interface of
28637         find_what_var_points_to.
28638         (compute_points_to_sets): Likewise.
28639         (ipa_pta_execute): Likewise.
28641 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
28643         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
28644         * configure: Regenerate.
28645         * config.in: Regenerate.
28646         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
28647         #nobits/#progbits if supported.
28649 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
28651         PR target/56121
28652         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
28653         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
28654         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
28656 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
28658         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
28659         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
28661 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
28663         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
28664         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
28666 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
28668         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
28669         declaration.
28670         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
28671         * config/arm/cortex-a7.md: New bypasses using
28672         arm_mac_accumulator_is_result.
28674 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
28676         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
28677         (cortex_a7_neon_mla): Likewise.
28678         (cortex_a7_fpfmad): New reservation.
28679         (cortex_a7_fpmacs): Use ffmas and update required units.
28680         (cortex_a7_fpmuld): Update required units and latency.
28681         (cortex_a7_fpmacd): Likewise.
28682         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
28683         (cortex_a7_neon). Likewise.
28684         (bypass) Update participating units.
28686 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
28688         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
28689         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
28690         from fmac to ffma.
28691         * config/arm/vfp11.md (vfp_farith): Use ffmas.
28692         (vfp_fmul): Use ffmad.
28693         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
28694         (cortex_r4_fmacd): Use ffmad.
28695         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
28696         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
28697         (cortex_a9_fmacd): Use ffmad.
28698         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
28699         (cortex_a8_vfp_macd): Use ffmad.
28700         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
28701         (cortex_a5_fpmacd): Use ffmad.
28702         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
28703         (cortex_a15_vfp_macd): Use ffmad.
28704         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
28706 2013-01-29  Jason Merrill  <jason@redhat.com>
28708         PR libstdc++/54314
28709         * varasm.c (default_assemble_visibility): Don't warn about
28710         visibility on artificial decls.
28712 2013-01-29  Richard Biener  <rguenther@suse.de>
28714         PR tree-optimization/56113
28715         * tree-ssa-structalias.c (equiv_class_lookup): Also return
28716         the bitmap leader.
28717         (label_visit): Free duplicate bitmaps and record the leader instead.
28718         (perform_var_substitution): Adjust.
28720 2013-01-29  Richard Biener  <rguenther@suse.de>
28722         PR tree-optimization/55270
28723         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
28724         the CFG, schedule loops for fixup.
28726 2013-01-29  Nick Clifton  <nickc@redhat.com>
28728         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
28729         SP_REG.
28731 2013-01-28  Leif Ekblad  <leif@rdos.net>
28733         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
28734         * config/i386/i386.h (TARGET_RDOS): New macro.
28735         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
28736         * config/i386/i386.c (ix86_option_override_internal): For 64bit
28737         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
28738         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
28739         DEFAULT_LARGE_SECTION_THRESHOLD.
28740         * config/i386/i386.md (R14_REG, R15_REG): New constants.
28741         * config/i386/rdos.h: New file.
28742         * config/i386/rdos64.h: New file.
28744 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
28746         PR other/54814
28747         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
28748         TEST_HARD_REG_BIT.
28750 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
28752         PR rtl-optimization/56117
28753         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
28754         call cselib_lookup_from_insn on the MEM before calling
28755         add_insn_mem_dependence.
28757 2013-01-28  Richard Biener  <rguenther@suse.de>
28759         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
28760         to a stmt that didn't have one.
28761         (copy_phis_for_bb): Likewise for PHI arguments.
28762         (copy_debug_stmt): Likewise for debug stmts.
28764 2013-01-28  Richard Biener  <rguenther@suse.de>
28766         PR tree-optimization/56034
28767         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
28768         (partition_builtin_p): Adjust.
28769         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
28770         it is the last partition.
28771         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
28772         up the vertex for the definition.
28773         (classify_partition): Classify whether a partition is a
28774         PKIND_REDUCTION, thus has uses outside of the loop.
28775         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
28776         Merge all PKIND_REDUCTION partitions into the last partition.
28777         (tree_loop_distribution): Seed partitions from reductions as well.
28779 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
28781         PR tree-optimization/56125
28782         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
28783         pow(x,c) into sqrt(x) * powi(x, n/2) or
28784         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
28785         optimizing for size.
28786         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
28787         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
28788         integer.
28790         PR tree-optimization/56094
28791         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
28792         to UNKNOWN_LOCATION while gimplifying expr.
28794 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
28796         PR target/56114
28797         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
28798         operand 0 in movabs insn template for -masm=intel asm alternative.
28799         (*movabs<mode>_2): Ditto for operand 1.
28801 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
28803         PR target/54663
28804         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
28805         of microblaze-c.o
28807 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
28809         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
28810         tm_file.
28812 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
28814         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
28815         Undef to avoid warning.
28817 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
28819         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
28820         * configure: Regenerate.
28822 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
28824         PR tree-optimization/56098
28825         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
28826         for stmts with volatile ops.
28827         (cond_store_replacement): Don't optimize if assign has volatile ops.
28828         (cond_if_else_store_replacement_1): Don't optimize if either
28829         then_assign or else_assign have volatile ops.
28830         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
28831         volatile ops.
28833 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
28835         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
28837 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
28839         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
28840         missing ':' in asm example.
28842 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
28844         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
28845         entries into lane and laneq entries.
28846         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
28847         Remove AdvSIMD scalar modes.
28848         (aarch64_sq<r>dmulh_laneq<mode>): New.
28849         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
28850         modes.
28851         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
28852         builtin implementations to relfect changes in RTL in aarch64-simd.md.
28853         * config/aarch64/iterators.md (VCOND): New.
28854         (VCONQ): New.
28856 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
28858         PR target/54222
28859         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
28860         Add NULL LIBNAME argument to existing definitions.
28861         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
28862         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
28863         * config/avr/avr.c (DEF_BUILTIN): Same.
28864         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
28865         (avr_expand_builtin): Expand to a vanilla call if a libgcc
28866         implementation is available (DECL_ASSEMBLER_NAME is set).
28867         (avr_fold_absfx): New static function.
28868         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
28869         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
28870         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
28871         AVR_BUILTIN_ABSLLK.
28872         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
28873         (abshk, absk, abslk, absllk): Provide as static inline functions.
28875 2013-01-25  Marek Polacek  <polacek@redhat.com>
28877         PR tree-optimization/56035
28878         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
28880 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
28882         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
28883         (*movtf_internal_rex64): Add (!o,C) alternative
28884         (*movxf_internal_rex64): Ditto.
28885         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
28887 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
28889         * doc/invoke.texi: fix typo.
28890         * doc/objc.texi: fix typo.
28892 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
28894         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
28895         for the first two alternatives.
28897 2013-01-24  Diego Novillo  <dnovillo@google.com>
28899         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
28900         (ggc-zone.o): Remove.
28901         * configure.ac: Remove option --with-gc.
28902         * configure: Re-generate.
28903         * doc/install.texi: Remove documentation for --with-gc.
28904         * gengtype.c (write_enum_defn): Remove.  Update all users.
28905         (write_Types_process_field): Remove generation of gt_e_* argument.
28906         (output_type_enum): Remove.  Update all users.
28907         (write_enum_defn): Remove.  Update all users.
28908         (enum alloc_zone): Remove.  Update all users.
28909         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
28910         * ggc-common.c (ggc_splay_alloc): Remove first argument.
28911         Update all callers.
28912         (struct ptr_data): Remove field TYPE.  Update all users.
28913         (gt_pch_note_object): Remove argument TYPE.  Update all users.
28914         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
28915         Update all users.
28916         * ggc-none.c (ggc_alloc_typed_stat): Remove.
28917         (struct alloc_zone): Remove.
28918         (ggc_internal_alloc_zone_stat): Remove.
28919         (ggc_internal_cleared_alloc_zone_stat): Remove.
28920         * ggc-page.c (ggc_alloc_typed_stat): Remove.
28921         (ggc_pch_count_object): Remove last argument.  Update all users.
28922         (ggc_pch_alloc_object): Remove last argument.  Update all users.
28923         (struct alloc_zone): Remove.
28924         * ggc-zone.c: Remove.
28925         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
28926         (struct alloc_zone): Remove.
28927         (ggc_alloc_typed_stat): Remove.
28928         (ggc_alloc_typed): Remove.
28929         (ggc_splay_alloc): Remove first argument.
28930         (rtl_zone): Remove.  Update all users.
28931         (tree_zone): Remove.  Update all users.
28932         (tree_id_zone): Remove.  Update all users.
28933         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
28934         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
28935         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
28936         * tree-ssanames.c: Remove references to zone allocator in comments.
28938 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
28940         * config/avr/avr.c (avr_out_fract): Make register numbers that
28941         might be outside of source operand signed.
28943 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
28945         * config/i386/constraints.md (Yf): New constraint.
28946         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
28947         of f constraint to conditionaly disable x87 register preferences.
28948         (*movdf_internal): Ditto.
28949         (*movsf_internal): Ditto.
28951 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
28953         PR inline-asm/55934
28954         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
28955         that have operands with impossible constraints.
28956         Add a FIXME for a speed-up opportunity.
28957         * lra-constraints.c (process_alt_operands): Verify that a class
28958         selected from constraints on asms is valid for the operand mode.
28959         (curr_insn_transform): Remove incorrect comment.
28961 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
28963         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
28964         TOC operand is a valid symbol ref in the constant pool.
28966 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
28968         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
28970 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
28972         PR target/54222
28973         * config/avr/stdfix.h: New file.
28974         * t-avr (stdfix-gcc.h): New rule to build it.
28975         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
28977 2013-01-23  Kostya Serebryany  <kcc@google.com>
28979         * config/darwin.h: remove dependency on
28980         CoreFoundation (asan on Mac OS).
28982 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
28984         PR target/49069
28985         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
28986         instead of cmpdi_operand for first comparison operand.
28987         Don't assert that comparison operands aren't both constants.
28989 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
28991         * doc/install.texi (Downloading the Source): Update references to
28992         downloading separate components.
28994 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
28996         * doc/extend.texi (__int128): Improve grammar.
28998 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
29000         PR target/56028
29001         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
29002         alternative to (o,r).
29003         (*movdi_internal_rex64): Remove (!o,n) alternative.
29004         (DImode immediate->memory splitter): Remove.
29005         (DImode immediate->memory peephole2): Remove.
29006         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
29007         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
29008         alternative to (!o,*r).
29009         (*movtf_internal_sse): New pattern.
29010         (*movxf_internal_rex64): New pattern.
29011         (*movxf_internal): Disable for TARGET_64BIT.
29012         (*movdf_internal_rex64): Remove (!o,F) alternative.
29014 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
29016         PR middle-end/56074
29017         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
29018         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
29019         * tree-vect-loop-manip.c (find_loop_location): Also ignore
29020         stmt locations where LOCATION_LOCUS of the stmt location is
29021         UNKNOWN_LOCATION or BUILTINS_LOCATION.
29023         PR target/55686
29024         * config/i386/i386.md (UNSPEC_STOS): New.
29025         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
29026         *strsetqi_1): Add UNSPEC_STOS.
29028 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
29030         PR c++/56067
29031         * doc/invoke.texi: Remove left over -Wsynth example.
29033 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
29035         PR tree-optimization/56051
29036         * fold-const.c (fold_binary_loc): Don't fold
29037         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
29038         a narrowing conversion, or widening conversion from signed
29039         to unsigned.
29041 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
29043         PR rtl-optimization/56023
29044         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
29045         dependent on debug instruction.
29047 2013-01-21  Martin Jambor  <mjambor@suse.cz>
29049         PR middle-end/56022
29050         * function.c (allocate_struct_function): Call
29051         invoke_set_current_function_hook earlier.
29053 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
29055         * reload1.c (init_reload): Only initialize reload_obstack
29056         during the first call.
29058 2013-01-21  Marek Polacek  <polacek@redhat.com>
29060         * cfgloop.c (verify_loop_structure): Fix up grammar.
29062 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
29064         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
29065         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
29067 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
29069         PR target/56058
29070         * config/arm/marvell-pj4.md: Update copyright year.
29071         Fix up use of alu to alu_reg and simple_alu_imm.
29073 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
29075         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
29077 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
29079         PR target/55433
29080         * lra-constraints.c (curr_insn_transform): Don't reuse original
29081         insn for secondary memory move when memory mode should be different.
29083 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
29085         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
29086         atomic_storedi_1): New patterns.
29088 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
29090         btver2 pipeline descriptions.
29091         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
29092         descriptions.
29093         * config/i386/i386.md (btver2_decode): New type attributes.
29094         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
29095         type attributes.
29096         * config/i386/btver2.md: New file describing btver2 pipelines.
29098 2013-01-19  Andrew Pinski  <apinski@cavium.com>
29100         PR tree-optimization/52631
29101         * tree-ssa-sccvn (visit_use): Before looking up the original
29102         statement, try looking up the simplified expression.
29104 2013-01-19  Anthony Green  <green@moxielogic.com>
29106         * config/moxie/moxie.c (moxie_expand_prologue): Set
29107         current_function_static_stack_size.
29109 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
29111         PR tree-optimization/56029
29112         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
29113         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
29115 2013-01-18  Sharad Singhai  <singhai@google.com>
29117         PR tree-optimization/55995
29118         * dumpfile.c (dump_loc): Print location only if available.
29119         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
29121 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
29123         PR target/55433
29124         * lra-constraints.c (curr_insn_transform): Reuse original insn for
29125         secondary memory move.
29126         (inherit_reload_reg): Use rclass instead of cl for
29127         check_secondary_memory_needed_p.
29129 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
29131         PR middle-end/56015
29132         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
29133         the case where writing real complex part of target modifies op1.
29135 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
29137         * config/aarch64/aarch64-simd.md
29138         (aarch64_vcond_internal<mode>): Handle unordered cases.
29139         * config/aarch64/iterators.md (v_cmp_result): New.
29141 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
29142             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
29144         * config/arm/marvell-pj4.md: New file.
29145         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
29146         * config/arm/arm.md (generic_sched): Add marvell_pj4.
29147         (generic_vfp): Likewise.
29148         * config/arm/arm-cores.def: Add marvell-pj4.
29149         * config/arm/arm-tune.md: Regenerate.
29150         * config/arm/arm-tables.opt: Regenerate.
29151         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
29152         * doc/invoke.texi: Document marvell-pj4.
29154 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
29156         * config/aarch64/arm_neon.h: Map scalar types to standard types.
29158 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
29160         PR debug/54114
29161         PR debug/54402
29162         PR debug/49888
29163         * var-tracking.c (negative_power_of_two_p): New.
29164         (global_get_addr_cache, local_get_addr_cache): New.
29165         (get_addr_from_global_cache, get_addr_from_local_cache): New.
29166         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
29167         heading comment.
29168         (vt_stack_offset_p): Remove.
29169         (vt_canon_true_dep): Always canonicalize loc's address.
29170         (clobber_overlapping_mems): Make sure we have a MEM.
29171         (local_get_addr_clear_given_value): New.
29172         (val_reset): Clear local cached entries.
29173         (compute_bb_dataflow): Create and release the local cache.
29174         Disable duplicate MEMs clobbering.
29175         (emit_notes_in_bb): Clobber MEMs likewise.
29176         (vt_emit_notes): Create and release the local cache.
29177         (vt_initialize, vt_finalize): Create and release the global
29178         cache, respectively.
29179         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
29181 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
29183         PR libmudflap/53359
29184         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
29185         not found in the symtab.
29187 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
29189         PR debug/56006
29190         PR rtl-optimization/55547
29191         PR rtl-optimization/53827
29192         PR debug/53671
29193         PR debug/49888
29194         * alias.c (offset_overlap_p): New, factored out of...
29195         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
29196         the conservative special case for symbolic constants.  Don't
29197         adjust zero sizes on alignment.
29199 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
29201         PR rtl-optimization/52573
29202         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
29203         REG_UNUSED for the same register.
29205 2013-01-17  Richard Biener  <rguenther@suse.de>
29206             Marek Polacek  <polacek@redhat.com>
29208         PR rtl-optimization/55833
29209         * loop-unswitch.c (unswitch_loops): Move loop verification...
29210         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
29211         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
29212         Set it to true when we're removing a loop from hierarchy tree in
29213         an irreducible region.
29214         (fix_bb_placements): Adjust caller.
29215         (fix_loop_placements): Likewise.
29217 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
29219         * config/avr/builtins.def (DEF_BUILTIN): Factor out
29220         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
29221         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
29222         Remove ID.  Adjust comments.
29223         * config/avr/avr-c.c (avr_builtin_name): Remove.
29224         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
29225         * config/avr/avr.c (avr_tolower): New static function.
29226         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
29227         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
29228         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
29229         default expansion.
29231 2013-01-17  Jan Hubicka  <jh@suse.cz>
29233         PR tree-optimization/55273
29234         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
29236 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
29238         PR target/55981
29239         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
29240         store through atomic_store<mode>_1.
29241         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
29243 2013-01-17  Martin Jambor  <mjambor@suse.cz>
29245         PR tree-optimizations/55264
29246         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
29247         for virtual methods.
29248         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
29249         virtual methods before inlining is over.
29250         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
29251         virtual functions.
29252         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
29253         non-virtual.
29255 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
29257         PR rtl-optimization/56005
29258         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
29259         pending reads for prefetch.
29261 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
29263         * config/aarch64/aarch64.md
29264         (*cstoresi_neg_uxtw): New pattern.
29265         (*cmovsi_insn_uxtw): New pattern.
29266         (*<optab>si3_uxtw): New pattern.
29267         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
29268         (*<optab>si3_insn_uxtw): New pattern.
29269         (*bswapsi2_uxtw): New pattern.
29271 2013-01-16  Richard Biener  <rguenther@suse.de>
29273         * tree-inline.c (tree_function_versioning): Remove set but
29274         never used variable.
29276 2013-01-16  Richard Biener  <rguenther@suse.de>
29278         PR tree-optimization/55964
29279         * tree-flow.h (rename_variables_in_loop): Remove.
29280         (rename_variables_in_bb): Likewise.
29281         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
29282         (copy_loop_before): Adjust and delete update-ssa status.
29283         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
29284         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
29285         (rename_variables_in_loop): Remove.
29286         (slpeel_update_phis_for_duplicate_loop): Likewise.
29287         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
29288         use available cfg machinery instead of duplicating it.
29289         Update PHI nodes and perform poor-mans SSA update here.
29290         (slpeel_tree_peel_loop_to_edge): Adjust.
29292 2013-01-16  Richard Biener  <rguenther@suse.de>
29294         PR tree-optimization/54767
29295         PR tree-optimization/53465
29296         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
29297         (vrp_visit_phi_node): For PHI arguments coming via backedges
29298         drop all symbolical range information.
29299         (execute_vrp): Compute backedges.
29301 2013-01-16  Richard Biener  <rguenther@suse.de>
29303         * doc/install.texi: Update CLooG and ISL requirements to
29304         0.18.0 and 0.11.1.
29306 2013-01-16  Christian Bruel  <christian.bruel@st.com>
29308         PR target/55301
29309         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
29310         (broken_move): Handle UNSPECV_SP_SWITCH_B.
29311         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
29313 2013-01-16  DJ Delorie  <dj@redhat.com>
29315         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
29316         (UNSPECV_SP_SWITCH_E): New.
29317         (sp_switch_1): Change to an unspec.
29318         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
29319         replace $r15.
29321 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
29323         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
29324         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
29325         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
29326         (expand_mem_thread_fence): Ditto.
29327         (expand_mem_signal_fence): Ditto.
29328         (expand_atomic_load): Ditto.
29329         (expand_atomic_store): Ditto.
29331 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
29333         PR rtl-optimization/55547
29334         PR rtl-optimization/53827
29335         PR debug/53671
29336         PR debug/49888
29337         * alias.c (memrefs_conflict_p): Set sizes to negative after
29338         AND adjustments.
29340 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
29342         PR target/55940
29343         * function.c (thread_prologue_and_epilogue_insns): Always
29344         add crtl->drap_reg to set_up_by_prologue.set, even if
29345         stack_realign_drap is false.
29347 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
29349         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
29350         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
29351         *call): Fix indention.
29353 2013-01-15  Tom de Vries  <tom@codesourcery.com>
29355         PR target/55876
29356         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
29357         Update comment.
29359 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
29361         PR rtl-optimization/55153
29362         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
29364 2013-01-15  Martin Jambor  <mjambor@suse.cz>
29366         PR tree-optimization/55920
29367         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
29368         accesses as grp_to_be_debug_replaced.
29370 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
29372         PR tree-optimization/55920
29373         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
29374         there is non-useless type conversion needed from debug rhs to lhs,
29375         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
29377 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
29378             Mikael Pettersson  <mikpe@it.uu.se>
29380         PR target/43961
29381         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
29382         Thumb.
29383         (ASM_OUTPUT_CASE_LABEL): Remove.
29384         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
29385         * final.c (shorten_branches): Update alignment of labels before
29386         jump tables if CASE_VECTOR_SHORTEN_MODE.
29388 2013-01-15  Richard Biener  <rguenther@suse.de>
29390         PR bootstrap/55961
29391         * system.h: Do not include gmp.h for building host tools.
29393 2013-01-15  Richard Biener  <rguenther@suse.de>
29395         PR middle-end/55882
29396         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
29397         account for bitpos when computing alignment.
29399 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
29401         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
29402         (ix86_target_macros_internal): Likewise.
29404         * config/i386/i386.c (m_CORE2I7): Removed.
29405         (m_CORE_HASWELL): New macro.
29406         (m_CORE_ALL): Likewise.
29407         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
29408         (initial_ix86_arch_features): Likewise.
29409         (processor_target_table): Initializations for Core avx2.
29410         (cpu_names): New names "core-avx2".
29411         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
29412         PROCESSOR_CORE_HASWELL.
29413         (ix86_issue_rate): New case.
29414         (ia32_multipass_dfa_lookahead): Likewise.
29415         (ix86_sched_init_global): Likewise.
29417         * config/i386/i386.h (TARGET_HASWELL): New macro.
29418         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
29419         (processor_type): New PROCESSOR_HASWELL.
29421 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
29423         PR tree-optimization/55955
29424         * tree-vect-loop.c (vectorizable_reduction): Give up early on
29425         *SHIFT_EXPR and *ROTATE_EXPR codes.
29427         PR tree-optimization/48766
29428         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
29429         -ftrapv disable -fwrapv.
29431 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
29433         PR target/55974
29434         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
29435         etc. to 1 and not to __flash.
29436         Use LL suffix for __INT24_MAX__ with -mint8.
29437         Use ULL suffix for __UINT24_MAX__ with -mint8.
29439 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
29441         * config/avr/avr-arch.h
29442         (struct base_arch_s): Use typedef avr_arch_t instead.
29443         (struct arch_info_s): Use typedef avr_arch_info_t instead.
29444         (struct mcu_type_s): Use typedef avr_mcu_t instead.
29445         * config/avr/avr.c: Same.
29446         * config/avr/avr-devices.c: Same.
29447         * config/avr/driver-avr.c: Same.
29448         * config/avr/gen-avr-mmcu-texi.c: Same.
29449         * config/avr/avr-mcus.def: Adjust comment.
29451 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
29453         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
29454         * config/aarch64/iterators.md (VALLDI): New.
29456 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
29457             Andi Kleen  <ak@linux.intel.com>
29459         PR target/55948
29460         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
29461         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
29462         memmodel flag.
29464 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
29466         * config/avr/avr-stdint.h: Remove trailing blanks.
29467         * config/avr/avr-log.h: Same.
29468         * config/avr/avr-arch.h: Same.
29469         * config/avr/avr-devices.c: Same.
29470         * config/avr/avr-dimode.md: Same.
29471         * config/avr/predicates.md: Same.
29472         * config/avr/avr-c.c: Same.  And fix typo.
29474         * config/avr/avr-protos.h: Same.  And:
29475         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
29476         (init_cumulative_args): Rename to avr_init_cumulative_args.
29477         (expand_prologue): Rename to avr_expand_prologue.
29478         (expand_epilogue): Rename to avr_expand_epilogue.
29479         (adjust_insn_length): Rename to avr_adjust_insn_length.
29480         (notice_update_cc): Rename to avr_notice_update_cc.
29481         (final_prescan_insn): Rename to avr_final_prescan_insn.
29482         * config/avr/avr.c: Same.
29483         * config/avr/avr.h: Same.
29484         * config/avr/avr.md: Remove trailing blanks.
29485         (prologue): Use avr_expand_prologue.
29486         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
29488 2013-01-14  Richard Biener  <rguenther@suse.de>
29490         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
29491         verify_location, collect_subblocks): New functions.
29492         (verify_gimple_in_cfg): Verify that locations only reference
29493         BLOCKs in the functions BLOCK tree.
29495 2013-01-14  Richard Biener  <rguenther@suse.de>
29497         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
29498         PHI argument.
29499         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
29500         unshare reference.
29501         (insert_out_of_ssa_copy_on_edge): Likewise.
29502         (rewrite_close_phi_out_of_ssa): Likewise.
29503         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
29504         debug expressions.
29505         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
29506         propagated constants.
29507         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
29508         can not be shared.
29510 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
29512         * config/avr/avr-modes.def: Add GPL copyright notice.
29514 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
29516         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
29517         MEMMODEL_MASK to determine memory model.
29518         (atomic_store<mode>): Ditto from operands[2].
29519         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
29521 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
29523         PR fortran/55935
29524         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
29525         (fold_gimple_assign): Don't call unshare_expr here.
29526         (fold_ctor_reference): Call unshare_expr.
29528 2013-01-13  Terry Guo  <terry.guo@arm.com>
29530         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
29531         * doc/fragments.texi: Document MULTILIB_REUSE.
29532         * gcc.c (multilib_reuse): New internal spec.
29533         (set_multilib_dir): Also search multilib from multilib_reuse.
29534         * genmultilib (tmpmultilib3): Refactor code.
29535         (tmpmultilib4): Ditto.
29536         (multilib_reuse): New multilib argument.
29538 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
29540         * Makefile.in: Update copyright.
29542 2013-01-12  Tom de Vries  <tom@codesourcery.com>
29544         PR middle-end/55890
29545         * calls.c (expand_call): Check if arg_nr is valid.
29547 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
29549         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
29550         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
29551         documentation.  Add missing '__' in front of
29552         __builtin_ia32_packssdw256.
29554 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
29556         PR target/55719
29557         * config/s390/s390.c (s390_preferred_reload_class): Do not return
29558         NO_REGS for larl operands.
29559         (s390_reload_larl_operand): Use s390_load_address instead of
29560         emit_move_insn.
29562 2013-01-11  Richard Biener  <rguenther@suse.de>
29564         * tree-cfg.c (verify_node_sharing_1): Split out from ...
29565         (verify_node_sharing): ... here.
29566         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
29568 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
29570         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
29571         Substitute TREECHECKING.
29572         * configure: Regenerate.
29573         * Makefile.in (TREECHECKING): New.
29575 2013-01-11  Richard Guenther  <rguenther@suse.de>
29577         PR tree-optimization/44061
29578         * tree-vrp.c (extract_range_basic): Compute zero as
29579         value-range for __builtin_constant_p of function parameters.
29581 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
29583         Update copyright years.
29585 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
29587         PR rtl-optimization/55672
29588         * lra-eliminations.c (mark_not_eliminable): Permit addition with
29589         const to be eliminable.
29591 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
29593         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
29594         * configure: Regenerate.
29596 2013-01-10  Richard Biener  <rguenther@suse.de>
29598         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
29600 2013-01-10  Richard Biener  <rguenther@suse.de>
29602         PR bootstrap/55792
29603         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
29604         locations for virtual PHI arguments.
29605         (rewrite_update_phi_arguments): Likewise.
29607 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
29609         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
29610         on to assembler.
29612 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
29614         PR tree-optimization/55921
29615         * tree-complex.c (expand_complex_asm): New function.
29616         (expand_complex_operations_1): Call it for GIMPLE_ASM.
29618 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
29620         PR target/55718
29621         * config/s390/s390.c (s390_symref_operand_p)
29622         (s390_loadrelative_operand_p): Merge the two functions.
29623         (s390_check_qrst_address, print_operand_address): Add parameters
29624         to s390_loadrelative_operand_p invokation.
29625         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
29626         (s390_reload_larl_operand, s390_secondary_reload): Use
29627         s390_loadrelative_operand_p instead of s390_symref_operand_p.
29628         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
29630 2013-01-09  Mike Stump  <mikestump@comcast.net>
29632         * dse.c (record_store): Remove unnecessary assert.
29634 2013-01-09  Jan Hubicka  <jh@suse.cz>
29636         PR tree-optimization/55569
29637         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
29638         * cfgloop.h (scale_loop_profile): Likewise.
29640 2013-01-09  Jan Hubicka  <jh@suse.cz>
29642         PR lto/45375
29643         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
29644         functions.
29645         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
29647 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
29649         PR middle-end/55114
29650         * expr.h (maybe_emit_group_store): Declare.
29651         * expr.c (maybe_emit_group_store): New function.
29652         * builtins.c (expand_builtin_int_roundingfn): Call it.
29653         (expand_builtin_int_roundingfn_2): Likewise.
29655 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
29657         PR rtl-optimization/55829
29658         * lra-constraints.c (match_reload): Add code for absent output.
29659         (curr_insn_transform): Add code for reloads of matched inputs
29660         without output.
29662 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
29664         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
29665         attribute of movddup insn to DF.
29666         (*vec_interleave_lowv2df): Ditto.
29667         (vec_dupv2df): Ditto.
29669 2013-01-09  Jan Hubicka  <jh@suse.cz>
29671         PR tree-optimiation/55875
29672         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
29673         EVERY_ITERATION parameter.
29674         (number_of_iterations_exit): Check if exit is executed every iteration.
29675         (idx_infer_loop_bounds): Similarly here.
29676         (n_of_executions_at_most): Simplify
29677         to only test for cases where statement is dominated by the
29678         particular bound; handle correctly the "postdominance" test.
29679         (scev_probably_wraps_p): Use max loop iterations info
29680         as a global bound first.
29682 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
29683             Nick Clifton  <nickc@redhat.com>
29685         * config/v850/v850.md (cbranchsf4): New pattern.
29686         (cstoresf4): New pattern.
29687         (cbranchdf4): New pattern.
29688         (cstoredf4): New pattern.
29689         (movsicc): Disallow floating point comparisons.
29690         (cmpsf_le_insn): Fix order of operators.
29691         (cmpsf_lt_insn): Likewise.
29692         (cmpsf_eq_insn): Likewise.
29693         (cmpdf_le_insn): Likewise.
29694         (cmpdf_lt_insn): Likewise.
29695         (cmpdf_eq_insn): Likewise.
29696         (cmpsf_ge_insn): Use LE comparison.
29697         (cmpdf_ge_insn): Likewise.
29698         (cmpsf_gt_insn): Use LT comparison.
29699         (cmpdf_gt_insn): Likewise.
29700         (cmpsf_ne_insn): Delete pattern.
29701         (cmpdf_ne_insn): Delete pattern.
29702         * config/v850/v850.c (v850_gen_float_compare): Use
29703         gen_cmpdf_eq_insn for NE comparison.
29704         (v850_float_z_comparison_operator)
29705         (v850_float_nz_comparison_operator): Move from here ...
29706         * config/v850/predicates.md: ... to here.  Move GT and GE
29707         comparisons into v850_float_z_comparison_operator.
29708         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
29709         Delete prototype.
29710         (v850_float_nz_comparison_operator): Likewise.
29712 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
29714         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
29715         with calls to gen_insvsi/gen_insvdi.
29717 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
29719         * config/i386/i386.c (initial_ix86_tune_features): Set up
29720         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
29722 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
29723             Jakub Jelinek  <jakub@redhat.com>
29725         PR tree-optimization/48189
29726         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
29727         If nitercst is 0, don't predict the exit edge.
29729 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
29731         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
29732         in asm_fprintf with reg_names.
29733         (aarch64_print_operand_address): Likewise.
29734         (aarch64_return_addr): Likewise.
29735         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
29737 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
29739         * config/pa/pa.h (VAL_U6_BITS_P): Define.
29740         (INT_U6_BITS): Likewise.
29741         * config/pa/predicates.md (uint6_operand): New predicate.
29742         (shift5_operand, shift6_operand): Likewise.
29743         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
29744         arith32_operand.
29745         (lshrdi3): Use shift6_operand.
29746         (shrpsi4, shrpdi4): New insn patterns.
29747         (extzv): Delete expander.
29748         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
29749         predicates in unamed zero extract patterns.  Tighten common constraint.
29750         (extv): Delete expander.
29751         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
29752         predicates in unamed sign extract patterns.  Tighten common constraint.
29753         (insv): Delete expander.
29754         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
29755         predicates in unamed insert patterns.  Tighten common constraint.
29756         Change uint32_operand predicate to uint6_operand predicate in unamed
29757         DImode pattern to insert constant values of type 1...1xxxx.
29759 2013-01-04  Jan Hubicka  <jh@suse.cz>
29761         PR tree-optimization/55823
29762         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
29763         issue.
29765 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
29766             Uros Bizjak  <ubizjak@gmail.com>
29768         PR rtl-optimization/55845
29769         * df-problems.c (can_move_insns_across): Stop scanning at
29770         volatile_insn_p source instruction or give up if
29771         across_from .. across_to range contains any volatile_insn_p
29772         instructions.
29774 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
29776         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
29777         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
29778         Declare.
29779         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
29780         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
29782 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
29784         PR fortran/55341
29785         * asan.c (asan_clear_shadow): New function.
29786         (asan_emit_stack_protection): Use it.
29788 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
29790         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
29791         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
29792         with tab instead of space.
29794 2013-01-08  Nick Clifton  <nickc@redhat.com>
29796         * config/rl78/rl78.c (rl78_expand_prologue): Always select
29797         register bank 0 at the start of an interrupt handler.
29798         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
29799         MDBH registers.
29801 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
29803         * config/aarch64/aarch64-simd.md
29804         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
29805         (aarch64_simd_bsl): Likewise.
29806         (aarch64_vcond_internal<mode>): Likewise.
29807         (vcond<mode><mode>): Likewise.
29808         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
29809         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
29811 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
29813         * config/aarch64/aarch64-builtins.c
29814         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
29816 2013-01-08  Martin Jambor  <mjambor@suse.cz>
29818         PR debug/55579
29819         * tree-sra.c (analyze_access_subtree): Return true also after
29820         potentially creating a debug-only replacement.
29822 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
29824         PR middle-end/55890
29825         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
29827         PR tree-optimization/54120
29828         * tree-vrp.c (range_fits_type_p): Don't allow
29829         src_precision < precision from signed vr to unsigned_p
29830         if vr->min or vr->max is negative.
29831         (simplify_float_conversion_using_ranges): Test can_float_p
29832         against CODE_FOR_nothing.
29834 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
29835             Richard Biener  <rguenther@suse.de>
29837         PR middle-end/55851
29838         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
29839         types instead of just INTEGER_TYPE types.
29841 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
29843         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
29844         TF_SIZE): Define.
29846 2013-01-07  Steve Ellcey  <sellcey@mips.com>
29848         PR target/42661
29849         * config/mips/mips.opt: Change mad to mmad to match documentation.
29851 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
29853         PR target/55897
29854         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
29855         .progmemx.data now.
29857 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
29859         PR target/55897
29860         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
29861         (avr_addrspace_t): Add .section_name field.
29862         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
29863         array size.
29864         (avr_addrspace): Same.  Initialize .section_name.  Remove last
29865         NULL entry.  Put __memx into .progmemx.data.
29866         (progmem_section_prefix): Remove.
29867         (avr_asm_init_sections): No need to initialize progmem_section.
29868         (avr_asm_named_section): Use avr_addrspace[].section_name to get
29869         section name prefix.
29870         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
29871         retrieve the progmem section.
29872         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
29873         boundary to run over avr_addrspace[].
29874         (avr_register_target_pragmas): Ditto.
29876 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
29878         * varasm.c (output_constant_def_contents): For asan_protect_global
29879         protected strings, adjust DECL_ALIGN if needed, before testing for
29880         anchored symbols.
29881         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
29882         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
29883         normal decls.
29884         (output_object_block): For asan protected decls, emit asan padding
29885         after their contents.
29886         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
29887         (asan_finish_file): Test it here instead.
29889 2013-01-07  Nick Clifton  <nickc@redhat.com>
29890             Matthias Klose  <doko@debian.org>
29891             Doug Kwan  <dougkwan@google.com>
29892             H.J. Lu  <hongjiu.lu@intel.com>
29894         PR driver/55470
29895         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
29897         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
29899         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
29901         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
29903         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
29905 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
29907         PR target/54461
29908         * doc/install.texi (Cross-Compiler-Specific Options): Document
29909         --with-avrlibc.
29911 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
29913         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
29914         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
29915         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
29916         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
29917         vqmovun_high_s64): Fix source operand number and update copyright.
29919 2013-01-07  Richard Biener  <rguenther@suse.de>
29921         PR middle-end/55890
29922         * gimple.h (gimple_call_builtin_p): New overload.
29923         * gimple.c (validate_call): New function.
29924         (gimple_call_builtin_p): Likewise.
29925         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
29926         Use gimple_call_builtin_p.
29927         (find_func_clobbers): Likewise.
29928         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
29929         (strlen_optimize_stmt): Likewise.
29931 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
29933         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
29934         (vld1q_dup_*): Likewise.
29935         (vld1_*): Likewise.
29936         (vld1q_*): Likewise.
29937         (vld1_lane_*): Likewise.
29938         (vld1q_lane_*): Likewise.
29940 2013-01-07  Richard Biener  <rguenther@suse.de>
29942         * lto-streamer.h (LTO_minor_version): Bump to 2.
29944 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
29946         * config/aarch64/aarch64-protos.h
29947         (aarch64_const_double_zero_rtx_p): Rename to...
29948         (aarch64_float_const_zero_rtx_p): ...this.
29949         (aarch64_float_const_representable_p): New.
29950         (aarch64_output_simd_mov_immediate): Likewise.
29951         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
29952         move immediate case.
29953         * config/aarch64/aarch64.c
29954         (aarch64_const_double_zero_rtx_p): Rename to...
29955         (aarch64_float_const_zero_rtx_p): ...this.
29956         (aarch64_print_operand): Allow printing of new constants.
29957         (aarch64_valid_floating_const): New.
29958         (aarch64_legitimate_constant_p): Check for valid floating-point
29959         constants.
29960         (aarch64_simd_valid_immediate): Likewise.
29961         (aarch64_vect_float_const_representable_p): New.
29962         (aarch64_float_const_representable_p): Likewise.
29963         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
29964         (aarch64_output_simd_mov_immediate): New.
29965         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
29966         (*movdf_aarch64): Likewise.
29967         * config/aarch64/constraints.md (Ufc): New.
29968         (Y): call aarch64_float_const_zero_rtx.
29969         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
29971 2013-01-07  Richard Biener  <rguenther@suse.de>
29973         PR tree-optimization/55888
29974         PR tree-optimization/55862
29975         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
29976         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
29977         not if it is contained therein.
29979 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
29981         * config/avr/t-avr: Typo.
29983 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
29985         PR55243
29986         * config/avr/t-avr: Don't automatically rebuild
29987         $(srcdir)/config/avr/t-multilib
29988         $(srcdir)/config/avr/avr-tables.opt
29989         $(srcdir)/doc/avr-mmcu.texi
29990         (avr-mcus): New phony target to build them on request.
29991         (s-avr-mlib, s-avr-mmcu-texi): Remove.
29992         * avr/avr-mcus.def: Adjust comments.
29994 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
29996         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
29998 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
30000         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
30002 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
30004         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
30006 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
30008         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
30009         to generate profiling.
30010         * config/rs6000/aix64.h (LIB_SPEC): Same.
30012 2013-01-04  Andrew Pinski  <apinski@cavium.com>
30014         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
30015         New function.
30016         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
30018 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
30020         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
30021         unconditionally.
30022         (ix86_expand_move): Ditto.
30023         (ix86_zero_extend_to_Pmode): Ditto.
30024         (ix86_expand_call): Ditto.
30025         (ix86_expand_special_args_builtin): Ditto.
30026         (ix86_expand_builtin): Ditto.
30028 2013-01-04  Richard Biener  <rguenther@suse.de>
30030         PR tree-optimization/55862
30031         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
30032         translating them through PHI nodes.
30034 2013-01-04  Martin Jambor  <mjambor@suse.cz>
30036         PR tree-optimization/55755
30037         * tree-sra.c (sra_modify_assign): Do not check that an access has no
30038         children when trying to avoid producing a VIEW_CONVERT_EXPR.
30040 2013-01-04  Marek Polacek  <polacek@redhat.com>
30042         PR middle-end/55859
30043         * opts.c (default_options_optimization): Clarify error message.
30045 2013-01-04  Richard Biener  <rguenther@suse.de>
30047         PR middle-end/55863
30048         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
30049         reassociation.
30051 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
30053         PR target/53789
30054         * config/pa/pa.md (movsi): Revert previous change.
30055         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
30056         references.
30058 2013-01-03  Richard Henderson  <rth@redhat.com>
30060         * config/i386/i386.c (ix86_expand_move): Always assign to op1
30061         after eliminating TLS symbols.
30063 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
30065         PR bootstrap/50167
30066         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
30067         * graphite-poly.c (debug_gmp_value): Likewise.
30069 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
30071         PR target/55712
30072         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
30073         selected code model, define __code_mode_small__, __code_model_medium__,
30074         __code_model_large__, __code_model_32__ or __code_model_kernel__.
30075         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
30076         xchg temporary register with %k.  Declare temporary register as
30077         early clobbered.
30078         [__x86_64__]: For medium and large code models, preserve %rbx register.
30080 2013-01-03  Richard Biener  <rguenther@suse.de>
30082         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
30083         (dump_subscript): Adjust.
30084         (finalize_ddr_dependent): Do not dump redundant info.
30085         (analyze_siv_subscript): Adjust.
30086         (subscript_dependence_tester): Likewise.
30087         (compute_affine_dependence): Likewise.
30089 2013-01-03  Richard Biener  <rguenther@suse.de>
30091         Revert
30092         2013-01-03  Richard Biener  <rguenther@suse.de>
30094         PR tree-optimization/55857
30095         * tree-vect-stmts.c (vectorizable_load): Do not setup
30096         re-alignment for invariant loads.
30098         2013-01-02  Richard Biener  <rguenther@suse.de>
30100         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
30101         invariant load do not generate a vector load from the scalar location.
30103 2013-01-03  Richard Biener  <rguenther@suse.de>
30105         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
30106         for not vectorizing.
30107         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
30108         not build INDIRECT_REFs, call get_name once only.
30109         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
30110         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
30112 2013-01-03  Richard Biener  <rguenther@suse.de>
30114         PR tree-optimization/55857
30115         * tree-vect-stmts.c (vectorizable_load): Do not setup
30116         re-alignment for invariant loads.
30118 2013-01-03  Richard Biener  <rguenther@suse.de>
30120         PR lto/55848
30121         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
30122         prefer a built-in decl.
30124 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
30126         * gcc.c (process_command): Update copyright notice dates.
30127         * gcov.c (print_version): Likewise.
30128         * gcov-dump.c (print_version): Likewise.
30130         PR rtl-optimization/55838
30131         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
30132         iv0.step, iv1.step and step.
30134 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
30135             Marc Glisse  <marc.glisse@inria.fr>
30137         PR tree-optimization/55832
30138         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
30139         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
30140         integer_{one,zero}_node.
30142 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
30144         PR debug/54402
30145         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
30146         * var-tracking.c (reverse_op): Don't add reverse ops to
30147         VALUEs that have already
30148         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
30150 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
30152         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
30154 2013-01-02  Teresa Johnson  <tejohnson@google.com>
30156         * dumpfile.c (dump_loc): Print filename with location.
30157         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
30158         new location_t parameter to emit complete unroll message with
30159         new dump framework.
30160         (canonicalize_loop_induction_variables): Compute loops location
30161         and pass to try_unroll_loop_completely.
30162         * loop-unroll.c (report_unroll_peel): New function.
30163         (peel_loops_completely): Use new dump format with location
30164         for main dumpfile message, and invoke report_unroll_peel on success.
30165         (decide_unrolling_and_peeling): Ditto.
30166         (decide_peel_once_rolling): Remove old dumpfile message subsumed
30167         by report_unroll_peel.
30168         (decide_peel_completely): Ditto.
30169         (decide_unroll_constant_iterations): Ditto.
30170         (decide_unroll_runtime_iterations): Ditto.
30171         (decide_peel_simple): Ditto.
30172         (decide_unroll_stupid): Ditto.
30173         * cfgloop.c (get_loop_location): New function.
30174         * cfgloop.h (get_loop_location): Declare.
30176 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
30178         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
30179         NULL.
30181 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
30183         PR middle-end/55198
30184         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
30185         BLKmode objects when EXPAND_MEMORY is specified.
30187 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
30189         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
30190         in loop predicate.
30191         (fold_builtin_cpu): Do not share cpu model decls across statements.
30193 2013-01-02  Jason Merrill  <jason@redhat.com>
30195         PR c++/55804
30196         * tree.c (build_array_type_1): Revert earlier change.
30198 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
30200         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
30201         "cortex-a57".
30202         * config/aarch64/aarch64-tune.md: Re-generate.
30204 2013-01-02  Richard Biener  <rguenther@suse.de>
30206         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
30207         invariant load do not generate a vector load from the scalar location.
30209 2013-01-02  Richard Biener  <rguenther@suse.de>
30211         PR bootstrap/55784
30212         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
30213         * configure: Regenerate.
30215 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
30217         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
30218         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
30219         (expand_builtin_int_roundingfn_2): Keep the original target around
30220         for the fallback case.
30222 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
30224         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
30225         to be clear for sign changes.
30227 2013-01-01  Jan Hubicka  <jh@suse.cz>
30229         * ipa-inline-analysis.c: Fix formatting.
30231 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
30233         PR tree-optimization/55831
30234         * tree-vect-loop.c (get_initial_def_for_induction): Use
30235         gsi_after_labels instead of gsi_start_bb.
30237 Copyright (C) 2013 Free Software Foundation, Inc.
30239 Copying and distribution of this file, with or without modification,
30240 are permitted in any medium without royalty provided the copyright
30241 notice and this notice are preserved.