2013-09-17 Richard Biener <rguenther@suse.de>
[official-gcc.git] / gcc / ChangeLog
blob6c75db04de1e7382bddd98ee80d4789207418d5b
1 2013-09-17  Richard Biener  <rguenther@suse.de>
3         PR tree-optimization/58432
4         * tree-loop-distribution.c (tree_loop_distribution): Also
5         scan PHIs for outside loop uses and seed a partition from them.
7 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
9         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
10         (restructure_reference): Call backtrace_base_for_ref.
12 2013-09-17  Alan Modra  <amodra@gmail.com>
14         PR target/57589
15         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11 patch.
17 2013-09-16  DJ Delorie  <dj@redhat.com>
19         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
20         vregs location for RL78/G10.
21         (rl78_expand_prologue): Avoid SEL on G10.
22         (rl78_expand_epilogue): Likewise.
23         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
24         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
25         __RL78_G10__ when appropriate.
26         (ASM_SPEC): Pass -mg10 along to the assembler.
27         * config/rl78/rl78.md (sel_rb): Disable for G10.
28         * config/rl78/rl78.opt: Add -mg10 option.
29         * config/rl78/t-rl78: Add -mg10 multilib.
31 2013-09-16  Xinliang David Li  <davidxl@google.com>
33         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
34         * omp-low.c (omp_max_vf): Ditto.
35         (expand_omp_simd): Ditto.
36         * tree-vectorizer.c (vectorize_loops): Ditto.
37         (gate_vect_slp): Ditto.
38         (gate_increase_alignment): Ditto.
39         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
40         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
41         (gate_tree_vectorize): Name change.
42         (tree_vectorize): Ditto.
43         (pass_vectorize::gate): Call new function.
44         (pass_vectorize::execute): Ditto.
45         opts.c: O3 default setting change.
46         (finish_options): Check new flag.
47         * doc/invoke.texi: Document new flags.
48         * common.opt: New flags.
50 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
52         * doc/tm.texi.in (Cond Exec Macros): Remove node.
53         (Condition Code): Don't reference it.
54         * doc/tm.texi: Regenerate.
56 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
58         PR middle-end/58418
59         * lra-constraints.c (undo_optional_reloads): Consider all optional
60         reload even if it did not get a hard reg.
62 2013-09-16  Teresa Johnson  <tejohnson@google.com>
64         * dumpfile.c (dump_loc): Remove newline emission.
65         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
66         emission to dump_printf_loc calls where missing.
67         (vect_mark_for_runtime_alias_test): Ditto.
68         (vect_analyze_data_ref_dependence): Ditto.
69         (vect_analyze_data_ref_dependences): Ditto.
70         (vect_slp_analyze_data_ref_dependence): Ditto.
71         (vect_slp_analyze_data_ref_dependences): Ditto.
72         (vect_compute_data_ref_alignment): Ditto.
73         (vect_update_misalignment_for_peel): Ditto.
74         (vect_verify_datarefs_alignment): Ditto.
75         (vector_alignment_reachable_p): Ditto.
76         (vect_get_data_access_cost): Ditto.
77         (vect_enhance_data_refs_alignment): Ditto.
78         (vect_find_same_alignment_drs): Ditto.
79         (vect_analyze_data_refs_alignment): Ditto.
80         (vect_analyze_group_access): Ditto.
81         (vect_analyze_data_ref_access): Ditto.
82         (vect_analyze_data_ref_accesses): Ditto.
83         (vect_prune_runtime_alias_test_list): Ditto.
84         (vect_analyze_data_refs): Ditto.
85         (vect_create_addr_base_for_vector_ref): Ditto.
86         (vect_create_data_ref_ptr): Ditto.
87         (vect_grouped_store_supported): Ditto.
88         (vect_grouped_load_supported): Ditto.
89         * value-prof.c (check_counter): Ditto.
90         (check_ic_target): Ditto.
91         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
92         (vect_recog_widen_mult_pattern): Ditto.
93         (vect_recog_widen_sum_pattern): Ditto.
94         (vect_recog_over_widening_pattern): Ditto.
95         (vect_recog_widen_shift_pattern): Ditto.
96         (vect_recog_rotate_pattern): Ditto.
97         (vect_recog_vector_vector_shift_pattern): Ditto.
98         (vect_recog_divmod_pattern): Ditto.
99         (vect_recog_mixed_size_cond_pattern): Ditto.
100         (vect_recog_bool_pattern): Ditto.
101         (vect_pattern_recog_1): Ditto.
102         (vect_pattern_recog): Ditto.
103         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
104         (vect_is_simple_iv_evolution): Ditto.
105         (vect_analyze_scalar_cycles_1): Ditto.
106         (vect_get_loop_niters): Ditto.
107         (vect_analyze_loop_1): Ditto.
108         (vect_analyze_loop_form): Ditto.
109         (vect_analyze_loop_operations): Ditto.
110         (vect_analyze_loop_2): Ditto.
111         (vect_analyze_loop): Ditto.
112         (report_vect_op): Ditto.
113         (vect_is_slp_reduction): Ditto.
114         (vect_is_simple_reduction_1): Ditto.
115         (vect_get_known_peeling_cost): Ditto.
116         (vect_estimate_min_profitable_iters): Ditto.
117         (vect_model_reduction_cost): Ditto.
118         (vect_model_induction_cost): Ditto.
119         (get_initial_def_for_induction): Ditto.
120         (vect_create_epilog_for_reduction): Ditto.
121         (vectorizable_reduction): Ditto.
122         (vectorizable_induction): Ditto.
123         (vectorizable_live_operation): Ditto.
124         (vect_loop_kill_debug_uses): Ditto.
125         (vect_transform_loop): Ditto.
126         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
127         (vect_stmt_relevant_p): Ditto.
128         (process_use): Ditto.
129         (vect_mark_stmts_to_be_vectorized): Ditto.
130         (vect_model_simple_cost): Ditto.
131         (vect_model_promotion_demotion_cost): Ditto.
132         (vect_model_store_cost): Ditto.
133         (vect_get_store_cost): Ditto.
134         (vect_model_load_cost): Ditto.
135         (vect_get_load_cost): Ditto.
136         (vect_init_vector_1): Ditto.
137         (vect_get_vec_def_for_operand): Ditto.
138         (vect_finish_stmt_generation): Ditto.
139         (vectorizable_call): Ditto.
140         (vectorizable_conversion): Ditto.
141         (vectorizable_assignment): Ditto.
142         (vectorizable_shift): Ditto.
143         (vectorizable_operation): Ditto.
144         (vectorizable_store): Ditto.
145         (vectorizable_load): Ditto.
146         (vectorizable_condition): Ditto.
147         (vect_analyze_stmt): Ditto.
148         (vect_transform_stmt): Ditto.
149         (vect_is_simple_use): Ditto.
150         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
151         (vect_can_advance_ivs_p): Ditto.
152         (vect_update_ivs_after_vectorizer): Ditto.
153         (vect_do_peeling_for_loop_bound): Ditto.
154         (vect_gen_niters_for_prolog_loop): Ditto.
155         (vect_update_inits_of_drs): Ditto.
156         (vect_create_cond_for_alias_checks): Ditto.
157         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
158         (vect_build_slp_tree_1): Ditto.
159         (vect_supported_load_permutation_p): Ditto.
160         (vect_analyze_slp_instance): Ditto.
161         (vect_analyze_slp): Ditto.
162         (vect_make_slp_decision): Ditto.
163         (vect_detect_hybrid_slp): Ditto.
164         (vect_bb_vectorization_profitable_p): Ditto.
165         (vect_slp_analyze_bb_1): Ditto.
166         (vect_update_slp_costs_according_to_vf): Ditto.
167         (vect_get_mask_element): Ditto.
168         (vect_transform_slp_perm_load): Ditto.
169         (vect_schedule_slp_instance): Ditto.
170         (vect_schedule_slp): Ditto.
171         (vect_slp_transform_bb): Ditto.
172         * profile.c (read_profile_edge_counts): Ditto.
173         (compute_branch_probabilities): Ditto.
174         * coverage.c (get_coverage_counts): Ditto.
176 2013-09-16  Diego Novillo  <dnovillo@google.com>
178         * tree-core.h: Add missing comment lines from refactoring
179         of tree.h.
181 2013-09-16  Jan Hubicka  <jh@suse.cz>
183         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
184         abstract functions; for static functions check the presence
185         of body.
187 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
189         * config/aarch64/aarch64-simd-builtins.def (fma): New.
190         * config/aarch64/aarch64-simd.md
191         (aarch64_mla_elt<mode>): New.
192         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
193         (aarch64_mls_elt<mode>): Likewise.
194         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
195         (aarch64_fma4_elt<mode>): Likewise.
196         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
197         (aarch64_fma4_elt_to_128v2df): Likewise.
198         (aarch64_fma4_elt_to_64df): Likewise.
199         (fnma<mode>4): Likewise.
200         (aarch64_fnma4_elt<mode>): Likewise.
201         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
202         (aarch64_fnma4_elt_to_128v2df): Likewise.
203         (aarch64_fnma4_elt_to_64df): Likewise.
204         * config/aarch64/iterators.md (VDQSF): New.
205         * config/aarch64/arm_neon.h
206         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
207         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
209 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
211         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
212         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
213         (aarch64_mul3_elt_to_128df): Likewise.
214         (aarch64_mul3_elt_to_64v2df): Likewise.
215         * config/aarch64/iterators.md (VEL): Also handle DFmode.
216         (VMUL): New.
217         (VMUL_CHANGE_NLANES) Likewise.
218         (h_con): Likewise.
219         (f): Likewise.
220         * config/aarch64/arm_neon.h
221         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
223 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
225         * config/aarch64/arm_neon.h
226         (vcvtx_high_f32_f64): Fix parameters.
228 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
229             Uros Bizjak  <ubizjak@gmail.com>
231         * config/alpha.c: Include tree-ssanames.h.
233 2013-09-16  Richard Biener  <rguenther@suse.de>
235         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
236         (dot_rdg_1): Handle control_dd.
237         (create_edge_for_control_dependence): New function.
238         (create_rdg_edges): Add control dependences if asked for.
239         (build_rdg): Likewise.
240         (generate_loops_for_partition): If there are not necessary
241         control stmts remove all their dependencies.
242         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
243         (distribute_loop): Pass on control dependences.
244         (tree_loop_distribution): Compute control dependences and remove
245         restriction on number of loop nodes.
247 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
249         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
250         for internal calls.
252 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
254         * cse.c (try_const_anchors): Punt on CC modes.
256 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
258         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
260 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
262         PR target/58382
263         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
264         calls to word_mode.
266 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
268         PR target/48094
269         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
270         seen.
271         (darwin_objc1_section): Likewise.
272         (darwin_file_end): Emit Image Info section when required.
274 2013-09-14  Jan Hubicka  <jh@suse.cz>
276         * tree-into-ssa.c (gate_into_ssa): New.
277         (pass_data_build_ssa): Use it.
278         * cgraph.h (expand_thunk): Update prototype.
279         * cgraphunit.c (analyze_function): Expand thunks early.
280         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
281         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
282         set lowered flag; do not add new function.
283         (assemble_thunks_and_aliases): Update.
284         * tree-ssa.c (gate_init_datastructures): New gate.
285         (pass_data_init_datastructures): Use it.
287 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
289         PR target/58269
290         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
291         xmm register set described in the psABI.
293 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
295         * dwarf2out.c (should_emit_struct_debug): Add check
296         for type_decl variable is not NULL.
298 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
300         * config.gcc: Use new winnt-c.c target hooks
301         * config/t-winnt: New file
302         * config/winnt-c.c: New file
303         * doc/tm.texi.in: Document new hook
304         * doc/tm.texi: Regenerated
306 2013-09-13  Jan Hubicka  <jh@suse.cz>
308         PR middle-end/58094
309         * ipa-inline.c (check_callers): New function.
310         (check_caller_edge): Remove.
311         (want_inline_function_to_all_callers_p): Also permit alises that are
312         called dirrectly.
313         (inline_to_all_callers): Terminate the walk when devirtualization
314         introduce new calls.
316 2013-09-13  Jan Hubicka  <jh@suse.cz>
318         * ipa-inline-analysis.c (struct growth_data): Add node.
319         (do_estimate_growth_1): Fix detection of recursion.
321 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
323         PR tree-optimization/58392
324         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
325         to avoid shadowing of outer loop variable.  If
326         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
327         replace_by_duplicate_decl simduid of loops that have it set and
328         set dest_cfun->has_simduid_loops and/or
329         dest_cfun->has_force_vect_loops.
330         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
331         instead of maybe_lookup_decl.
332         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
333         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
334         call.  Copy over force_vect and copy and remap simduid.  Set
335         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
336         (copy_cfg_body): Remove blocks_to_copy argument.  Use
337         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
338         caller.  Don't set cfun->has_simduid_loops and/or
339         cfun->has_force_vect_loops here.
340         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
341         caller.
342         (expand_call_inline, tree_function_versioning): Adjust copy_body
343         callers.
345 2013-09-13  Martin Jambor  <mjambor@suse.cz>
347         PR bootstrap/58388
348         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
349         the assert if the edge was a speculative one.
351 2013-09-13  Richard Biener  <rguenther@suse.de>
353         * tree-data-ref.h (known_dependences_p): Move here ...
354         * tree-loop-distribution.c (known_dependences_p): ... from here.
355         (dump_rdg_component, debug_rdg_component): Remove.
356         (dump_rdg): Adjust.
357         (generate_loops_for_partition): Use gimple_uid instead of
358         relying on matching stmt visit order.
359         (rdg_build_partitions): Take starting stmt vector.
360         (ldist_gen): Merge into ...
361         (distribute_loop): ... this function.  Do not compute starting
362         vertices vector.
363         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
365 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
367         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
368         Set type attribute correctly. Set predicable_short_it attribute.
369         (cmpsi_shiftsi): Remove %? from output template.
371 2013-09-13  Richard Biener  <rguenther@suse.de>
373         * tree-loop-distribution.c (struct rdg_component,
374         rdg_defs_used_in_other_loops_p, free_rdg_components,
375         rdg_build_components): Remove.
376         (stmts_from_loop): Do not record virtual PHIs.
377         (generate_loops_for_partition): Skip virtual PHIs.
378         (build_rdg_partition_for_component): Rename to ...
379         (build_rdg_partition_for_vertex): ... this and adjust.
380         (rdg_build_partitions): Take a vector of starting vertices
381         instead of components.  Remove unnecessary leftover handling.
382         (ldist_gen): Do not build components or record other stores.
383         (distribute_loop): Do not distribute loops containing stmts
384         with side-effects.
386 2013-09-13  Christian Bruel  <christian.bruel@st.com>
388         PR target/58314
389         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
391 2013-09-13  Kai Tietz  <ktietz@redhat.com>
393         * config.gcc: Separate cases for mingw and cygwin targets,
394         and add 64-bit cygwin target case.
396         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
397         dll-export inline-functions.
398         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
401 2013-09-13  Jeff Law  <law@redhat.com>
403         PR middle-end/58387
404         Revert:
405         2013-09-06  Jeff Law  <law@redhat.com>
407         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
408         edge implied equivalences into successor phis.
410 2013-09-12  DJ Delorie  <dj@redhat.com>
412         * config/rl78/rl78-virt.md: Change from | to \; for asm line
413         separators.
415 2013-09-12  Brooks Moses  <bmoses@google.com>
417         PR driver/42955
418         * Makefile.in: Do not install driver binaries in $(target)/bin.
420 2013-09-12  DJ Delorie  <dj@redhat.com>
422         * config/rl78/rl78.opt (mrelax): New.
423         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
424         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
426         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
427         between SP and FP.
428         (rl78_expand_epilogue): Likewise.
430 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
432         PR middle-end/58335
433         * lra-eliminations.c (remove_reg_equal_offset_note): New.
434         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
435         pointer elimination with using remove_reg_equal_offset_note.
437 2013-09-12  DJ Delorie  <dj@redhat.com>
439         * config/msp430/: New port.
440         * config.gcc (msp430): Added.
441         * doc/invoke.texi: Document MSP430 options.
442         * doc/install.texi: Document msp430-elf
443         * doc/md.texi: Document msp430-elf
444         * doc/contrib.texi: Document msp430-elf
446         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
447         PSImode.
449 2013-09-12  Martin Jambor  <mjambor@suse.cz>
451         PR ipa/58389
452         * ipa-prop.c (remove_described_reference): Give up if the edge in the
453         reference descriptor is NULL.
454         (ipa_edge_removal_hook): If owning a reference descriptor, set its
455         edge to NULL.
457 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
459         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
460         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
461         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
462         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
463         make_temp_ssa_name): move to tree-ssanames.h
464         * tree-ssa-alias.h: Move prototype.
465         * tree-ssa.h: Include tree-ssanames.h.
466         * tree-ssanames.c (FREE_SSANAMES): Move to here.
467         * tree-ssanames.h: New.  Move items from tree-flow*.h
468         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
470 2013-09-12  Richard Biener  <rguenther@suse.de>
472         PR tree-optimization/58404
473         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
474         propagate non-invariant addresses into dereferences wrapped
475         in component references.
477 2013-09-12  Richard Biener  <rguenther@suse.de>
479         PR tree-optimization/58402
480         * passes.def: Move pass_late_warn_uninitialized later.
482 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
484         * tree-ssa.h: New.  Move content from tree-flow.h and
485         tree-flow-inline.h.
486         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
487         Move prototypes belonging to tree-ssa.c.
488         * tree-flow-inline.h (redirect_edge_var_map_def,
489         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
490         tree-ssa.h.
491         * gimple.h: Adjust prototypes.
492         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
493         to...
494         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
495         * tree.h: Move prototype to tree-ssa.h.
496         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
497         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
498         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
499         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
500         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
501         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
502         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
503         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
504         graphite-blocking.c, graphite-clast-to-gimple.c,
505         graphite-dependences.c, graphite-interchange.c,
506         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
507         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
508         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
509         ipa-reference.c, ipa-split.c, ipa-utils.c,
510         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
511         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
512         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
513         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
514         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
515         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
516         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
517         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
518         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
519         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
520         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
521         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
522         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
523         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
524         value-prof.c, var-tracking.c,
525         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
527 2013-09-12  Richard Biener  <rguenther@suse.de>
529         PR tree-optimization/58396
530         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
531         (build_rdg): Take a loop-nest parameter, fix memleaks.
532         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
534 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
536         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
537         for SLM.
539 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
541         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
542         The FMA instruction names should have a 'v' prefix.
544 2013-09-12  Richard Biener  <rguenther@suse.de>
546         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
547         (dot_rdg): Use popen instead of system in optional code.
548         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
549         (already_processed_vertex_p): Adjust.
550         (has_anti_or_output_dependence, predecessor_has_mem_write,
551         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
552         rdg_flag_uses): Remove.
553         (rdg_flag_vertex): Simplify.
554         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
555         remove recursion.
556         (build_rdg_partition_for_component): Process the first vertex
557         of a component only.
558         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
560 2013-09-12  Alan Modra  <amodra@gmail.com>
562         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
564 2013-09-11  DJ Delorie  <dj@redhat.com>
565             Nick Clifton  <nickc@redhat.com>
567         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
568         (rl78_stack_based_mem): New.
569         * config/rl78/constraints.md (Iv08): New.
570         (Iv16): New.
571         (Iv24): New.
572         (Is09): New.
573         (Is17): New.
574         (Is25): New.
575         (ISsi): New.
576         (IShi): New.
577         (ISqi): New.
578         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
579         (movhi): Likewise.
580         (movsi): Change from expand to insn-and-split.
581         (ashrsi3): Clobber AX.
582         (lshrsi3): New.
583         (ashlsi3): New.
584         (cbranchsi4): New.
585         * config/rl78/rl78.md (CC_REG): Fix.
586         (addsi3): Allow memory and immediate operands.
587         (addsi3_internal): Split into...
588         (addsi3_internal_virt): ...new, and ...
589         (addsi3_internal_real): ...new.
590         (subsi): New.
591         (subsi3_internal_virt): New.
592         (subsi3_internal_real): New.
593         (mulsi3): Add memory operand.
594         (mulsi3_rl78): Likewise.
595         (mulsi3_g13): Likewise.
596         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
597         (cbranchqi4_real): Add more constraint options.
598         (cbranchhi4_real): Expand pattern.
599         (cbranchhi4_real_signed): New.
600         (cbranchhi4_real_inverted): New.
601         (cbranchsi4_real_lt): New.
602         (cbranchsi4_real_ge): New.
603         (cbranchsi4_real_signed): New.
604         (cbranchsi4_real): New.
605         (peephole2): New.
606         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
607         constant shifts.
608         (lshrsi3_virt): Likewise.
609         (ashlsi3_virt): Likewise.
610         (cbranchqi4_virt_signed): New.
611         (cbranchhi4_virt_signed): New.
612         (cbranchsi4_virt): New.
613         * config/rl78/rl78.c: Whitespace fixes throughout.
614         (move_elim_pass): New.
615         (pass_data_rl78_move_elim): New.
616         (pass_rl78_move_elim): New.
617         (make_pass_rl78_move_elim): New.
618         (rl78_devirt_info): Run devirt earlier.
619         (rl78_move_elim_info): New.
620         (rl78_asm_file_start): Register it.
621         (rl78_split_movsi): New.
622         (rl78_as_legitimate_address): Allow virtual base registers when
623         appropriate.
624         (rl78_addr_space_convert): Remove spurious debug stuff.
625         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
626         (rl78_print_operand): More cases for not printing '#'.
627         (rl78_expand_compare): Remove most of the logic.
628         (content_memory): New.
629         (clear_content_memory): New.
630         (get_content_index): New.
631         (get_content_name): New.
632         (display_content_memory): New.
633         (update_content): New.
634         (record_content): New.
635         (already_contains): New.
636         (insn_ok_now): Re-recog insns with virtual registers.
637         (add_postponed_content_update): New.
638         (process_postponed_content_update): New.
639         (gen_and_emit_move): New.
640         (transcode_memory_rtx): Record new location content.
641         Use gen_and_emit_move.
642         (force_into_acc): New.
643         (move_to_acc): Use gen_and_emit_move.
644         (move_from_acc): Likewise.
645         (move_acc_to_reg): Likewise.
646         (move_to_x): Likewise.
647         (move_to_hl): Likewise.
648         (move_to_de): Likewise.
649         (rl78_alloc_physical_registers_op1): Record location content.
650         (has_constraint): New.
651         (rl78_alloc_physical_registers_op2): Record location content.
652         Optimize use of HL.
653         (rl78_alloc_physical_registers_ro1): Likewise.
654         (rl78_alloc_physical_registers_cmp): Likewise.
655         (rl78_alloc_physical_registers_umul): Likewise.
656         (rl78_alloc_address_registers_macax): New.
657         (rl78_alloc_physical_registers): Initialize and set location
658         content memory as needed.
659         (rl78_reorg): Make sure split2 is called.
660         (rl78_rtx_costs): New.
662 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
664         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
665         for (not (neg ...)) and (neg (not ...)) cases.
667 2013-09-11  Richard Biener  <rguenther@suse.de>
669         PR middle-end/58377
670         * passes.def: Split critical edges before late uninit warning passes.
671         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
673 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
675         PR tree-optimization/58385
676         * fold-const.c (build_range_check): If both low and high are NULL,
677         use omit_one_operand_loc to preserve exp side-effects.
679 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
681         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
683 2013-09-11  Richard Biener  <rguenther@suse.de>
685         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
686         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
687         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
688         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
689         stmts_from_loop, known_dependences_p, build_empty_rdg,
690         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
691         * tree-loop-distribution.c: ... here.
692         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
693         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
694         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
695         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
696         * tree-loop-distribution.c: ... here.
697         * tree-loop-distribution.c: Include gimple-pretty-print.h.
698         (struct partition_s): Add loops member.
699         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
700         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
701         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
703 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
704             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
705             Sergey Lega  <sergey.s.lega@intel.com>
706             Anna Tikhonova  <anna.tikhonova@intel.com>
707             Ilya Tocar  <ilya.tocar@intel.com>
708             Andrey Turetskiy  <andrey.turetskiy@intel.com>
709             Ilya Verbin  <ilya.verbin@intel.com>
710             Kirill Yukhin  <kirill.yukhin@intel.com>
711             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
713         * config/i386/constraints.md (k): New.
714         (Yk): Ditto.
715         * config/i386/i386.c (const regclass_map): Add new mask registers.
716         (dbx_register_map): Ditto.
717         (dbx64_register_map): Ditto.
718         (svr4_dbx_register_map): Ditto.
719         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
720         disabled.
721         (ix86_preferred_reload_class): Disable constants for mask registers.
722         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
723         (ix86_hard_regno_mode_ok): Support new mask registers.
724         (x86_order_regs_for_local_alloc): Ditto.
725         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
726         (FIXED_REGISTERS): Add new mask registers.
727         (CALL_USED_REGISTERS): Ditto.
728         (REG_ALLOC_ORDER): Ditto.
729         (VALID_MASK_REG_MODE): New.
730         (FIRST_MASK_REG): Ditto.
731         (LAST_MASK_REG): Ditto.
732         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
733         (MAYBE_MASK_CLASS_P): New.
734         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
735         (REG_CLASS_CONTENTS): Ditto.
736         (MASK_REGNO_P): New.
737         (ANY_MASK_REG_P): Ditto.
738         (HI_REGISTER_NAMES): Add new mask registers.
739         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
740         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
741         mask registers.
742         (attribute "type"): Add mskmov, msklog.
743         (attribute "length_immediate"): Support them.
744         (attribute "memory"): Ditto.
745         (attribute "prefix_0f"): Ditto.
746         (*movhi_internal): Support new mask registers.
747         (*movqi_internal): Ditto.
748         (define_split): Split out clobber pattern is a logic
749         insn on mask registers.
750         (*k<logic><mode>): New.
751         (*andhi_1): Extend to support mask regs.
752         (*andqi_1): Extend to support mask regs.
753         (kandn<mode>): New.
754         (define_split): Split and-not to and and not if operands
755         are not mask regs.
756         (*<code><mode>_1): Separate HI mode to new pattern...
757         (*<code>hi_1): This.
758         (*<code>qi_1): Extend to support mask regs.
759         (kxnor<mode>): New.
760         (kortestzhi): Ditto.
761         (kortestchi): Ditto.
762         (kunpckhi): Ditto.
763         (*one_cmpl<mode>2_1): Remove HImode and handle it...
764         (*one_cmplhi2_1): ...Here, now with mask registers support.
765         (*one_cmplqi2_1): Support new mask registers.
766         (HI/QImode arithmetics splitter): Don't split if mask registers
767         are used.
768         (HI/QImode not splitter): Ditto.
769         * config/i386/predicated.md (mask_reg_operand): New.
770         (general_reg_operand): Ditto.
772 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
774         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
775         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
777 2013-09-10  Jeff Law  <law@redhat.com>
779         PR tree-optimization/58380
780         * tree-ssa-threadupdate.c (thread_block): Recognize another case
781         of threading through a buried loop header.
783         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
784         return value for single successor case.
786 2013-09-10  Jan Hubicka  <jh@suse.cz>
788         * ipa-devirt.c (ipa_devirt): Enable with LTO.
790 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
792         PR target/58361
793         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
794         support conditional execution.
795         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
797 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
799         * lra.c (lra): Clear lra_optional_reload_pseudos before every
800         constraint pass.
801         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
802         Check destination too to check move insn.
803         (undo_optional_reloads): Add check that the original peudo did not
804         changed its allocation and the optional reload was inherited on last
805         inheritance pass.  Break loop after deciding to keep optional reload.
806         (lra_undo_inheritance): Add check that inherited pseudo still in
807         memory.
809 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
811         * config/aarch64/aarch64.md (generic_sched): New.
812         * config/aarch64/aarch64-generic.md (load): Make conditional
813         on generic_sched attribute.
814         (nonload): Likewise.
816 2013-09-10  Jan Hubicka  <jh@suse.cz>
818         * lto-cgraph.c: Include ipa-utils.h.
819         (compute_ltrans_boundary): Also add possible targets into the boundary.
821 2013-09-10  Jan Hubicka  <jh@suse.cz>
823         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
824         VAR_DECL of vtable rather than full expression.
826 2013-09-10  Jan Hubicka  <jh@suse.cz>
827             Paolo Carlini  <paolo.carlini@oracle.com>
829         * cgraphunit.c (analyze_functions): Save input_location, set it
830         to UNKNOWN_LOCATION and restore it at the end.
832 2013-09-10  Martin Jambor  <mjambor@suse.cz>
834         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
835         represented by a thunk.
837 2013-09-10  Jeff Law  <law@redhat.com>
839         PR tree-optimization/58343
840         * tree-ssa-threadupdate.c (thread_block): Identify and disable
841         jump threading requests through loop headers buried in the middle
842         of a jump threading path.
844         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
845         in return value/type.
847 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
849         PR rtl-optimization/58365
850         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
851         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
852         it differs.
854 2013-09-10  Richard Biener  <rguenther@suse.de>
856         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
857         * tree-data-ref.c (create_rdg_vertices): Collect all data
858         references, signal failure to the caller, use data-ref API.
859         (build_rdg): Compute data references only once.  Maintain lifetime
860         of data references and data dependences from within RDG.
861         (free_rdg): Free dependence relations.
862         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
863         inventing extra dependences.
864         (distribute_loop): Update for RDG API changes.
866 2013-09-10  Kai Tietz  <ktietz@redhat.com>
868         * doc/invoke.texi (fms-extensions): Document changed
869         behavior for ms-abi targets.
870         * config/i386/i386.c (ix86_option_override_internal):
871         Set default value of option -fms-extension for ms-abi targets.
873 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
875         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
877 2013-09-10  Alan Modra  <amodra@gmail.com>
879         PR target/58330
880         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
882 2013-09-10  Alan Modra  <amodra@gmail.com>
884         * config/rs6000/predicates.md (add_cint_operand): New.
885         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
886         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
887         using add_cint_operand.
888         (largetoc_high_plus_aix): Likewise.
890 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
892         PR tree-optimization/58364
893         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
894         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
895         the current range can't be an unconditional true or false.
897 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
899         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
901 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
903         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
905 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
907         PR c++/43452
908         * doc/invoke.texi (-Wdelete-incomplete): Document it.
910 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
912         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
913         NO_REGS for immediate that can't be moved directly into FP_REGS.
915 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
917         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
918         comparison with negated operand.
919         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
920         RTL form.
922 2013-09-09  Richard Biener  <rguenther@suse.de>
924         PR middle-end/58326
925         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
926         of a subloop record all its block as affecting loop-closed SSA form.
928 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
930         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
931         of an rtx/bitpos pair.
932         (store_fixed_bit_field): Update accordingly.
934 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
936         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
937         GEN_INT.
938         * builtins.c (expand_errno_check): Likewise.
939         * dwarf2cfi.c (init_return_column_size): Likewise.
940         * except.c (sjlj_mark_call_sites): Likewise.
941         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
942         * lra-constraints.c (emit_inc): Likewise.
943         * ree.c (combine_set_extension): Likewise.
944         * regmove.c (fixup_match_2): Likewise.
945         * reload1.c (inc_for_reload): Likewise.
947 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
949         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
950         (force_to_mode, simplify_shift_const_1, simplify_comparison):
951         Use gen_int_mode with the mode of the associated simplify_* call.
952         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
953         * expmed.c (expand_shift_1): Likewise.
954         * function.c (instantiate_virtual_regs_in_insn): Likewise.
955         * loop-iv.c (iv_number_of_iterations): Likewise.
956         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
957         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
959 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
961         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
962         of the associated expand_* call.
963         (asan_emit_stack_protection): Likewise.
964         * builtins.c (round_trampoline_addr): Likewise.
965         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
966         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
967         (emit_store_flag): Likewise.
968         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
969         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
970         Likewise.
971         * function.c (instantiate_virtual_regs_in_insn): Likewise.
972         * ifcvt.c (noce_try_store_flag_constants): Likewise.
973         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
974         * modulo-sched.c (generate_prolog_epilog): Likewise.
975         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
976         (expand_ctz, expand_ffs, expand_unop): Likewise.
978 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
980         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
981         of the associated gen_rtx_* call.
982         * caller-save.c (init_caller_save): Likewise.
983         * combine.c (find_split_point, make_extraction): Likewise.
984         (make_compound_operation): Likewise.
985         * dwarf2out.c (mem_loc_descriptor): Likewise.
986         * explow.c (plus_constant, probe_stack_range): Likewise.
987         * expmed.c (expand_mult_const): Likewise.
988         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
989         * reload1.c (init_reload): Likewise.
990         * valtrack.c (cleanup_auto_inc_dec): Likewise.
991         * var-tracking.c (adjust_mems): Likewise.
992         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
993         rather than gen_rtx_fmt_ee.
995 2013-09-09  Jan Hubicka  <jh@suse.cz>
997         PR middle-end/58294
998         * value-prof.c (gimple_ic): Copy also abnormal edges.
1000 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
1002         * asan.c (asan_shadow_cst): Use gen_int_mode.
1004 2013-09-08  Jan Hubicka  <jh@suse.cz>
1006         * ipa-profile.c: Add toplevel comment.
1007         (ipa_propagate_frequency_1): Be more conservative when profile is read.
1008         (contains_hot_call_p): New function.
1009         (ipa_propagate_frequency): Set frequencies based on counts when
1010         profile is read.
1011         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
1012         profile; do not tamper with profile after inlining if it is read.
1014 2013-09-08  Jan Hubicka  <jh@suse.cz>
1016         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
1017         speculative edges.
1019 2013-09-08  Jan Hubicka  <jh@suse.cz>
1021         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
1022         summary generation.
1024 2013-09-08  Jeff Law  <law@redhat.com>
1026         PR bootstrap/58340
1027         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
1028         of 'found'.
1030 2013-09-08  Andi Kleen  <ak@linux.intel.com>
1032         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
1034 2013-09-08  Jan Hubicka  <jh@suse.cz>
1036         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
1038 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
1040         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
1041         for non-debug insns.
1042         * lra.c (new_insn_reg): Take the containing insn as a parameter.
1043         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
1044         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
1045         accordingly.
1047 2013-09-08  Jan Hubicka  <jh@suse.cz>
1049         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
1050         targets and devirtualize to BUILT_IN_UNREACHABLE.
1051         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
1052         * ipa.c (walk_polymorphic_call_targets): New function.
1053         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
1054         functions; use the new timevar.
1055         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
1056         was removed from the program.
1057         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
1058         not consider it in the walk when its vtable is dead.
1059         (possible_polymorphic_call_targets_1): Pass anonymous flag to
1060         record_binfo.
1061         (devirt_variable_node_removal_hook): New function.
1062         (possible_polymorphic_call_targets): Also register
1063         devirt_variable_node_removal_hook.
1064         (ipa_devirt): Do not do non-speculative devirtualization.
1065         (gate_ipa_devirt): One execute if devirtualizing speculatively.
1067 2013-09-08  Jan Hubicka  <jh@suse.cz>
1069         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
1070         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
1071         varpool_remove_variable_insertion_hook): Declare.
1072         * varpool.c (varpool_node_hook_list): New structure.
1073         (first_varpool_node_removal_hook,
1074         first_varpool_variable_insertion_hook): New variables.
1075         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
1076         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
1077         varpool_remove_variable_insertion_hook,
1078         varpool_call_variable_insertion_hooks): New functions.
1079         (varpool_remove_node): Use it.
1081 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
1083         PR c++/54941
1084         * diagnostic.c (diagnostic_build_prefix): When s.file is
1085         "<built-in>" don't output line and column numbers.
1087 2013-09-06  Jan Hubicka  <jh@suse.cz>
1089         * cgraphunit.c (expand_thunk): Get body before touching arguments.
1090         * lto-streamer-out.c: Stream thunks, too.
1091         * lto-streamer-in.c (input_function): Pop cfun here
1092         (lto_read_body): Instead of here.
1094 2013-09-06  Caroline Tice  <cmtice@google.com>
1096         * doc/install.texi: Add documentation for the --enable-vtable-verify
1097         and the --disable-libvtv configure options.
1099 2013-09-06  Jeff Law  <law@redhat.com>
1101         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
1102         edge implied equivalences into successor phis.
1104 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
1106         * resource.c (mark_referenced_resources): Handle COND_EXEC.
1108 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
1110         * resource.c (mark_target_live_regs): Compute resources taking
1111         into account if a call is predicated or not.
1113 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
1115         * toplev.c (output_stack_usage): Be prepared for suffixes created by
1116         the compiler in the function names.
1118 2013-09-06  Jan Hubicka  <jh@suse.cz>
1120         PR middle-end/58094
1121         * ipa-inline.c (has_caller_p): New function.
1122         (want_inline_function_to_all_callers_p): Use it.
1123         (sum_callers, inline_to_all_callers): Break out from ...
1124         (ipa_inline): ... here.
1126 2013-09-06  Jan Hubicka  <jh@suse.cz>
1128         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
1129         only when AVX is enabled.
1131 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1133         * config/aarch64/aarch64.md
1134         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
1135         is fpsimd_<load/store>2.
1136         (load_pair<mode>): Likewise.
1137         (store_pair<mode>): Likewise.
1139 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1141         * config/arm/types.md (type): Add "mrs" type.
1142         * config/aarch64/aarch64.md
1143         (aarch64_load_tp_hard): Make type "mrs".
1144         * config/arm/arm.md
1145         (load_tp_hard): Make type "mrs".
1146         * config/arm/cortex-a15.md: Update with new attributes.
1147         * config/arm/cortex-a5.md: Update with new attributes.
1148         * config/arm/cortex-a53.md: Update with new attributes.
1149         * config/arm/cortex-a7.md: Update with new attributes.
1150         * config/arm/cortex-a8.md: Update with new attributes.
1151         * config/arm/cortex-a9.md: Update with new attributes.
1152         * config/arm/cortex-m4.md: Update with new attributes.
1153         * config/arm/cortex-r4.md: Update with new attributes.
1154         * config/arm/fa526.md: Update with new attributes.
1155         * config/arm/fa606te.md: Update with new attributes.
1156         * config/arm/fa626te.md: Update with new attributes.
1157         * config/arm/fa726te.md: Update with new attributes.
1159 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1161         * config/aarch64/aarch64.md
1162         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
1163         (*movtf_aarch64): Likewise.
1164         * config/arm/arm.md
1165         (thumb1_movdi_insn): Use "multiple" for type where more than one
1166         instruction is used for a move.
1167         (*arm32_movhf): Likewise.
1168         (*thumb_movdf_insn): Likewise.
1170 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1172         * config/arm/types.md (type): Rename fcpys to fmov.
1173         * config/arm/vfp.md
1174         (*arm_movsi_vfp): Rename type fcpys as fmov.
1175         (*thumb2_movsi_vfp): Likewise
1176         (*movhf_vfp_neon): Likewise
1177         (*movhf_vfp): Likewise
1178         (*movsf_vfp): Likewise
1179         (*thumb2_movsf_vfp): Likewise
1180         (*movsfcc_vfp): Likewise
1181         (*thumb2_movsfcc_vfp): Likewise
1182         * config/aarch64/aarch64-simd.md
1183         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
1184         * config/aarch64/aarch64.md
1185         (*movsi_aarch64): Replace type mov_reg with fmovs.
1186         (*movdi_aarch64): Likewise
1187         (*movsf_aarch64): Likewise
1188         (*movdf_aarch64): Likewise
1189         * config/arm/arm.c
1190         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
1191         * config/arm/iwmmxt.md
1192         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
1193         * config/arm/arm1020e.md: Update with new attributes.
1194         * config/arm/cortex-a15-neon.md: Update with new attributes.
1195         * config/arm/cortex-a5.md: Update with new attributes.
1196         * config/arm/cortex-a53.md: Update with new attributes.
1197         * config/arm/cortex-a7.md: Update with new attributes.
1198         * config/arm/cortex-a8-neon.md: Update with new attributes.
1199         * config/arm/cortex-a9.md: Update with new attributes.
1200         * config/arm/cortex-m4-fpu.md: Update with new attributes.
1201         * config/arm/cortex-r4f.md: Update with new attributes.
1202         * config/arm/marvell-pj4.md: Update with new attributes.
1203         * config/arm/vfp11.md: Update with new attributes.
1205 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1207         * config/aarch64/aarch64.md
1208         (*madd<mode>): Fix type attribute.
1209         (*maddsi_uxtw): Likewise.
1210         (*msub<mode>): Likewise.
1211         (*msubsi_uxtw): Likewise.
1212         (<su_optab>maddsidi4): Likewise.
1213         (<su_optab>msubsidi4): Likewise.
1215 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1217         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
1218         * config/arm/arm.md (core_cycles): Remove fdiv.
1219         * config/arm/vfp.md:
1220         (*sqrtsf2_vfp): Update for attribute changes.
1221         (*sqrtdf2_vfp): Likewise.
1222         * config/aarch64/aarch64.md:
1223         (sqrt<mode>2): Update for attribute changes.
1224         * config/arm/arm1020e.md: Update with new attributes.
1225         * config/arm/cortex-a15-neon.md: Update with new attributes.
1226         * config/arm/cortex-a5.md: Update with new attributes.
1227         * config/arm/cortex-a53.md: Update with new attributes.
1228         * config/arm/cortex-a7.md: Update with new attributes.
1229         * config/arm/cortex-a8-neon.md: Update with new attributes.
1230         * config/arm/cortex-a9.md: Update with new attributes.
1231         * config/arm/cortex-m4-fpu.md: Update with new attributes.
1232         * config/arm/cortex-r4f.md: Update with new attributes.
1233         * config/arm/marvell-pj4.md: Update with new attributes.
1234         * config/arm/vfp11.md: Update with new attributes.
1236 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1238         * config/arm/types.md
1239         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
1240         * config/aarch64/aarch64.md
1241         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
1242         new attributes.
1243         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
1244         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
1245         (float<GPI:mode><GPF:mode>2): Likewise.
1246         * config/arm/vfp.md
1247         (*truncsisf2_vfp): Update with new attributes.
1248         (*truncsidf2_vfp): Likewise.
1249         (fixuns_truncsfsi2): Likewise.
1250         (fixuns_truncdfsi2): Likewise.
1251         (*floatsisf2_vfp): Likewise.
1252         (*floatsidf2_vfp): Likewise.
1253         (floatunssisf2): Likewise.
1254         (floatunssidf2): Likewise.
1255         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
1256         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
1257         * config/arm/arm1020e.md: Update with new attributes.
1258         * config/arm/cortex-a15-neon.md: Update with new attributes.
1259         * config/arm/cortex-a5.md: Update with new attributes.
1260         * config/arm/cortex-a53.md: Update with new attributes.
1261         * config/arm/cortex-a7.md: Update with new attributes.
1262         * config/arm/cortex-a8-neon.md: Update with new attributes.
1263         * config/arm/cortex-a9.md: Update with new attributes.
1264         * config/arm/cortex-m4-fpu.md: Update with new attributes.
1265         * config/arm/cortex-r4f.md: Update with new attributes.
1266         * config/arm/marvell-pj4.md: Update with new attributes.
1267         * config/arm/vfp11.md: Update with new attributes.
1269 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1271         * config/aarch64/arm_neon.h
1272         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
1273         (vqtbx<1,2,3,4><q>_s8): Likewise.
1275 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1277         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
1278         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
1279         (add<mode>3): Add type attribute.
1280         (add<mode>3): Likewise.
1281         (usadd<mode>3): Likewise.
1282         (ssadd<mode>3): Likewise.
1283         (sub<mode>3): Likewise.
1284         (sub<mode>3): Likewise.
1285         (ussub<mode>3): Likewise.
1286         (sssub<mode>3): Likewise.
1287         (ssmulsa3): Likewise.
1288         (usmulusa3): Likewise.
1289         (arm_usatsihi): Likewise.
1290         * config/arm/vfp.md
1291         (*movdi_vfp): Add types for all instructions.
1292         (*movdi_vfp_cortexa8): Likewise.
1293         (*movhf_vfp_neon): Likewise.
1294         (*movhf_vfp): Likewise.
1295         (*movdf_vfp): Likewise.
1296         (*thumb2_movdf_vfp): Likewise.
1297         (*thumb2_movdfcc_vfp): Likewise.
1298         * config/arm/arm.md: Add type attribute to all insn patterns.
1299         (*thumb1_adddi3): Add type attribute.
1300         (*arm_adddi3): Likewise.
1301         (*adddi_sesidi_di): Likewise.
1302         (*adddi_zesidi_di): Likewise.
1303         (*thumb1_addsi3): Likewise.
1304         (addsi3_compare0): Likewise.
1305         (*addsi3_compare0_scratch): Likewise.
1306         (*compare_negsi_si): Likewise.
1307         (cmpsi2_addneg): Likewise.
1308         (*addsi3_carryin_<optab>): Likewise.
1309         (*addsi3_carryin_alt2_<optab>): Likewise.
1310         (*addsi3_carryin_clobercc_<optab>): Likewise.
1311         (*subsi3_carryin): Likewise.
1312         (*subsi3_carryin_const): Likewise.
1313         (*subsi3_carryin_compare): Likewise.
1314         (*subsi3_carryin_compare_const): Likewise.
1315         (*arm_subdi3): Likewise.
1316         (*thumb_subdi3): Likewise.
1317         (*subdi_di_zesidi): Likewise.
1318         (*subdi_di_sesidi): Likewise.
1319         (*subdi_zesidi_di): Likewise.
1320         (*subdi_sesidi_di): Likewise.
1321         (*subdi_zesidi_ze): Likewise.
1322         (thumb1_subsi3_insn): Likewise.
1323         (*arm_subsi3_insn): Likewise.
1324         (*anddi3_insn): Likewise.
1325         (*anddi_zesidi_di): Likewise.
1326         (*anddi_sesdi_di): Likewise.
1327         (*ne_zeroextracts): Likewise.
1328         (*ne_zeroextracts): Likewise.
1329         (*ite_ne_zeroextr): Likewise.
1330         (*ite_ne_zeroextr): Likewise.
1331         (*anddi_notdi_di): Likewise.
1332         (*anddi_notzesidi): Likewise.
1333         (*anddi_notsesidi): Likewise.
1334         (andsi_notsi_si): Likewise.
1335         (thumb1_bicsi3): Likewise.
1336         (*iordi3_insn): Likewise.
1337         (*iordi_zesidi_di): Likewise.
1338         (*iordi_sesidi_di): Likewise.
1339         (*thumb1_iorsi3_insn): Likewise.
1340         (*xordi3_insn): Likewise.
1341         (*xordi_zesidi_di): Likewise.
1342         (*xordi_sesidi_di): Likewise.
1343         (*arm_xorsi3): Likewise.
1344         (*andsi_iorsi3_no): Likewise.
1345         (*smax_0): Likewise.
1346         (*smax_m1): Likewise.
1347         (*arm_smax_insn): Likewise.
1348         (*smin_0): Likewise.
1349         (*arm_smin_insn): Likewise.
1350         (*arm_umaxsi3): Likewise.
1351         (*arm_uminsi3): Likewise.
1352         (*minmax_arithsi): Likewise.
1353         (*minmax_arithsi_): Likewise.
1354         (*satsi_<SAT:code>): Likewise.
1355         (arm_ashldi3_1bit): Likewise.
1356         (arm_ashrdi3_1bit): Likewise.
1357         (arm_lshrdi3_1bit): Likewise.
1358         (*arm_negdi2): Likewise.
1359         (*thumb1_negdi2): Likewise.
1360         (*arm_negsi2): Likewise.
1361         (*thumb1_negsi2): Likewise.
1362         (*negdi_extendsid): Likewise.
1363         (*negdi_zero_extend): Likewise.
1364         (*arm_abssi2): Likewise.
1365         (*thumb1_abssi2): Likewise.
1366         (*arm_neg_abssi2): Likewise.
1367         (*thumb1_neg_abss): Likewise.
1368         (one_cmpldi2): Likewise.
1369         (extend<mode>di2): Likewise.
1370         (*compareqi_eq0): Likewise.
1371         (*arm_extendhisi2addsi): Likewise.
1372         (*arm_movdi): Likewise.
1373         (*thumb1_movdi_insn): Likewise.
1374         (*arm_movt): Likewise.
1375         (*thumb1_movsi_insn): Likewise.
1376         (pic_add_dot_plus_four): Likewise.
1377         (pic_add_dot_plus_eight): Likewise.
1378         (tls_load_dot_plus_eight): Likewise.
1379         (*thumb1_movhi_insn): Likewise.
1380         (*thumb1_movsf_insn): Likewise.
1381         (*movdf_soft_insn): Likewise.
1382         (*thumb_movdf_insn): Likewise.
1383         (cbranchsi4_insn): Likewise.
1384         (cbranchsi4_scratch): Likewise.
1385         (*negated_cbranchsi4): Likewise.
1386         (*tbit_cbranch): Likewise.
1387         (*tlobits_cbranch): Likewise.
1388         (*tstsi3_cbranch): Likewise.
1389         (*cbranchne_decr1): Likewise.
1390         (*addsi3_cbranch): Likewise.
1391         (*addsi3_cbranch_scratch): Likewise.
1392         (*arm_cmpdi_insn): Likewise.
1393         (*arm_cmpdi_unsig): Likewise.
1394         (*arm_cmpdi_zero): Likewise.
1395         (*thumb_cmpdi_zero): Likewise.
1396         (*deleted_compare): Likewise.
1397         (*mov_scc): Likewise.
1398         (*mov_negscc): Likewise.
1399         (*mov_notscc): Likewise.
1400         (*cstoresi_eq0_thumb1_insn): Likewise.
1401         (cstoresi_nltu_thumb1): Likewise.
1402         (cstoresi_ltu_thu): Likewise.
1403         (thumb1_addsi3_addgeu): Likewise.
1404         (*arm_jump): Likewise.
1405         (*thumb_jump): Likewise.
1406         (*check_arch2): Likewise.
1407         (arm_casesi_internal): Likewise.
1408         (thumb1_casesi_dispatch): Likewise.
1409         (*arm_indirect_jump): Likewise.
1410         (*thumb1_indirect_jump): Likewise.
1411         (nop): Likewise.
1412         (*and_scc): Likewise.
1413         (*ior_scc): Likewise.
1414         (*compare_scc): Likewise.
1415         (*cond_move): Likewise.
1416         (*cond_arith): Likewise.
1417         (*cond_sub): Likewise.
1418         (*cmp_ite0): Likewise.
1419         (*cmp_ite1): Likewise.
1420         (*cmp_and): Likewise.
1421         (*cmp_ior): Likewise.
1422         (*ior_scc_scc): Likewise.
1423         (*ior_scc_scc_cmp): Likewise.
1424         (*and_scc_scc): Likewise.
1425         (*and_scc_scc_cmp): Likewise.
1426         (*and_scc_scc_nod): Likewise.
1427         (*negscc): Likewise.
1428         (movcond_addsi): Likewise.
1429         (movcond): Likewise.
1430         (*ifcompare_plus_move): Likewise.
1431         (*if_plus_move): Likewise.
1432         (*ifcompare_move_plus): Likewise.
1433         (*if_move_plus): Likewise.
1434         (*ifcompare_arith_arith): Likewise.
1435         (*if_arith_arith): Likewise.
1436         (*ifcompare_arith_move): Likewise.
1437         (*if_arith_move): Likewise.
1438         (*ifcompare_move_arith): Likewise.
1439         (*if_move_arith): Likewise.
1440         (*ifcompare_move_not): Likewise.
1441         (*if_move_not): Likewise.
1442         (*ifcompare_not_move): Likewise.
1443         (*if_not_move): Likewise.
1444         (*ifcompare_shift_move): Likewise.
1445         (*if_shift_move): Likewise.
1446         (*ifcompare_move_shift): Likewise.
1447         (*if_move_shift): Likewise.
1448         (*ifcompare_shift_shift): Likewise.
1449         (*ifcompare_not_arith): Likewise.
1450         (*ifcompare_arith_not): Likewise.
1451         (*if_arith_not): Likewise.
1452         (*ifcompare_neg_move): Likewise.
1453         (*if_neg_move): Likewise.
1454         (*ifcompare_move_neg): Likewise.
1455         (*if_move_neg): Likewise.
1456         (prologue_thumb1_interwork): Likewise.
1457         (*cond_move_not): Likewise.
1458         (*sign_extract_onebit): Likewise.
1459         (*not_signextract_onebit): Likewise.
1460         (stack_tie): Likewise.
1461         (align_4): Likewise.
1462         (align_8): Likewise.
1463         (consttable_end): Likewise.
1464         (consttable_1): Likewise.
1465         (consttable_2): Likewise.
1466         (consttable_4): Likewise.
1467         (consttable_8): Likewise.
1468         (consttable_16): Likewise.
1469         (*thumb1_tablejump): Likewise.
1470         (prefetch): Likewise.
1471         (force_register_use): Likewise.
1472         (thumb_eh_return): Likewise.
1473         (load_tp_hard): Likewise.
1474         (load_tp_soft): Likewise.
1475         (tlscall): Likewise.
1476         (*arm_movtas_ze): Likewise.
1477         (*arm_rev): Likewise.
1478         (*arm_revsh): Likewise.
1479         (*arm_rev16): Likewise.
1480         * config/arm/thumb2.md
1481         (*thumb2_smaxsi3): Likewise.
1482         (*thumb2_sminsi3): Likewise.
1483         (*thumb32_umaxsi3): Likewise.
1484         (*thumb2_uminsi3): Likewise.
1485         (*thumb2_negdi2): Likewise.
1486         (*thumb2_abssi2): Likewise.
1487         (*thumb2_neg_abss): Likewise.
1488         (*thumb2_movsi_insn): Likewise.
1489         (tls_load_dot_plus_four): Likewise.
1490         (*thumb2_movhi_insn): Likewise.
1491         (*thumb2_mov_scc): Likewise.
1492         (*thumb2_mov_negs): Likewise.
1493         (*thumb2_mov_negs): Likewise.
1494         (*thumb2_mov_nots): Likewise.
1495         (*thumb2_mov_nots): Likewise.
1496         (*thumb2_movsicc_): Likewise.
1497         (*thumb2_movsfcc_soft_insn): Likewise.
1498         (*thumb2_indirect_jump): Likewise.
1499         (*thumb2_and_scc): Likewise.
1500         (*thumb2_ior_scc): Likewise.
1501         (*thumb2_ior_scc_strict_it): Likewise.
1502         (*thumb2_cond_move): Likewise.
1503         (*thumb2_cond_arith): Likewise.
1504         (*thumb2_cond_ari): Likewise.
1505         (*thumb2_cond_sub): Likewise.
1506         (*thumb2_negscc): Likewise.
1507         (*thumb2_movcond): Likewise.
1508         (thumb2_casesi_internal): Likewise.
1509         (thumb2_casesi_internal_pic): Likewise.
1510         (*thumb2_alusi3_short): Likewise.
1511         (*thumb2_mov<mode>_shortim): Likewise.
1512         (*thumb2_addsi_short): Likewise.
1513         (*thumb2_subsi_short): Likewise.
1514         (thumb2_addsi3_compare0): Likewise.
1515         (*thumb2_cbz): Likewise.
1516         (*thumb2_cbnz): Likewise.
1517         (*thumb2_one_cmplsi2_short): Likewise.
1518         (*thumb2_negsi2_short): Likewise.
1519         (*orsi_notsi_si): Likewise.
1520         * config/arm/arm1020e.md: Update with new attributes.
1521         * config/arm/arm1026ejs.md: Update with new attributes.
1522         * config/arm/arm1136jfs.md: Update with new attributes.
1523         * config/arm/arm926ejs.md: Update with new attributes.
1524         * config/arm/cortex-a15.md: Update with new attributes.
1525         * config/arm/cortex-a5.md: Update with new attributes.
1526         * config/arm/cortex-a53.md: Update with new attributes.
1527         * config/arm/cortex-a7.md: Update with new attributes.
1528         * config/arm/cortex-a8.md: Update with new attributes.
1529         * config/arm/cortex-a9.md: Update with new attributes.
1530         * config/arm/cortex-m4.md: Update with new attributes.
1531         * config/arm/cortex-r4.md: Update with new attributes.
1532         * config/arm/fa526.md: Update with new attributes.
1533         * config/arm/fa606te.md: Update with new attributes.
1534         * config/arm/fa626te.md: Update with new attributes.
1535         * config/arm/fa726te.md: Update with new attributes.
1537 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1539         * config/aarch64/aarch64-simd.md
1540         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
1541         <vwx> iterator to ensure correct register choice.
1542         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
1543         (aarch64_sqdmull_n<mode>): Likewise.
1544         (aarch64_sqdmull2_n<mode>_internal): Likewise.
1545         * config/aarch64/arm_neon.h
1546         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
1547         (vml<as><q>_n_<su>16): Likewise.
1548         (vml<as>l_high_lane<q>_<su>16): Likewise.
1549         (vml<as>l_high_n_<su>16): Likewise.
1550         (vml<as>l_lane<q>_<su>16): Likewise.
1551         (vml<as>l_n_<su>16): Likewise.
1552         (vmul<q>_lane<q>_<su>16): Likewise.
1553         (vmul<q>_n_<su>16): Likewise.
1554         (vmull_lane<q>_<su>16): Likewise.
1555         (vmull_n_<su>16): Likewise.
1556         (vmull_high_lane<q>_<su>16): Likewise.
1557         (vmull_high_n_<su>16): Likewise.
1558         (vqrdmulh<q>_n_s16): Likewise.
1560 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
1562         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
1563         have the correct lane parameter.
1565 2013-09-06  Richard Biener <rguenther@suse.de>
1567         * cfganal.c (control_dependences::~control_dependences):
1568         Properly free all of the vector.
1570 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
1572         PR target/58269
1573         * config/i386/i386.c (ix86_conditional_register_usage):
1574         Proper initialize extended SSE registers.
1576 2013-09-06  Jan Hubicka  <jh@suse.cz>
1578         PR tree-optimization/58311
1579         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
1581 2013-09-06  Jan Hubicka  <jh@suse.cz>
1583         * Makefile.in (tree-sra.o): Update dependencies.
1584         * tree-sra.c: Include ipa-utils.h
1585         (scan_function): Use recursive_call_p.
1586         (has_caller_p): New function.
1587         (cgraph_for_node_and_aliases): Count also callers of aliases.
1589 2013-09-06  Jan Hubicka  <jh@suse.cz>
1591         PR middle-end/58094
1592         * cgraph.h (symtab_semantically_equivalent_p): Declare.
1593         * tree-tailcall.c: Include ipa-utils.h.
1594         (find_tail_calls): Use it.
1595         * ipa-pure-const.c (check_call): Likewise.
1596         * ipa-utils.c (recursive_call_p): New function.
1597         * ipa-utils.h (recursive_call_p): Dclare.
1598         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
1599         (symtab_semantically_equivalent_p): New function.
1600         * Makefile.in (tree-tailcall.o): Update dependencies.
1602 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
1604         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
1605         non-inlinable part.
1607 2013-09-06  Richard Biener  <rguenther@suse.de>
1609         * lto-streamer.h (lto_global_var_decls): Remove.
1610         * Makefile.in (OBJS): Remove lto-symtab.o.
1611         (lto-symtab.o): Remove.
1612         (GTFILES): Remove lto-symtab.c
1613         * lto-symtab.c: Move to lto/
1615 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1617         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
1618         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
1619         (UNSPEC_FPINT_RINT): New constant definitions.
1620         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
1621         definition with 2 attributes.
1622         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
1623         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
1624         definitions.
1626 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1628         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
1629         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
1630         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
1631         attribute to "z196_alone".
1632         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
1633         "zEC12_simple".
1635 2013-09-06  Richard Biener  <rguenther@suse.de>
1637         * basic-block.h (class control_dependences): New.
1638         * tree-ssa-dce.c (control_dependence_map): Remove.
1639         (cd): New global.
1640         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
1641         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
1642         find_pdom, find_control_dependence, find_all_control_dependences):
1643         Move to cfganal.c.
1644         (mark_control_dependent_edges_necessary,
1645         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
1646         tree_dce_done, perform_tree_ssa_dce): Adjust.
1647         * cfganal.c (set_control_dependence_map_bit,
1648         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
1649         find_all_control_dependences): Move from tree-ssa-dce.c and
1650         implement as methods of control_dependences class.
1651         (control_dependences::control_dependences): New.
1652         (control_dependences::~control_dependences): Likewise.
1653         (control_dependences::get_edges_dependent_on): Likewise.
1654         (control_dependences::get_edge): Likewise.
1656 2013-09-04  Jan Hubicka  <jh@suse.cz>
1658         * tree.c (types_same_for_odr): Drop overactive check.
1659         * ipa-devirt.c (hash_type_name): Likewise.
1661 2013-09-04  Jan Hubicka  <jh@suse.cz>
1663         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
1664         (analyze_functions): ... here.
1666 2013-09-04  Jan Hubicka  <jh@suse.cz>
1668         PR middle-end/58201
1669         * cgraphunit.c (analyze_functions): Clear AUX fields
1670         after processing; initialize assembler name has.
1672 2013-09-05  Jeff Law  <law@redhat.com>
1674         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
1675         from thread_around_empty_block.  Record threading path into PATH.
1676         Recurse if threading through the initial block is successful.
1677         (thread_across_edge): Corresponding changes to slightly simplify.
1679 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
1681         * config/aarch64/aarch64.md
1682         (type): Remove frecpe, frecps, frecpx.
1683         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
1684         fix to be a TARGET_SIMD instruction.
1685         (aarch64_frecps): Remove.
1686         * config/aarch64/aarch64-simd.md
1687         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
1688         (aarch64_frecps<mode>): Handle all float/vector of float modes.
1690 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
1691             Sofiane Naci  <sofiane.naci@arm.com>
1693         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
1694         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
1695         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
1696         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
1697         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
1698         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
1699         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
1700         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
1701         into "clz, "rbit".  Rename "shift" to "shift_imm".
1702         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
1703         changes.  Update for attribute changes all occurrences of arlo_* and
1704         shift* types.
1705         * config/arm/arm-fixed.md: Update for attribute changes
1706         all occurrences of arlo_* types.
1707         * config/arm/thumb2.md: Update for attribute changes all occurrences
1708         of arlo_* types.
1709         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
1710         (cortexa7_older_only): Likewise.
1711         (cortexa7_younger):  Likewise.
1712         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
1713         (1020alu_shift_op): Likewise.
1714         (1020alu_shift_reg_op): Likewise.
1715         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
1716         (alu_shift_op): Likewise.
1717         (alu_shift_reg_op): Likewise.
1718         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
1719         (11_alu_shift_op): Likewise.
1720         (11_alu_shift_reg_op): Likewise.
1721         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
1722         (9_alu_shift_reg_op): Likewise.
1723         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
1724         attribute changes.
1725         (cortex_a15_alu_shift): Likewise.
1726         (cortex_a15_alu_shift_reg): Likewise.
1727         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
1728         attribute changes.
1729         (cortex_a5_alu_shift): Likewise.
1730         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
1731         attribute changes.
1732         (cortex_a53_alu_shift): Likewise.
1733         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
1734         attribute changes.
1735         (cortex_a7_alu_reg): Likewise.
1736         (cortex_a7_alu_shift): Likewise.
1737         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
1738         attribute changes.
1739         (cortex_a8_alu_shift): Likewise.
1740         (cortex_a8_alu_shift_reg): Likewise.
1741         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
1742         (cortex_a9_dp_shift): Likewise.
1743         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
1744         attribute changes.
1745         * config/arm/cortex-r4.md
1746         (cortex_r4_alu): Update for attribute changes.
1747         (cortex_r4_mov): Likewise.
1748         (cortex_r4_alu_shift_reg): Likewise.
1749         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
1750         (526_alu_shift_op): Likewise.
1751         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
1752         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
1753         (626te_alu_shift_op): Likewise.
1754         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
1755         (726te_alu_shift_op): Likewise.
1756         (726te_alu_shift_reg_op): Likewise.
1757         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
1758         (mp626_alu_shift_op): Likewise.
1759         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
1760         (pj4_alu_conds): Likewise.
1761         (pj4_shift): Likewise.
1762         (pj4_shift_conds): Likewise.
1763         (pj4_alu_shift): Likewise.
1764         (pj4_alu_shift_conds): Likewise.
1765         * config/aarch64/aarch64.md: Update for attribute change
1766         all occurrences of arlo_* and shift* types.
1768 2013-09-05  Mike Stump  <mikestump@comcast.net>
1770         * tree.h: Move documentation for tree_function_decl to tree-core.h
1771         with the declaration.
1773 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
1775         PR target/58139
1776         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
1777         looking for widest mode.
1779 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
1781         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
1783 2013-09-05  Richard Biener  <rguenther@suse.de>
1785         PR tree-optimization/58137
1786         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
1787         Do not create vectors of pointers.
1788         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
1789         types for the components of the vector initializer.
1790         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
1791         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
1793 2013-09-05  Martin Jambor  <mjambor@suse.cz>
1795         * ipa-prop.c (remove_described_reference): Accept missing references,
1796         return false if that hppens, otherwise return true.
1797         (cgraph_node_for_jfunc): New function.
1798         (try_decrement_rdesc_refcount): Likewise.
1799         (try_make_edge_direct_simple_call): Use them.
1800         (ipa_edge_removal_hook): Remove references from rdescs.
1801         (ipa_edge_duplication_hook): Clone rdescs and their references
1802         when the new edge has the same caller as the old one.
1803         * cgraph.c (cgraph_resolve_speculation): Remove speculative
1804         reference before removing any edges.
1806 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
1808         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
1809         initial store.
1810         * thumb2.md (thumb2_storewb_parisi): New pattern.
1812 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
1814         * config/aarch64/aarch64-option-extensions.def: Add
1815         AARCH64_OPT_EXTENSION of 'crc'.
1816         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
1817         (AARCH64_ISA_CRC): Ditto.
1818         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
1819         description of the CRC extension.
1821 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1823         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
1824         * config/rs6000/linux.h: Ditto.
1825         * alpha/linux.h: Ditto.
1826         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
1827         no_c99_libc_has_function.
1828         * config/c6x/uclinux-elf.h: Ditto.
1829         * config/lm32/uclinux-elf.h: Ditto.
1830         * config/m68k/uclinux.h: Ditto.
1831         * config/moxie/uclinux.h: Ditto.
1832         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
1833         (crisv32-*-linux*, cris-*-linux*): Ditto.
1834         * config/bfin/bfin.c: Include "tm_p.h".
1836 2013-09-05  Richard Biener  <rguenther@suse.de>
1838         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
1839         check for a definition without a basic-block.
1841 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
1842             Sofiane Naci <sofiane.naci@arm.com>
1844         * config/aarch64/aarch64.md
1845         (*movti_aarch64): Rename r_2_f and f_2_r.
1846         (*movsf_aarch64): Likewise.
1847         (*movdf_aarch64): Likewise.
1848         (*movtf_aarch64): Likewise.
1849         (aarch64_movdi_<mode>low): Likewise.
1850         (aarch64_movdi_<mode>high): Likewise.
1851         (aarch64_mov<mode>high_di): Likewise.
1852         (aarch64_mov<mode>low_di): Likewise.
1853         (aarch64_movtilow_tilow): Likewise.
1854         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
1855         values to config/arm/types.md
1856         (attribute "conds"): Update for attribute change.
1857         (anddi3_insn): Likewise.
1858         (iordi3_insn): Likewise.
1859         (xordi3_insn): Likewise.
1860         (one_cmpldi2): Likewise.
1861         * config/arm/types.md (type): Add Neon types.
1862         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
1863         use "type" attribute.
1864         (movmisalign<mode>_neon_store): Likewise.
1865         (movmisalign<mode>_neon_load): Likewise.
1866         (vec_set<mode>_internal): Likewise.
1867         (vec_setv2di_internal): Likewise.
1868         (vec_extract<mode>): Likewise.
1869         (vec_extractv2di): Likewise.
1870         (add<mode>3_neon): Likewise.
1871         (adddi3_neon): Likewise.
1872         (sub<mode>3_neon): Likewise.
1873         (subdi3_neon): Likewise.
1874         (mul<mode>3_neon): Likewise.
1875         (mul<mode>3add<mode>_neon): Likewise.
1876         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
1877         (fma<VCVTF:mode>4)): Likewise.
1878         (fma<VCVTF:mode>4_intrinsic): Likewise.
1879         (fmsub<VCVTF:mode>4)): Likewise.
1880         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
1881         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
1882         (ior<mode>3): Likewise.
1883         (and<mode>3): Likewise.
1884         (anddi3_neon): Likewise.
1885         (orn<mode>3_neon): Likewise.
1886         (orndi3_neon): Likewise.
1887         (bic<mode>3_neon): Likewise.
1888         (bicdi3_neon): Likewise.
1889         (xor<mode>3): Likewise.
1890         (one_cmpl<mode>2): Likewise.
1891         (abs<mode>2): Likewise.
1892         (neg<mode>2): Likewise.
1893         (umin<mode>3_neon): Likewise.
1894         (umax<mode>3_neon): Likewise.
1895         (smin<mode>3_neon): Likewise.
1896         (smax<mode>3_neon): Likewise.
1897         (vashl<mode>3): Likewise.
1898         (vashr<mode>3_imm): Likewise.
1899         (vlshr<mode>3_imm): Likewise.
1900         (ashl<mode>3_signed): Likewise.
1901         (ashl<mode>3_unsigned): Likewise.
1902         (neon_load_count): Likewise.
1903         (ashldi3_neon_noclobber): Likewise.
1904         (signed_shift_di3_neon): Likewise.
1905         (unsigned_shift_di3_neon): Likewise.
1906         (ashrdi3_neon_imm_noclobber): Likewise.
1907         (lshrdi3_neon_imm_noclobber): Likewise.
1908         (widen_ssum<mode>3): Likewise.
1909         (widen_usum<mode>3): Likewise.
1910         (quad_halves_<code>v4si): Likewise.
1911         (quad_halves_<code>v4sf): Likewise.
1912         (quad_halves_<code>v8hi): Likewise.
1913         (quad_halves_<code>v16qi): Likewise.
1914         (reduc_splus_v2di): Likewise.
1915         (neon_vpadd_internal<mode>): Likewise.
1916         (neon_vpsmin<mode>): Likewise.
1917         (neon_vpsmax<mode>): Likewise.
1918         (neon_vpumin<mode>): Likewise.
1919         (neon_vpumax<mode>): Likewise.
1920         (ss_add<mode>_neon): Likewise.
1921         (us_add<mode>_neon): Likewise.
1922         (ss_sub<mode>_neon): Likewise.
1923         (us_sub<mode>_neon): Likewise.
1924         (neon_vadd<mode>_unspec): Likewise.
1925         (neon_vaddl<mode>): Likewise.
1926         (neon_vaddw<mode>): Likewise.
1927         (neon_vhadd<mode>): Likewise.
1928         (neon_vqadd<mode>): Likewise.
1929         (neon_vaddhn<mode>): Likewise.
1930         (neon_vmul<mode>): Likewise.
1931         (neon_vmla<mode>): Likewise.
1932         (neon_vmlal<mode>): Likewise.
1933         (neon_vmls<mode>): Likewise.
1934         (neon_vmlsl<mode>): Likewise.
1935         (neon_vqdmulh<mode>): Likewise.
1936         (neon_vqdmlal<mode>): Likewise.
1937         (neon_vqdmlsl<mode>): Likewise.
1938         (neon_vmull<mode>): Likewise.
1939         (neon_vqdmull<mode>): Likewise.
1940         (neon_vsub<mode>_unspec): Likewise.
1941         (neon_vsubl<mode>): Likewise.
1942         (neon_vsubw<mode>): Likewise.
1943         (neon_vqsub<mode>): Likewise.
1944         (neon_vhsub<mode>): Likewise.
1945         (neon_vsubhn<mode>): Likewise.
1946         (neon_vceq<mode>): Likewise.
1947         (neon_vcge<mode>): Likewise.
1948         (neon_vcgeu<mode>): Likewise.
1949         (neon_vcgt<mode>): Likewise.
1950         (neon_vcgtu<mode>): Likewise.
1951         (neon_vcle<mode>): Likewise.
1952         (neon_vclt<mode>): Likewise.
1953         (neon_vcage<mode>): Likewise.
1954         (neon_vcagt<mode>): Likewise.
1955         (neon_vtst<mode>): Likewise.
1956         (neon_vabd<mode>): Likewise.
1957         (neon_vabdl<mode>): Likewise.
1958         (neon_vaba<mode>): Likewise.
1959         (neon_vabal<mode>): Likewise.
1960         (neon_vmax<mode>): Likewise.
1961         (neon_vmin<mode>): Likewise.
1962         (neon_vpaddl<mode>): Likewise.
1963         (neon_vpadal<mode>): Likewise.
1964         (neon_vpmax<mode>): Likewise.
1965         (neon_vpmin<mode>): Likewise.
1966         (neon_vrecps<mode>): Likewise.
1967         (neon_vrsqrts<mode>): Likewise.
1968         (neon_vqabs<mode>): Likewise.
1969         (neon_vqneg<mode>): Likewise.
1970         (neon_vcls<mode>): Likewise.
1971         (clz<mode>2): Likewise.
1972         (popcount<mode>2): Likewise.
1973         (neon_vrecpe): Likewise.
1974         (neon_vrsqrte): Likewise.
1975         (neon_vget_lane<mode>_sext_internal): Likewise.
1976         (neon_vget_lane<mode>_zext_internal): Likewise.
1977         (neon_vdup_n<mode>): Likewise.
1978         (neon_vdup_nv2di): Likewise.
1979         (neon_vdpu_lane<mode>_internal): Likewise.
1980         (neon_vswp<mode>): Likewise.
1981         (float<mode><V_cvtto>2): Likewise.
1982         (floatuns<mode><V_cvtto>2): Likewise.
1983         (fix_trunc<mode><V_cvtto>)2): Likewise
1984         (fixuns_trunc<mode><V_cvtto)2): Likewise.
1985         (neon_vcvt<mode>): Likewise.
1986         (neon_vcvtv4sfv4hf): Likewise.
1987         (neon_vcvtv4hfv4sf): Likewise.
1988         (neon_vcvt_n<mode>): Likewise.
1989         (neon_vmovn<mode>): Likewise.
1990         (neon_vqmovn<mode>): Likewise.
1991         (neon_vqmovun<mode>): Likewise.
1992         (neon_vmovl<mode>): Likewise.
1993         (neon_vmul_lane<mode>): Likewise.
1994         (neon_vmull_lane<mode>): Likewise.
1995         (neon_vqdmull_lane<mode>): Likewise.
1996         (neon_vqdmulh_lane<mode>): Likewise.
1997         (neon_vmla_lane<mode>): Likewise.
1998         (neon_vmlal_lane<mode>): Likewise.
1999         (neon_vqdmlal_lane<mode>): Likewise.
2000         (neon_vmls_lane<mode>): Likewise.
2001         (neon_vmlsl_lane<mode>): Likewise.
2002         (neon_vqdmlsl_lane<mode>): Likewise.
2003         (neon_vext<mode>): Likewise.
2004         (neon_vrev64<mode>): Likewise.
2005         (neon_vrev32<mode>): Likewise.
2006         (neon_vrev16<mode>): Likewise.
2007         (neon_vbsl<mode>_internal): Likewise.
2008         (neon_vshl<mode>): Likewise.
2009         (neon_vqshl<mode>): Likewise.
2010         (neon_vshr_n<mode>): Likewise.
2011         (neon_vshrn_n<mode>): Likewise.
2012         (neon_vqshrn_n<mode>): Likewise.
2013         (neon_vqshrun_n<mode>): Likewise.
2014         (neon_vshl_n<mode>): Likewise.
2015         (neon_vqshl_n<mode>): Likewise.
2016         (neon_vqshlu_n<mode>): Likewise.
2017         (neon_vshll_n<mode>): Likewise.
2018         (neon_vsra_n<mode>): Likewise.
2019         (neon_vsri_n<mode>): Likewise.
2020         (neon_vsli_n<mode>): Likewise.
2021         (neon_vtbl1v8qi): Likewise.
2022         (neon_vtbl2v8qi): Likewise.
2023         (neon_vtbl3v8qi): Likewise.
2024         (neon_vtbl4v8qi): Likewise.
2025         (neon_vtbx1v8qi): Likewise.
2026         (neon_vtbx2v8qi): Likewise.
2027         (neon_vtbx3v8qi): Likewise.
2028         (neon_vtbx4v8qi): Likewise.
2029         (neon_vtrn<mode>_internal): Likewise.
2030         (neon_vzip<mode>_internal): Likewise.
2031         (neon_vuzp<mode>_internal): Likewise.
2032         (neon_vld1<mode>): Likewise.
2033         (neon_vld1_lane<mode>): Likewise.
2034         (neon_vld1_dup<mode>): Likewise.
2035         (neon_vld1_dupv2di): Likewise.
2036         (neon_vst1<mode>): Likewise.
2037         (neon_vst1_lane<mode>): Likewise.
2038         (neon_vld2<mode>): Likewise.
2039         (neon_vld2_lane<mode>): Likewise.
2040         (neon_vld2_dup<mode>): Likewise.
2041         (neon_vst2<mode>): Likewise.
2042         (neon_vst2_lane<mode>): Likewise.
2043         (neon_vld3<mode>): Likewise.
2044         (neon_vld3qa<mode>): Likewise.
2045         (neon_vld3qb<mode>): Likewise.
2046         (neon_vld3_lane<mode>): Likewise.
2047         (neon_vld3_dup<mode>): Likewise.
2048         (neon_vst3<mode>): Likewise.
2049         (neon_vst3qa<mode>): Likewise.
2050         (neon_vst3qb<mode>): Likewise.
2051         (neon_vst3_lane<mode>): Likewise.
2052         (neon_vld4<mode>): Likewise.
2053         (neon_vld4qa<mode>): Likewise.
2054         (neon_vld4qb<mode>): Likewise.
2055         (neon_vld4_lane<mode>): Likewise.
2056         (neon_vld4_dup<mode>): Likewise.
2057         (neon_vst4<mode>): Likewise.
2058         (neon_vst4qa<mode>): Likewise.
2059         (neon_vst4qb<mode>): Likewise.
2060         (neon_vst4_lane<mode>): Likewise.
2061         (neon_vec_unpack<US>_lo_<mode>): Likewise.
2062         (neon_vec_unpack<US>_hi_<mode>): Likewise.
2063         (neon_vec_<US>mult_lo_<mode>): Likewise.
2064         (neon_vec_<US>mult_hi_<mode>): Likewise.
2065         (neon_vec_<US>shiftl_<mode>): Likewise.
2066         (neon_unpack<US>_<mode>): Likewise.
2067         (neon_vec_<US>mult_<mode>): Likewise.
2068         (vec_pack_trunc_<mode>): Likewise.
2069         (neon_vec_pack_trunk_<mode>): Likewise.
2070         (neon_vabd<mode>_2): Likewise.
2071         (neon_vabd<mode>_3): Likewise.
2072         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
2073         (thumb2_movsi_vfp): Likewise.
2074         (movdi_vfp): Likewise.
2075         (movdi_vfp_cortexa8): Likewise.
2076         (movhf_vfp_neon): Likewise.
2077         (movhf_vfp): Likewiwse.
2078         (movsf_vfp): Likewiwse.
2079         (thumb2_movsf_vfp): Likewiwse.
2080         (movdf_vfp): Likewise.
2081         (thumb2_movdf_vfp): Likewise.
2082         (movsfcc_vfp): Likewise.
2083         (thumb2_movsfcc_vfp): Likewise.
2084         (movdfcc_vfp): Likewise.
2085         (thumb2_movdfcc_vfp): Likewise.
2086         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
2087         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
2088         (v10_v2c): Likewise.
2089         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
2090         attribute change.
2091         (cortex_a15_neon_int_2): Likewise.
2092         (cortex_a15_neon_int_3): Likewise.
2093         (cortex_a15_neon_int_4): Likewise.
2094         (cortex_a15_neon_int_5): Likewise.
2095         (cortex_a15_neon_vqneg_vqabs): Likewise.
2096         (cortex_a15_neon_vmov): Likewise.
2097         (cortex_a15_neon_vaba): Likewise.
2098         (cortex_a15_neon_vaba_qqq): Likewise.
2099         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2100         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
2101         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
2102         scalar_64_32_long_scalar): Likewise.
2103         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2104         (cortex_a15_neon_mla_qqq_8_16): Likewise.
2105         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
2106         lotype_qdd_64_32_long): Likewise.
2107         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
2108         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
2109         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
2110         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
2111         (cortex_a15_neon_shift_1): Likewise.
2112         (cortex_a15_neon_shift_2): Likewise.
2113         (cortex_a15_neon_shift_3): Likewise.
2114         (cortex_a15_neon_vshl_ddd): Likewise.
2115         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
2116         (cortex_a15_neon_vsra_vrsra): Likewise.
2117         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
2118         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
2119         (cortex_a15_neon_fp_vmul_ddd): Likewise.
2120         (cortex_a15_neon_fp_vmul_qqd): Likewise.
2121         (cortex_a15_neon_fp_vmla_ddd): Likewise.
2122         (cortex_a15_neon_fp_vmla_qqq): Likewise.
2123         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
2124         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
2125         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
2126         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
2127         (cortex_a15_neon_bp_simple): Likewise.
2128         (cortex_a15_neon_bp_2cycle): Likewise.
2129         (cortex_a15_neon_bp_3cycle): Likewise.
2130         (cortex_a15_neon_vld1_1_2_regs): Likewise.
2131         (cortex_a15_neon_vld1_3_4_regs): Likewise.
2132         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
2133         (cortex_a15_neon_vld2_4_regs): Likewise.
2134         (cortex_a15_neon_vld3_vld4): Likewise.
2135         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
2136         (cortex_a15_neon_vst1_3_4_regs): Likewise.
2137         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
2138         (cortex_a15_neon_vst3_vst4): Likewise.
2139         (cortex_a15_neon_vld1_vld2_lane): Likewise.
2140         (cortex_a15_neon_vld3_vld4_lane" 10
2141         (cortex_a15_neon_vst1_vst2_lane): Likewise.
2142         (cortex_a15_neon_vst3_vst4_lane): Likewise.
2143         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
2144         (cortex_a15_neon_ldm_2): Likewise.0
2145         (cortex_a15_neon_stm_2): Likewise.
2146         (cortex_a15_neon_mcr): Likewise.
2147         (cortex_a15_neon_mcr_2_mcrr): Likewise.
2148         (cortex_a15_neon_mrc): Likewise.
2149         (cortex_a15_neon_mrrc): Likewise.
2150         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
2151         change.
2152         (cortex_a15_alu_shift): Likewise.
2153         (cortex_a15_alu_shift_reg): Likewise.
2154         (cortex_a15_mult32): Likewise.
2155         (cortex_a15_mult64): Likewise.
2156         (cortex_a15_block): Likewise.
2157         (cortex_a15_branch): Likewise.
2158         (cortex_a15_load1): Likewise.
2159         (cortex_a15_load3): Likewise.
2160         (cortex_a15_store1): Likewise.
2161         (cortex_a15_store3): Likewise.
2162         (cortex_a15_call): Likewise.
2163         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
2164         (cortex_a5_f2r): Likewise.
2165         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
2166         change.
2167         (cortex_a53_f2r): Likewise.
2168         * config/arm/cortex-a7.md
2169         (cortex_a7_branch): Update for attribute change.
2170         (cortex_a7_call): Likewise.
2171         (cortex_a7_alu_imm): Likewise.
2172         (cortex_a7_alu_reg): Likewise.
2173         (cortex_a7_alu_shift): Likewise.
2174         (cortex_a7_mul): Likewise.
2175         (cortex_a7_load1): Likewise.
2176         (cortex_a7_store1): Likewise.
2177         (cortex_a7_load2): Likewise.
2178         (cortex_a7_store2): Likewise.
2179         (cortex_a7_load3): Likewise.
2180         (cortex_a7_store3): Likewise.
2181         (cortex_a7_load4): Likewise.
2182         (cortex_a7_store4): Likewise.
2183         (cortex_a7_fpalu): Likewise.
2184         (cortex_a7_fconst): Likewise.
2185         (cortex_a7_fpmuls): Likewise.
2186         (cortex_a7_neon_mul): Likewise.
2187         (cortex_a7_fpmacs): Likewise.
2188         (cortex_a7_neon_mla: Likewise.
2189         (cortex_a7_fpmuld: Likewise.
2190         (cortex_a7_fpmacd: Likewise.
2191         (cortex_a7_fpfmad: Likewise.
2192         (cortex_a7_fdivs: Likewise.
2193         (cortex_a7_fdivd: Likewise.
2194         (cortex_a7_r2f: Likewise.
2195         (cortex_a7_f2r: Likewise.
2196         (cortex_a7_f_flags: Likewise.
2197         (cortex_a7_f_loads: Likewise.
2198         (cortex_a7_f_loadd: Likewise.
2199         (cortex_a7_f_stores: Likewise.
2200         (cortex_a7_f_stored: Likewise.
2201         (cortex_a7_neon): Likewise.
2202         * config/arm/cortex-a8-neon.md
2203         (cortex_a8_neon_mrc): Update for attribute change.
2204         (cortex_a8_neon_mrrc): Likewise.
2205         (cortex_a8_neon_int_1): Likewise.
2206         (cortex_a8_neon_int_2): Likewise.
2207         (cortex_a8_neon_int_3): Likewise.
2208         (cortex_a8_neon_int_4): Likewise.
2209         (cortex_a8_neon_int_5): Likewise.
2210         (cortex_a8_neon_vqneg_vqabs): Likewise.
2211         (cortex_a8_neon_vmov): Likewise.
2212         (cortex_a8_neon_vaba): Likewise.
2213         (cortex_a8_neon_vaba_qqq): Likewise.
2214         (cortex_a8_neon_vsma): Likewise.
2215         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2216         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
2217         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
2218         long_scalar): Likewise.
2219         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2220         (cortex_a8_neon_mla_qqq_8_16): Likewise.
2221         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
2222         long_scalar_qdd_64_32_long): Likewise.
2223         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
2224         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
2225         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
2226         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
2227         (cortex_a8_neon_shift_1): Likewise.
2228         (cortex_a8_neon_shift_2): Likewise.
2229         (cortex_a8_neon_shift_3): Likewise.
2230         (cortex_a8_neon_vshl_ddd): Likewise.
2231         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
2232         (cortex_a8_neon_vsra_vrsra): Likewise.
2233         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
2234         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
2235         (cortex_a8_neon_fp_vsum): Likewise.
2236         (cortex_a8_neon_fp_vmul_ddd): Likewise.
2237         (cortex_a8_neon_fp_vmul_qqd): Likewise.
2238         (cortex_a8_neon_fp_vmla_ddd): Likewise.
2239         (cortex_a8_neon_fp_vmla_qqq): Likewise.
2240         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
2241         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
2242         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
2243         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
2244         (cortex_a8_neon_bp_simple): Likewise.
2245         (cortex_a8_neon_bp_2cycle): Likewise.
2246         (cortex_a8_neon_bp_3cycle): Likewise.
2247         (cortex_a8_neon_ldr): Likewise.
2248         (cortex_a8_neon_str): Likewise.
2249         (cortex_a8_neon_vld1_1_2_regs): Likewise.
2250         (cortex_a8_neon_vld1_3_4_regs): Likewise.
2251         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
2252         (cortex_a8_neon_vld2_4_regs): Likewise.
2253         (cortex_a8_neon_vld3_vld4): Likewise.
2254         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
2255         (cortex_a8_neon_vst1_3_4_regs): Likewise.
2256         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
2257         (cortex_a8_neon_vst3_vst4): Likewise.
2258         (cortex_a8_neon_vld1_vld2_lane): Likewise.
2259         (cortex_a8_neon_vld3_vld4_lane): Likewise.
2260         (cortex_a8_neon_vst1_vst2_lane): Likewise.
2261         (cortex_a8_neon_vst3_vst4_lane): Likewise.
2262         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
2263         (cortex_a8_neon_mcr): Likewise.
2264         (cortex_a8_neon_mcr_2_mcrr): Likewise.
2265         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
2266         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
2267         change.
2268         (ca9_neon_mrrc): Likewise.
2269         (cortex_a9_neon_int_1): Likewise.
2270         (cortex_a9_neon_int_2): Likewise.
2271         (cortex_a9_neon_int_3): Likewise.
2272         (cortex_a9_neon_int_4): Likewise.
2273         (cortex_a9_neon_int_5): Likewise.
2274         (cortex_a9_neon_vqneg_vqabs): Likewise.
2275         (cortex_a9_neon_vmov): Likewise.
2276         (cortex_a9_neon_vaba): Likewise.
2277         (cortex_a9_neon_vaba_qqq): Likewise.
2278         (cortex_a9_neon_vsma): Likewise.
2279         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2280         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
2281         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
2282         long_scalar): Likewise.
2283         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2284         (cortex_a9_neon_mla_qqq_8_16): Likewise.
2285         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
2286         long_scalar_qdd_64_32_long): Likewise.
2287         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
2288         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
2289         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
2290         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
2291         (cortex_a9_neon_shift_1): Likewise.
2292         (cortex_a9_neon_shift_2): Likewise.
2293         (cortex_a9_neon_shift_3): Likewise.
2294         (cortex_a9_neon_vshl_ddd): Likewise.
2295         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
2296         (cortex_a9_neon_vsra_vrsra): Likewise.
2297         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
2298         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
2299         (cortex_a9_neon_fp_vsum): Likewise.
2300         (cortex_a9_neon_fp_vmul_ddd): Likewise.
2301         (cortex_a9_neon_fp_vmul_qqd): Likewise.
2302         (cortex_a9_neon_fp_vmla_ddd): Likewise.
2303         (cortex_a9_neon_fp_vmla_qqq): Likewise.
2304         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
2305         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
2306         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
2307         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
2308         (cortex_a9_neon_bp_simple): Likewise.
2309         (cortex_a9_neon_bp_2cycle): Likewise.
2310         (cortex_a9_neon_bp_3cycle): Likewise.
2311         (cortex_a9_neon_ldr): Likewise.
2312         (cortex_a9_neon_str): Likewise.
2313         (cortex_a9_neon_vld1_1_2_regs): Likewise.
2314         (cortex_a9_neon_vld1_3_4_regs): Likewise.
2315         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
2316         (cortex_a9_neon_vld2_4_regs): Likewise.
2317         (cortex_a9_neon_vld3_vld4): Likewise.
2318         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
2319         (cortex_a9_neon_vst1_3_4_regs): Likewise.
2320         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
2321         (cortex_a9_neon_vst3_vst4): Likewise.
2322         (cortex_a9_neon_vld1_vld2_lane): Likewise.
2323         (cortex_a9_neon_vld3_vld4_lane): Likewise.
2324         (cortex_a9_neon_vst1_vst2_lane): Likewise.
2325         (cortex_a9_neon_vst3_vst4_lane): Likewise.
2326         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
2327         (cortex_a9_neon_mcr): Likewise.
2328         (cortex_a9_neon_mcr_2_mcrr): Likewise.
2329         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
2330         (cortex_a9_fps): Likewise.
2331         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
2332         change.
2333         (cortex_m4_fmuls): Likewise.
2334         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
2335         change.
2336         (cortex_r4_mrc): Likewise.
2337         * config/arm/iterators.md: Update comment referring to neon_type.
2338         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
2339         (iwmmxt_movsi_insn): Likewise.
2340         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
2341         attribute change.
2342         (pj4_core_to_vfp): Likewise.
2343         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
2344         attribute change.
2345         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
2346         (vfp_fstore): Likewise.
2347         * doc/md.texi: Change references to neon_type to refer to type.
2349 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
2351         * tree.h (DECL_BUILT_IN): Fix typo in comment.
2353 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
2355         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
2356         lglobl if not weak.
2358 2013-09-04  Easwaran Raman  <eraman@google.com>
2360         PR middle-end/57370
2361         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
2362         (build_and_add_sum): Use it.
2363         (appears_later_in_bb): Simplify code.
2365 2013-09-04  Teresa Johnson  <tejohnson@google.com>
2367         * dumpfile.c (dump_finish): Don't close stderr/stdout.
2369 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
2371         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
2373 2013-09-04  Jan Hubicka  <jh@suse.cz>
2375         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
2376         * ipa-devirt.c: Include diganostic.h
2377         (odr_type_d): Add types and types_set.
2378         (hash_type_name): Work for types with vtables during LTO.
2379         (odr_hasher::remove): Fix comment; destroy types_set.
2380         (add_type_duplicate): New function,
2381         (get_odr_type): Use it.
2382         (dump_type_inheritance_graph): Dump type duplicates.
2383         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
2384         graph.
2385         * tree.c (types_same_for_odr): Give exact answers on types with
2386         virtual tables.
2388 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
2390         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
2391         explaining their differences.
2393 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
2395         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
2397 2013-09-03  Jeff Law  <law@redhat.com>
2399         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
2400         when not threading through a joiner block.  Pass joiner/no joiner
2401         state to register_jump_thread.
2402         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
2403         state from argument rather than implying on path length.
2404         Dump the entire jump thread path into debugging dump.
2405         * tree-flow.h (register_jump_thread): Update prototype.
2407 2013-08-29  Xinliang David Li  <davidxl@google.com>
2409         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
2410         Remove a trivial gcc_assert.
2412 2013-08-29  Xinliang David Li  <davidxl@google.com>
2414         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
2415         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
2416         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
2417         Delay base decl alignment adjustment.
2418         * tree-vectorizer.c (vect_destroy_datarefs): New function.
2419         * tree-vectorizer.h: New data structure.
2420         (set_dr_misalignment): New function.
2421         (dr_misalignment): Ditto.
2422         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
2423         (vectorizable_load): Ditto.
2424         (ensure_base_align): New function.
2425         (vectorize_loops): Add dbg_cnt support.
2426         (execute_vect_slp): Ditto.
2427         * dbgcnt.def: New debug counter.
2428         * Makefile: New dependency.
2430 2013-09-03  Meador Inge  <meadori@codesourcery.com>
2432         Revert:
2434         2013-08-30  Meador Inge  <meadori@codesourcery.com>
2436         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
2438 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
2440         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
2441         function descriptor.
2443 2013-09-03  Richard Biener  <rguenther@suse.de>
2445         * tree-affine.c (add_elt_to_tree): Fix association issue,
2446         avoid useless converts and make sure to always return a
2447         properly typed result.
2449 2013-09-03  Richard Biener  <rguenther@suse.de>
2451         PR middle-end/57656
2452         * fold-const.c (negate_expr_p): Fix division case.
2453         (negate_expr): Likewise.
2455 2013-09-03  Richard Biener  <rguenther@suse.de>
2457         PR lto/58285
2458         * tree-streamer-out.c: Include tm.h.
2459         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
2461 2013-09-03  Jan Hubicka  <jh@suse.cz>
2463         * tree-profile.c (tree_profiling): Cleanup CFG when done.
2465 2013-09-03  Alan Modra  <amodra@gmail.com>
2467         * config.gcc (powerpc*-*-linux*): Add support for little-endian
2468         multilibs to big-endian target and vice versa.
2469         * config/rs6000/t-linux64: Use := assignment on all vars.
2470         (MULTILIB_EXTRA_OPTS): Remove fPIC.
2471         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
2472         * config/rs6000/t-linux64le: New file.
2473         * config/rs6000/t-linux64bele: New file.
2474         * config/rs6000/t-linux64lebe: New file.
2476 2013-09-02  Jan Hubicka  <jh@suse.cz>
2478         * ipa-inline-transform.c (inline_transform): Do not
2479         optimize_inline_calls when not optimizing.
2481 2013-09-02  Jan Hubicka  <jh@suse.cz>
2483         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
2484         duplicated nodes for assembler names.
2485         * symtab.c (symtab_unregister_node): Do not attempt to unlink
2486         hard registers from assembler name hash.
2488 2013-09-02  Jan Hubicka  <jh@suse.cz>
2490         * ipa-split.c (execute_split_functions): Split externally visible
2491         functions called once.
2493 2013-09-02  Martin Jambor  <mjambor@suse.cz>
2495         PR ipa/58106
2496         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
2497         linked list.  When finding the correct duplicate, also consider also
2498         the caller in additon to its inlined_to node.
2500 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
2502         * config/aarch64/aarch64-simd-builtins.def
2503         (dup_lane_scalar): Remove.
2504         * config/aarch64/aarch64-simd.md
2505         (aarch64_simd_dup): Add 'w->w' alternative.
2506         (aarch64_dup_lane<mode>): Allow for VALL.
2507         (aarch64_dup_lane_scalar<mode>): Remove.
2508         (aarch64_dup_lane_<vswap_width_name><mode>): New.
2509         (aarch64_get_lane_signed<mode>): Add w->w altenative.
2510         (aarch64_get_lane_unsigned<mode>): Likewise.
2511         (aarch64_get_lane<mode>): Likewise.
2512         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
2513         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
2514         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
2515         (VCON): Change container of V2SF.
2516         (vswap_width_name): Likewise.
2517         * config/aarch64/arm_neon.h
2518         (__aarch64_vdup_lane_any): New.
2519         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
2520         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
2521         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
2523 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
2525         PR middle-end/56382
2526         * expr.c (emit_move_complex): Do not move complex FP values as parts if
2527         the source or the destination is a single hard register.
2529 2013-09-02  Richard Biener  <rguenther@suse.de>
2531         PR middle-end/57511
2532         * tree-scalar-evolution.c (instantiate_scev_name): Allow
2533         non-linear SCEVs.
2535 2013-09-02  Richard Biener  <rguenther@suse.de>
2537         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
2538         arithmetic to sizetype.
2540 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
2542         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
2543         Find auto-increment use both before and after candidate.
2545 2013-09-02  Marek Polacek  <polacek@redhat.com>
2547         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
2549 2013-09-01  Jan Hubicka  <jh@suse.cz>
2551         * Makefile.in: Add ipa-profile.o
2552         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
2553         * cgraph.c (struct cgraph_propagate_frequency_data,
2554         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
2555         ipa-profile.c; replace cgraph_ by ipa_ prefix.
2556         * cgraph.h (cgraph_propagate_frequency): Remove.
2557         * ipa-inline-analysis.c: Include ipa-utils.h;
2558         drop duplicated cfgloop.h.
2559         (inline_update_callee_summaries): Update.
2560         * ipa-profile.c: New file.
2561         * ipa-utils.h (ipa_propagate_frequency): Declare.
2562         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
2563         data-streamer.h, value-prof.h.
2564         (symtab_remove_unreachable_nodes): Update profile.
2565         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
2566         account_time_size, cmp_counts, dump_histogram,
2567         ipa_profile_generate_summary, ipa_profile_write_summary,
2568         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
2569         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
2570         Move to ipa-profile.c.
2572 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
2574         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
2576 2013-09-01  Jan Hubicka  <jh@suse.cz>
2578         * common.opt (fdevirtualize-speculatively): New function.
2579         * invoke.texi (fdevirtualize-speculatively): Document.
2580         * ipa-devirt.c: Include ipa-inline.h
2581         (likely_target_p): New function.
2582         (ipa_devirt): New function.
2583         (gate_ipa_devirt): New function.
2584         (pass_data_ipa_devirt): New static var.
2585         (pass_ipa_devirt): Likewise.
2586         (make_pass_ipa_devirt): New function.
2587         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
2588         (common_handle_option): Disable devirtualization when
2589         value range profiling is available.
2590         * passes.def (pass_ipa_devirt): Add.
2591         * timever.def (TV_IPA_DEVIRT): New timevar.
2592         * tree-pass.h (make_pass_ipa_devirt):
2594 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
2596         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
2597         include sanitize(undefined).
2599 2013-08-31  Diego Novillo  <dnovillo@google.com>
2601         * Makefile.in (TREE_CORE_H): Define.
2602         (TREE_H): Use.
2603         (GTFILES): Add tree-core.h.
2604         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
2605         size the array.
2606         * tree-core.h: New file.
2607         Move all data structures, enum, typedefs, global
2608         declarations and constants from ...
2609         * tree.h: ... here.
2611 2013-08-31  Jan Hubicka  <jh@suse.cz>
2613         * bulitins.c (expand_builtin): Do not early exit for gcov
2614         instrumented functions.
2616 2013-08-31  Marek Polacek  <polacek@redhat.com>
2618         * ubsan.c: Include tm_p.h.
2620 2013-08-31  Jan Hubicka  <jh@suse.cz>
2622         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
2623         warning.
2625         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
2626         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
2627         * tree-cfg.c (verify_gimple_label): ... here.
2628         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
2629         (ipa_merge_profiles): New function.
2630         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
2631         (lto_input_function_body): Likewise.
2632         * ipa-utils.h (ipa_merge_profiles): Declare.
2633         * lto-streamer.h (lto_input_function_body): Update prototype.
2634         (emit_label_in_global_context_p): Remove.
2635         * lto-symtab.c: Include ipa-utils.h
2636         (lto_cgraph_replace_node): Use ipa_merge_profiles.
2638 2013-08-31  Jan Hubicka  <jh@suse.cz>
2640         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
2642 2013-08-31  Jan Hubicka  <jh@suse.cz>
2644         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
2646 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
2648         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
2649         "cmp" RTX before signed_comparison_operator check to account
2650         for "code" changes.
2652 2013-08-30  Jan Hubicka  <jh@suse.cz>
2654         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
2655         (detect_type_change_1): Rename to ...
2656         (detect_type_change): ... this one; early return on non-polymorphic
2657         types.
2658         (detect_type_change_ssa): Add comp_type parameter; update
2659         use of detect_type_change.
2660         (compute_complex_assign_jump_func): Add param_type parameter;
2661         update use of detect_type_change_ssa.
2662         (compute_complex_ancestor_jump_func): Likewise.
2663         (ipa_get_callee_param_type): New function.
2664         (ipa_compute_jump_functions_for_edge): Compute parameter type;
2665         update calls to the jump function computation functions.
2667 2013-08-30  Teresa Johnson  <tejohnson@google.com>
2668             Steven Bosscher  <steven@gcc.gnu.org>
2670         * cfgrtl.c (fixup_new_cold_bb): New routine.
2671         (commit_edge_insertions): Invoke fixup_partitions.
2672         (find_partition_fixes): New routine.
2673         (fixup_partitions): Ditto.
2674         (verify_hot_cold_block_grouping): Update comments.
2675         (rtl_verify_edges): Invoke find_partition_fixes.
2676         (rtl_verify_bb_pointers): Update comments.
2677         (rtl_verify_bb_layout): Ditto.
2678         * basic-block.h (probably_never_executed_edge_p): Declare.
2679         (fixup_partitions): Ditto.
2680         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
2681         * bb-reorder.c (sanitize_hot_paths): New function.
2682         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
2683         sanitize_hot_paths.
2684         * predict.c (probably_never_executed_edge_p): New routine.
2685         * cfg.c (check_bb_profile): Add partition insanity warnings.
2687 2013-08-30  Meador Inge  <meadori@codesourcery.com>
2689         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
2691 2013-08-30  Marek Polacek  <polacek@redhat.com>
2693         * Makefile.in (ubsan.o): Add.
2694         (c-family/c-ubsan.o): Add.
2695         (builtins.o): Add ubsan.h dependency.
2696         * ubsan.h: New file.
2697         * ubsan.c: New file.
2698         * common.opt: Add -fsanitize=undefined option.
2699         (flag_sanitize): Add variable.
2700         (fsanitize=): Add option.  Add Driver.
2701         (fsanitize=thread): Remove option.
2702         (fsanitize=address): Likewise.
2703         (static-libubsan): New option.
2704         * doc/invoke.texi: Document the new flag and -static-libubsan.
2705         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
2706         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
2707         * builtin-attrs.def (ATTR_COLD): Define.
2708         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
2709         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
2710         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
2711         * flag-types.h (sanitize_code): New enum.
2712         * opts.c (common_handle_option): Parse command line arguments
2713         of -fsanitize=.  Add -fsanitize=unreachable option.
2714         * varasm.c (get_variable_section): Adjust.
2715         (assemble_noswitch_variable): Likewise.
2716         (assemble_variable): Likewise.
2717         (output_constant_def_contents): Likewise.
2718         (categorize_decl_for_section): Likewise.
2719         (place_block_symbol): Likewise.
2720         (output_object_block): Likewise.
2721         * builtins.def: Likewise.
2722         * toplev.c (compile_file): Likewise.
2723         (process_options): Likewise.
2724         * cppbuiltin.c: Likewise.
2725         * tsan.c (tsan_pass): Likewise.
2726         (tsan_gate): Likewise.
2727         (tsan_gate_O0): Likewise.
2728         * cfgexpand.c (partition_stack_vars): Likewise.
2729         (expand_stack_vars): Likewise.
2730         (defer_stack_allocation): Likewise.
2731         (expand_used_vars): Likewise.
2732         * cfgcleanup.c (old_insns_match_p): Likewise.
2733         * asan.c (asan_finish_file): Likewise.
2734         (asan_instrument): Likewise.
2735         (gate_asan): Likewise.
2736         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
2737         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
2738         (asan_global_struct): Use pointer_sized_int_node instead
2739         calling build_nonstandard_integer_type.
2740         (initialize_sanitizer_builtins): Likewise.
2741         (asan_finish_file): Likewise.
2742         * gcc.c: Document %{%:function(args):X}.
2743         (static_spec_functions): Add sanitize.
2744         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
2745         store funcval != NULL there.
2746         (do_spec_1): Adjust handle_spec_function caller.
2747         (handle_braces): Allow %:function(args) as condition.
2748         (sanitize_spec_function): New function.
2749         (ADD_STATIC_LIBUBSAN_LIBS): Define.
2750         (LIBUBSAN_SPEC): Likewise.
2751         (LIBUBSAN_EARLY_SPEC): Likewise.
2752         (SANITIZER_SPEC): Handle libubsan.
2753         (SANITIZER_EARLY_SPEC): Likewise.
2754         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
2755         instead of fsanitize=address.
2756         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
2757         instead of fsanitize=address*.
2758         * builtins.c: Include ubsan.h.
2759         (fold_builtin_0): Instrument __builtin_unreachable.
2760         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
2761         instead of flag_asan.
2762         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
2763         (pointer_sized_int_node): Define.
2764         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
2766 2013-08-30  Mike Stump  <mikestump@comcast.net>
2768         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
2769         with RE patterns.
2771 2013-08-29  Jan Hubicka  <jh@suse.cz>
2773         * cgraph.c (cgraph_function_body_availability): Handle weakref
2774         correctly.
2775         * passes.def: Remove pass_fixup_cfg.
2776         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
2777         track when we need to remove functions.
2778         (gate_ipa_inline): Execute inlining always; add comment why.
2779         (pass_data_ipa_inline): Remove TODO_remove_functions.
2780         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
2781         do not produce summaries.
2782         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
2783         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
2784         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
2785         constant pool and vtable.
2787 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
2789         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
2790         New arm_neon.h's internal macros to specify 64-bit constants.
2791         Avoid using stdint.h's macros.
2793 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
2795         * recog.c (verify_changes): Verify that changes[i].old is non-zero
2796         before applying REG_P.
2798 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
2800         PR tree-optimization/58277
2801         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
2802         after seeing too many stmts with vdef in between dombb and current
2803         bb, invalidate everything.
2805 2013-08-30  Richard Biener  <rguenther@suse.de>
2807         * fold-const.c (fold_single_bit_test): Fix overflow test.
2809 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
2811         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
2812         and which can live in a register, always retrieve the value on entry.
2813         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
2814         passed by invisible reference specially.
2815         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
2816         (vt_add_function_parameter): Correctly deal with a parameter passed by
2817         invisible reference.
2819 2013-08-30  Jan Hubicka  <jh@suse.cz>
2821         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
2823 2013-08-30  Richard Biener  <rguenther@suse.de>
2825         PR tree-optimization/58228
2826         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
2827         allow invariant loads in nested loop vectorization.
2829 2013-08-30  Richard Biener  <rguenther@suse.de>
2831         PR tree-optimization/58223
2832         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
2833         (has_anti_or_output_dependence): ... this and adjust to also
2834         look for output dependences.
2835         (mark_nodes_having_upstream_mem_writes): Adjust.
2836         (rdg_flag_uses): Likewise.
2838 2013-08-30  Richard Biener  <rguenther@suse.de>
2840         PR tree-optimization/58010
2841         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
2842         assert that we have a loop-closed PHI.
2844 2013-08-29  Jan Hubicka  <jh@suse.cz>
2846         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
2847         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
2848         * lto-section-in.c (lto_free_function_in_decl_state): New function.
2849         (lto_free_function_in_decl_state_for_node): New function.
2851 2013-08-29  Xinliang David Li  <davidxl@google.com>
2853         * loop-unroll.c (report_unroll_peel): Minor message change.
2854         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
2855         Emit alignment peeling message with default -fopt-info.
2856         (vect_loop_versioning): Emit loop version info message.
2857         * tree-vectorizer.c (vectorize_loops): Minor message change.
2858         (execute_vect_slp): Ditto.
2860 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
2862         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
2863         of the clone from the DECL_NAME of the original function.
2865 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
2867         * passes.c (register_pass): Add overload.
2868         * tree-pass.h (register_pass): Forward declare it.  Add comment.
2870 2013-08-29  Jan Hubicka  <jh@suse.cz>
2872         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
2873         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
2874         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
2875         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
2876         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
2877         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
2878         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
2879         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
2881 2013-08-29  Teresa Johnson  <tejohnson@google.com>
2883         * dumpfile.c (dump_loc): Output column number.
2884         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
2885         * doc/invoke.texi: Document optall -fopt-info flag.
2886         * profile.c (read_profile_edge_counts): Use new dump framework.
2887         (compute_branch_probabilities): Ditto.
2888         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
2889         when pass not in any opt group.
2890         * pass_manager.h (pass_manager::get_pass_profile): New method.
2891         * value-prof.c (check_counter): Use new dump framework.
2892         (check_ic_target): Ditto.
2893         * coverage.c (get_coverage_counts): Ditto.
2894         (coverage_init): Setup new dump framework.
2896 2013-08-29  Richard Biener  <rguenther@suse.de>
2898         PR tree-optimization/58246
2899         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
2900         handle the dominance check inside a basic-block.
2902 2013-08-29  Richard Biener  <rguenther@suse.de>
2904         PR middle-end/57287
2905         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
2906         of default defs that appear in abnormal PHI nodes.
2908 2013-08-29  Richard Biener  <rguenther@suse.de>
2910         PR tree-optimization/57685
2911         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
2912         single-use operands to avoid exponential complexity.
2914 2013-08-28  Dehao Chen  <dehao@google.com>
2916         * ipa-inline.c (edge_badness): Fix integer underflow.
2918 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
2920         * gtm-builtins.def (_ITM_free): Declare leaf.
2922 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
2924         PR target/58067
2925         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
2926         (*tls_local_dynamic_base_64_largepic): Likewise.
2927         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
2928         Remove predicate from call operand.
2929         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
2930         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
2932 2013-08-28  Jeff Law  <law@redhat.com>
2934         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
2935         checks for the number of predecessors and successors allowed.
2936         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
2937         which require copying a joiner block if there is a request which
2938         is a subpath that requires no joiner block copying.
2940 2013-08-28  Jan Hubicka  <jh@suse.cz>
2942         * lto-streamer-out.c (DFS_write_tree_body): Drop
2943         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
2944         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
2945         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
2946         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
2947         Do not read DECL_ERROR_ISSUED.
2948         (unpack_ts_decl_with_vis_value_fields): Do not read
2949         DECL_DEFER_OUTPUT.
2950         (lto_input_ts_binfo_tree_pointers): Do not read
2951         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
2952         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
2953         write DECL_ERROR_ISSUED..
2954         (pack_ts_decl_with_vis_value_fields): Do not write
2955         DECL_DEFER_OUTPUT.
2956         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
2957         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
2958         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
2959         * tree.h (tree_decl_common): Update comment.
2960         (DECL_ERROR_ISSUED): Remove.
2962 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
2964         PR middle-end/58257
2965         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
2967 2013-08-28  Jan Hubicka  <jh@suse.cz>
2969         * builtins.def (free): Declare leaf.
2971 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
2973         * gdbhooks.py: New.
2974         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
2975         * configure: Regenerate.
2977 2013-08-27  Martin Jambor  <mjambor@suse.cz>
2979         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
2980         (ipa_ancestor_jf_data): Likewise.
2981         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
2982         (ipa_get_jf_pass_through_type_preserved): New function.
2983         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
2984         (ipa_get_jf_ancestor_type_preserved): New function.
2985         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
2986         (ipa_get_jf_ancestor_result): Likewise.
2987         (propagate_vals_accross_pass_through): Use
2988         ipa_get_jf_pass_through_result to do all the value mappings.
2989         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
2990         type_preserved flag.
2991         (ipa_set_jf_cst_copy): New function.
2992         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
2993         (ipa_set_jf_arith_pass_through): Likewise.
2994         (ipa_set_ancestor_jf): Likewise.
2995         (compute_complex_assign_jump_func): Set type_preserved instead of
2996         punting.
2997         (ipa_compute_jump_functions_for_edge): Likewise.
2998         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
2999         (update_jump_functions_after_inlining): Update type_preserved.
3000         Explicitely create jump functions when combining one with pass_through.
3001         (ipa_write_jump_function): Stream the type_preserved flags.
3002         (ipa_read_jump_function): Likewise.
3004 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
3005             Aldy Hernandez  <aldyh@redhat.com>
3007         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
3008         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
3009         * function.h (struct function): Add has_force_vect_loops and
3010         has_simduid_loops.
3011         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
3012         * gimple.c (gimple_build_omp_critical): Add KIND argument and
3013         handle it.
3014         * gimple.def: Update CLAUSES comments.
3015         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
3016         (gimple_build_omp_for): Add argument to prototype.
3017         (gimple_omp_for_kind): New.
3018         (gimple_omp_for_set_kind): New.
3019         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
3020         GOVD_DATA_SHARE_CLASS.
3021         (enum omp_region_type): Add ORT_SIMD.
3022         (gimple_add_tmp_var): Handle ORT_SIMD.
3023         (gimplify_var_or_parm_decl): Same.
3024         (is_gimple_stmt): Same.
3025         (omp_firstprivatize_variable): Same.
3026         (omp_add_variable): Only use splay_tree_insert if lookup failed.
3027         (omp_notice_variable): Handle ORT_SIMD.
3028         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
3029         (omp_check_private): Handle ORT_SIMD.
3030         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
3031         OMP_CLAUSE_SAFELEN.
3032         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
3033         Handle OMP_CLAUSE_LASTPRIVATE.
3034         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
3035         OMP_CLAUSE_SAFELEN.
3036         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
3037         (gimplify_expr): Handle OMP_SIMD.
3038         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
3039         (expand_GOMP_SIMD_VF): New.
3040         (expand_GOMP_SIMD_LAST_LANE): New.
3041         * internal-fn.def (GOMP_SIMD_LANE): New.
3042         (GOMP_SIMD_VF): New.
3043         (GOMP_SIMD_LAST_LANE): New.
3044         * omp-low.c: Include target.h.
3045         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
3046         OMP_CLAUSE_SAFELEN.
3047         (check_omp_nesting_restrictions): Same.
3048         (omp_max_vf): New.
3049         (lower_rec_simd_input_clauses): New.
3050         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
3051         OMP_CLAUSE_LINEAR.
3052         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
3053         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
3054         (expand_omp_build_assign): New.
3055         (expand_omp_for_init_counts): New.
3056         (expand_omp_for_init_vars): New.
3057         (extract_omp_for_update_vars): New.
3058         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
3059         and rewrite accordingly.
3060         (expand_omp_simd): New.
3061         (expand_omp_for): Use expand_omp_simd.
3062         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
3063         (lower_omp_for): Do not lower the body.
3064         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
3065         in their own loops.
3066         * tree-flow.h (find_omp_clause): Remove prototype.
3067         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
3068         forcing vectorization of the loop, or if flag_tree_vectorize.
3069         (gate_tree_if_conversion): Similarly.
3070         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
3071         gimple_build_omp_for.
3072         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
3073         * tree-parloops (create_parallel_loop): Pass kind argument to
3074         gimple_build_omp_for.
3075         * tree-pretty-print.c (dump_omp_clause): Add cases for
3076         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
3077         OMP_CLAUSE__SIMDUID_.
3078         (dump_generic_node): Handle OMP_SIMD.
3079         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
3080         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
3081         unroll OMP_SIMD loops here.
3082         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
3083         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
3084         loop->safelen.
3085         (vect_analyze_data_refs): Handle simd loops.
3086         * tree-vect-loop.c (vectorizable_live_operation): Handle
3087         IFN_GOMP_SIMD*.
3088         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
3089         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
3090         (vectorizable_load): Same.
3091         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
3092         (struct simduid_to_vf): New.
3093         (simduid_to_vf::hash): New.
3094         (simduid_to-vf::equal): New.
3095         (struct simd_array_to_simduid): New.
3096         (simd_array_to_simduid::hash): New.
3097         (simd_array_to_simduid::equal): New.
3098         (adjust_simduid_builtins): New.
3099         (struct note_simd_array_uses_struct): New.
3100         (note_simd_array_uses_cb): New.
3101         (note_simd_array_uses): New.
3102         (vectorize_loops): Handle simd hints and adjust simd builtins
3103         accordingly.
3104         * tree-vectorizer.h (struct _stmt_vec_info): Add
3105         simd_lane_access_p field.
3106         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
3107         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
3108         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
3109         (omp_clause_code_name): Same.
3110         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
3111         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
3112         * tree.def (OMP_SIMD): New entry.
3113         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
3114         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
3115         (OMP_CLAUSE_DECL): Adjust range for new clauses.
3116         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
3117         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
3118         (OMP_CLAUSE_LINEAR_STEP): New.
3119         (OMP_CLAUSE_SAFELEN_EXPR): New.
3120         (OMP_CLAUSE__SIMDUID__DECL): New.
3121         (find_omp_clause): New prototype.
3123 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
3125         * config/i386/driver-i386.c (host_detect_local_cpu): Update
3126         Haswell processor detection.
3128 2013-08-27  Christian Widmer  <shadow@umbrox.de>
3130         PR target/57927
3131         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
3132         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
3133         AVX2 capable processors.
3135 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
3137         * config/aarch64/arm_neon.h: Replace all inline asm implementations
3138         of vget_low_* with implementations in terms of other intrinsics.
3140 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
3142         PR middle-end/57219
3143         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
3144         values to -1, 0 and 1.
3146 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
3148         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
3149         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
3150         (<optab><mode>3_insn): Remove.
3151         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
3152         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
3153         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
3154         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
3155         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
3156         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
3157         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
3158         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
3159         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
3160         (ror<mode>3_insn): Likewise.
3161         * config/aarch64/predicates.md (aarch64_simd_register): New.
3163 2013-08-27  Richard Biener  <rguenther@suse.de>
3165         PR tree-optimization/57521
3166         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
3167         one edge is non-critical.
3168         (find_phi_replacement_condition): Make sure to use a non-critical
3169         edge.  Cleanup and remove old bug workarounds.
3170         (bb_postdominates_preds): Remove.
3171         (if_convertible_loop_p_1): Do not compute post-dominators.
3172         (combine_blocks): Do not free post-dominators.
3173         (main_tree_if_conversion): Likewise.
3174         (pass_data_if_conversion): Add TODO_verify_ssa.
3176 2013-08-27  DJ Delorie  <dj@redhat.com>
3178         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
3180 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
3182         * function.c (assign_parm_find_data_types): Set passed_mode and
3183         nominal_mode to the TYPE_MODE of nominal_type for the built
3184         pointer type in case of the struct-pass-by-reference.
3186 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
3188         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
3189         (UINT16_TYPE): Change default to "unsigned int".
3191         * config/avr/avr.opt (mfract-convert-truncate): New option.
3192         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
3193         is set, round negative fractional integers according to n1169
3194         when converting to integer types.
3196 2013-08-26  Jan Hubicka  <jh@suse.cz>
3198         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
3199         methods can not be called indirectly when their address is not taken.
3201 2013-08-26  Jan Hubicka  <jh@suse.cz>
3203         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
3204         ctor_for_folding.
3206 2013-08-26  Jan Hubicka  <jh@suse.cz>
3208         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
3209         can be unshared.
3211 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
3213         * reload.c (find_valid_class): Allow classes that do not include
3214         FIRST_PSEUDO_REGISTER - 1.
3216 2013-08-26  Jan Hubicka  <jh@suse.cz>
3218         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
3219         fix edge count/frequency when speculation failed; fix type check
3220         for the direct call.
3222 2013-08-26  Jan Hubicka  <jh@suse.cz>
3224         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
3226 2013-08-26  Jan Hubicka  <jh@suse.cz>
3228         * ipa-inline-transform.c (inline_transform): Be ready for basic block
3229         to be changed by edge redirection.
3231 2013-08-26  Jan Hubicka  <jh@suse.cz>
3233         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
3234         formating; add sanity check.
3235         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
3236         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
3237         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
3238         (resolve_noninline_speculation): Update callee keys, too.
3240 2013-08-26  Jan Hubicka  <jh@suse.cz>
3242         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
3243         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
3245 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
3247         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
3248         into proper place.
3250 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
3252         * config/i386/i386.c (ix86_debug_options): Remove prototype.
3253         (x86_64_elf_select_section): Ditto.
3254         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
3255         arguments.
3256         (ix86_pass_by_reference): Ditto.
3257         (output_set_got): Ditto.
3258         (ix86_unary_operator_ok): Ditto.
3259         (ix86_expand_builtin): Ditto.
3261 2013-08-23  Jan Hubicka  <jh@suse.cz>
3263         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
3265 2013-08-23  Jan Hubicka  <jh@suse.cz>
3267         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
3268         (tree_decl_with_vis): Add FINAL field.
3270 2013-08-23  Jeff Law  <law@redhat.com>
3272         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
3273         the debugging dump when the expression is fully redundant.
3275 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3277         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
3278         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
3279         * pretty-print.c (pp_formatted_text_data): Likewise.
3280         (pp_write_text_to_stream): Likewise.
3281         (pp_write_text_as_dot_label_to_stream): Likewise.
3282         (pp_append_r): Likewise.
3283         (pp_format): Likewise.
3284         (pp_flush): Likewise.
3285         (pp_clear_output_area): Likewise.
3286         (pp_append_text): Likewise.
3287         (pp_formatted_text): Likewise.
3288         (pp_remaining_character_count_for_line): Likewise.
3289         (pp_newline): Likewise.
3290         (pp_character): Likewise.
3291         (output_buffer::~output_buffer): Define.
3292         (pretty_printer::~pretty_printer): Destruct output buffer.
3293         * pretty-print.h (output_buffer::~output_buffer): Declare.
3294         (pretty_printer::~pretty_printer): Declare virtual.
3296 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
3298         PR other/57324
3299         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
3300         HOST_WIDE_INT_M1U): New macros.
3301         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
3302         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
3303         unsigned -1 for lshift.
3304         * cse.c (cse_insn): Likewise.
3305         * double-int.c (rshift_double, lshift_double): Likewise.
3306         * builtins.c (fold_builtin_bitop): Likewise.
3307         * combine.c (force_to_mode): Likewise.
3308         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
3309         * simplify-rtx.c (simplify_const_unary_operation,
3310         simplify_const_binary_operation): Likewise.
3311         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
3312         check_va_list_escapes): Likewise.
3313         * rtlanal.c (nonzero_bits1): Likewise.
3314         * expmed.c (expand_smod_pow2): Likewise.
3315         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
3317 2013-08-23  Jan Hubicka  <jh@suse.cz>
3319         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
3320         as having address taken.
3322 2013-08-23  Jan Hubicka  <jh@suse.cz>
3324         * ipa-utils.h (method_class_type): Declare.
3325         * ipa-devirt.c (method_class_type): Export.
3327         * cgraphunit.c (analyze_functions): Do basic devirtualization;
3328         do not walk base classes of anonymous types.
3330 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
3332         PR rtl-optimization/58220
3333         PR regression/58221
3334         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
3335         handle SEQUENCE insns properly.
3337 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3339         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
3340         definition.
3341         (pp_newline_and_indent): Likewise.
3342         (pp_separate_with): Likewise.
3343         * pretty-print.c (pp_newline_and_flush): Define.
3344         (pp_newline_and_indent): Likewise.
3345         (pp_separate_with): Likewise.
3347 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
3349         PR target/58218
3350         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
3351         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
3353 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
3355         * config/i386/predicates.md (ext_sse_reg_operand): New.
3356         * config/i386/i386.md (*movti_internal): Use
3357         predicate to determine if EVEX is needed.
3358         (*movsi_internal): Ditto.
3359         (*movdf_internal): Ditto.
3360         (*movsf_internal): Ditto.
3361         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
3363 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
3365         PR tree-optimization/58209
3366         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
3367         (find_tail_calls): Give up for pointer result types if m is non-NULL.
3368         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
3369         emit POINTER_PLUS_EXPR.
3370         (create_tailcall_accumulator): For pointer result type accumulate in
3371         sizetype type.
3373 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
3375         * configure.ac: Add backslashes missing from the last change.
3376         * configure: Regenerate.
3378 2013-08-22  Jan Hubicka  <jh@susue.cz>
3380         * ipa.c (function_and_variable_visibility): First remember function
3381         was global and then make it local.
3383 2013-08-22  Julian Brown  <julian@codesourcery.com>
3385         * configure.ac: Add aarch64 to list of arches which use "nop" in
3386         debug_line test.
3387         * configure: Regenerate.
3389 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3391         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
3392         gnu_libc_has_function.
3393         * config/s390/tpf.h: Likewise.
3395 2013-08-22  Jan Hubicka  <jh@susue.cz>
3397         * timevar.c (validate_phases): Add cast.
3399 2013-08-22  Jan Hubicka  <jh@susue.cz>
3401         * timevar.c (validate_phases): Use size_t for memory.
3402         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
3404 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3406         * pretty-print.h (output_buffer::output_buffer): Declare.
3407         (pretty_printer::pretty_printer): Likewise.
3408         (pp_construct): Remove.
3409         * pretty-print.c (output_buffer::output_buffer): Define.
3410         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
3411         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
3412         (print_gimple_expr): Likewise.
3413         (print_gimple_seq): Likewise.
3414         (gimple_dump_bb): Likewise.
3415         * sched-vis.c (dump_value_slim): Likewise.
3416         (dump_insn_slim): Likewise.
3417         (dump_rtl_slim): Likewise.
3418         (str_pattern_slim): Likewise.
3419         * tree-mudflap.c (mf_varname_tree): Likewise.
3420         * graph.c (print_graph_cfg): Likewise.
3421         (start_graph_dump): Likewise.
3422         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
3423         placement-new.
3424         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
3425         pretty printer initialization.
3426         * coretypes.h (diagnostic_context): Remove superflous type alias
3427         declaration.
3428         (pretty_printer): Likewise.  Declare directly as a class.
3429         (pretty_print_info): Remove declaration as class.
3430         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
3431         and pp_clear_output_area.
3432         (asan_add_global): Likewise.
3434 2013-08-22  Jan Hubicka  <jh@suse.cz>
3436         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
3437         * ipa-utils.h (update_type_inheritance_graph): Declare.
3438         (possible_polymorphic_call_target_p): Declare.
3439         (possible_polymorphic_call_target_p): New.
3440         * ipa-devirt.c: Update toplevel comments.
3441         (cached_polymorphic_call_targets): Move up.
3442         (odr_type_d): Move ID down.
3443         (polymorphic_type_binfo_p): Update comment.
3444         (odr_hasher::remove): Likewise;
3445         (get_odr_type): Set anonymous_namespace.
3446         (dump_odr_type): Dump it.
3447         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
3448         (maybe_record_node): Record node in cached_polymorphic_call_targets.
3449         (record_binfo): Add comment.
3450         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
3451         built.
3452         (devirt_node_removal_hook): Do not iCE when cache is freed.
3453         (possible_polymorphic_call_target_p): New predicate.
3454         (update_type_inheritance_graph): New function.
3456 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3457             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3458             Sergey Lega  <sergey.s.lega@intel.com>
3459             Anna Tikhonova  <anna.tikhonova@intel.com>
3460             Ilya Tocar  <ilya.tocar@intel.com>
3461             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3462             Ilya Verbin  <ilya.verbin@intel.com>
3463             Kirill Yukhin  <kirill.yukhin@intel.com>
3464             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3466         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
3467         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
3468         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
3469         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
3470         (OPTION_MASK_ISA_AVX2_UNSET): Update.
3471         (OPTION_MASK_ISA_AVX512F_UNSET): New.
3472         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
3473         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
3474         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
3475         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
3476         OPT_mavx512pf, OPT_mavx512er cases.
3477         * config/i386/constraints.md (v): New constraint.
3478         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
3479         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
3480         (bit_AVX512CD): New.
3481         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
3482         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
3483         * config/i386/i386-c.c (ix86_target_macros_internal):
3484         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
3485         __AVX512PF__.
3486         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
3487         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
3488         * config/i386/i386.c (regclass_map, dbx_register_map)
3489         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
3490         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
3491         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
3492         -mavx512pf options.
3493         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
3494         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
3495         -mavx512pf options.  Fix formatting.
3496         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
3497         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
3498         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
3499         -mavx512cd, -mavx512pf options.
3500         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
3501         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
3502         (ix86_preferred_output_reload_class): Replace SSE_REGS with
3503         ALL_SSE_REGS.
3504         (ix86_hard_regno_mode_ok): Support 512-bit registers.
3505         (ix86_set_reg_reg_cost): Ditto.
3506         (x86_order_regs_for_local_alloc): Ditto.
3507         (MAX_VECT_LEN): Extend to 64-byte.
3508         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
3509         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
3510         (TARGET_AVX512ER, TARGET_AVX512CD): New.
3511         (BIGGEST_ALIGNMENT): Extend to 512-bits.
3512         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
3513         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
3514         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
3515         (SSE_REG_MODE_P): Support new modes.
3516         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
3517         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
3518         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
3519         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
3520         (REG_CLASS_CONTENTS): Add new registers.
3521         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
3522         (EXT_REX_SSE_REGNO_P): New.
3523         (HI_REGISTER_NAMES): Add new registers.
3524         * config/i386/i386.md: Define constants for new registers.
3525         (mode): Add new 512-bit modes.
3526         (prefix): Support evex prefix.
3527         (isa): Support avx512f, noavx512f, fma_avx512f.
3528         (ssemodesuffix): Add new 512-bit modes.
3529         (movxi): New.
3530         (*movxi_internal_avx512f): Ditto.
3531         (*movdi_internal): Replace constraint "x" with the new constraint "v".
3532         Support MODE_XI.
3533         (*movsi_internal): Likewise.
3534         (*movdf_internal): Likewise.
3535         (*movsf_internal): Likewise.
3536         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
3537         (<code><mode>3): Likewise.
3538         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
3539         New.
3540         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
3541         with the new constraint "v".
3542         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
3543         modes.
3544         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
3545         with the new constraint "v".
3546         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
3547         (<sse2>_storedqu<avxsizesuffix>): Likewise.
3548         (*<plusminus_insn><mode>3): Likewise.
3549         (<sse>_vm<plusminus_insn><mode>3): Likewise.
3550         (*mul<mode>3): Likewise.
3551         (<sse>_vmmul<mode>3): Likewise.
3552         (<sse>_div<mode>3): Likewise.
3553         (<sse>_vmdiv<mode>3): Likewise.
3554         (<sse>_sqrt<mode>2): Likewise.
3555         (<sse>_vmsqrt<mode>2): Likewise.
3556         (*<code><mode>3_finite): Likewise.
3557         (*<code><mode>3) <smaxmin>: Likewise.
3558         (<sse>_vm<code><mode>3): Likewise.
3559         (*<code><mode>3) <any_logic>: Likewise.
3560         (*fma_fmadd_<mode>): Likewise.
3561         (*fma_fmsub_<mode>): Likewise.
3562         (*fma_fnmadd_<mode>): Likewise.
3563         (*fma_fnmsub_<mode>): Likewise.
3564         (*fma_fmaddsub_<mode>): Likewise.
3565         (*fma_fmsubadd_<mode>): Likewise.
3566         (*fmai_fmadd_<mode>): Likewise.
3567         (*fmai_fmsub_<mode>): Likewise.
3568         (*fmai_fnmadd_<mode>): Likewise.
3569         (*fmai_fnmsub_<mode>): Likewise.
3570         (sse_cvtsi2ss): Likewise.
3571         (sse_cvtsi2ssq): Likewise.
3572         (sse_cvtss2si): Likewise.
3573         (sse_cvtss2si_2): Likewise.
3574         (sse_cvtss2siq): Likewise.
3575         (sse_cvtss2siq_2): Likewise.
3576         (sse_cvttss2si): Likewise.
3577         (sse_cvtss2siq_2): Likewise.
3578         (float<sseintvecmodelower><mode>2): Likewise.
3579         (sse2_cvtsd2si_2): Likewise.
3580         (sse2_cvtsd2siq_2): Likewise.
3581         (*<plusminus_insn><mode>3): Likewise.
3582         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
3583         (*<sse4_1_avx2>_mul<mode>3): Likewise.
3584         (ashr<mode>3): Likewise.
3585         (<shift_insn><mode>3): Likewise.
3586         (avx2_<code><mode>3): Likewise.
3587         (*avx2_<code><mode>3): Likewise.
3588         (*andnot<mode>3): Likewise.
3589         (*<code><mode>3) <any_logic>: Likewise.
3590         (abs<mode>2): Likewise.
3591         (avx2_permvar<mode>): Likewise.
3592         (avx2_perm<mode>_1): Likewise.
3593         (*avx_vpermilp<mode>): Likewise.
3594         (avx_vpermilvar<mode>3): Likewise.
3595         (avx2_ashrv<mode>): Likewise.
3596         (avx2_<shift_insn>v<mode>): Likewise.
3597         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
3598         -mavx512cd.
3599         * doc/rtl.texi: Document XImode.
3601 2013-08-21  Jeff Law  <law@redhat.com>
3603         * tree-flow.h (register_jump_thread): Pass vector of edges
3604         instead of each important edge.
3605         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
3606         thread path into a vector and pass that to register_jump_thread.
3607         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
3608         passed in edge vector to the current 3-edge form.
3610         Revert:
3611         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
3613         * dce.c (fini_dce): Call df_analyze again just in case
3614         delete_unmarked_insns removed anything.
3616 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
3618         * reload.h (struct reg_equivs): Rename to ..
3619         (struct reg_equivs_s): .. this.
3621 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
3623         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
3625 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3627         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
3629 2013-08-21  Jeff Law  <law@redhat.com>
3631         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
3632         simplify assignments too.  If the RHS collapses to a singleton
3633         range, then return the value for the range.
3635 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
3637         * config/i386/sse.md (V16): Rename to...
3638         (VMOVE): this.
3639         (mov<mode>): Update iterator name.
3640         (*mov<mode>_internal): Ditto.
3641         (push<mode>1): Ditto.
3642         (movmisalign<mode>): Ditto.
3644 2013-08-20  Jan Hubicka  <jh@suse.cz>
3646         PR bootstrap/58186
3647         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
3648         entry for direct edges.
3649         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
3651 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3653         Revert my last two changes, r201865 and r201864:
3655         Revert r201865:
3656         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3658         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
3659         instances can own GC refs.
3661         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
3662         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
3663         (gcc::context::gt_pch_nx): Likewise.
3664         (gcc::context::gt_pch_nx):  Likewise.
3665         * ggc.h (gt_ggc_mx <T>): New.
3666         (gt_pch_nx_with_op <T>): New.
3667         (gt_pch_nx <T>): New.
3668         * passes.c (opt_pass::gt_ggc_mx): New.
3669         (opt_pass::gt_pch_nx): New.
3670         (opt_pass::gt_pch_nx_with_op): New.
3671         (pass_manager::gt_ggc_mx): New.
3672         (pass_manager::gt_pch_nx): New.
3673         (pass_manager::gt_pch_nx_with_op): New.
3674         (pass_manager::operator new): Use
3675         ggc_internal_cleared_alloc_stat rather than xcalloc.
3676         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
3677         (pass_manager::gt_ggc_mx): New.
3678         (pass_manager::gt_pch_nx): New.
3679         (pass_manager::gt_pch_nx_with_op): New.
3680         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
3681         (opt_pass::operator new): New.
3682         (opt_pass::gt_ggc_mx): New.
3683         (opt_pass::gt_pch_nx): New.
3684         (opt_pass::gt_pch_nx_with_op): New.
3686         Revert r201864:
3687         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3689         * Makefile.in (GTFILES): Add context.h.
3690         * context.c (gcc::context::operator new): New.
3691         (gcc::context::gt_ggc_mx): New.
3692         (gcc::context::gt_pch_nx): New.
3693         (gcc::context::gt_pch_nx): New.
3694         * context.h (gcc::context): Add GTY((user)) marking.
3695         (gcc::context::operator new): New.
3696         (gcc::context::gt_ggc_mx): New.
3697         (gcc::context::gt_pch_nx): New.
3698         (gcc::context::gt_pch_nx): New.
3699         (g): Add GTY marking.
3700         (gt_ggc_mx (gcc::context *)): New.
3701         (gt_pch_nx (gcc::context *)): New.
3702         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
3703         void *cookie)): New.
3704         * gengtype.c (open_base_files) <ifiles>: Add context.h.
3706 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
3708         * dce.c (fini_dce): Call df_analyze again just in case
3709         delete_unmarked_insns removed anything.
3711 2013-08-20  Teresa Johnson  <tejohnson@google.com>
3713         PR rtl-optimizations/57451
3714         * final.c (reemit_insn_block_notes): Prevent lexical blocks
3715         from crossing split section boundaries.
3717 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
3719         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
3720         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
3721         with MULTLIB_DEFAULTS.
3723 2013-08-20  Nick Clifton  <nickc@redhat.com>
3725         * target.def (narrow_volatile_bitfield): Note that the default
3726         value is false, not !TARGET_STRICT_ALIGN.
3727         * doc/tm.texi: Regenerate.
3729 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
3731         Fix LIB_SPEC for systems without libpthread.
3733         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
3734         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
3735         for Android.
3736         * config/i386/linux-common.h: Likewise.
3737         * config/mips/linux-common.h: Likewise.
3739 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
3741         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
3742         checks.
3744 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3746         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
3747         instances can own GC refs.
3749         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
3750         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
3751         (gcc::context::gt_pch_nx): Likewise.
3752         (gcc::context::gt_pch_nx):  Likewise.
3753         * ggc.h (gt_ggc_mx <T>): New.
3754         (gt_pch_nx_with_op <T>): New.
3755         (gt_pch_nx <T>): New.
3756         * passes.c (opt_pass::gt_ggc_mx): New.
3757         (opt_pass::gt_pch_nx): New.
3758         (opt_pass::gt_pch_nx_with_op): New.
3759         (pass_manager::gt_ggc_mx): New.
3760         (pass_manager::gt_pch_nx): New.
3761         (pass_manager::gt_pch_nx_with_op): New.
3762         (pass_manager::operator new): Use
3763         ggc_internal_cleared_alloc_stat rather than xcalloc.
3764         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
3765         (pass_manager::gt_ggc_mx): New.
3766         (pass_manager::gt_pch_nx): New.
3767         (pass_manager::gt_pch_nx_with_op): New.
3768         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
3769         (opt_pass::operator new): New.
3770         (opt_pass::gt_ggc_mx): New.
3771         (opt_pass::gt_pch_nx): New.
3772         (opt_pass::gt_pch_nx_with_op): New.
3774 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3776         * Makefile.in (GTFILES): Add context.h.
3777         * context.c (gcc::context::operator new): New.
3778         (gcc::context::gt_ggc_mx): New.
3779         (gcc::context::gt_pch_nx): New.
3780         (gcc::context::gt_pch_nx): New.
3781         * context.h (gcc::context): Add GTY((user)) marking.
3782         (gcc::context::operator new): New.
3783         (gcc::context::gt_ggc_mx): New.
3784         (gcc::context::gt_pch_nx): New.
3785         (gcc::context::gt_pch_nx): New.
3786         (g): Add GTY marking.
3787         (gt_ggc_mx (gcc::context *)): New.
3788         (gt_pch_nx (gcc::context *)): New.
3789         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
3790         void *cookie)): New.
3791         * gengtype.c (open_base_files) <ifiles>: Add context.h.
3793 2013-08-20  Alan Modra  <amodra@gmail.com>
3795         PR target/57865
3796         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
3797         (rs6000_emit_epilogue): Likewise.
3799 2013-08-19  Dehao Chen  <dehao@google.com>
3801         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
3803 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
3804             Jakub Jelinek  <jakub@redhat.com>
3806         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
3807         (BUILT_IN_FABSD64): Likewise.
3808         (BUILT_IN_FABSD128): Likewise.
3809         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
3810         (fold_builtin_1): Likewise.
3811         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
3812         destination and source operands.
3813         (*abstd2_fpr): Likewise.
3814         (*nabstd2_fpr): Likewise.
3816 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
3818         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
3819         JUMP_P and INSN_P.
3821 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
3823         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
3824         incomplete.
3826 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3828         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
3829         * builtins.c (default_libc_has_function): New.
3830         (gnu_libc_has_function): Ditto.
3831         (no_c99_libc_has_function): Ditto.
3832         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
3833         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
3834         (fold_builtin_sincos): Likewise.
3835         (fold_builtin_cexp): Likewise.
3836         * builtins.def (DEF_C94_BUILTIN): Likewise.
3837         (DEF_C99_BUILTIN): Likewise.
3838         (DEF_C99_C90RES_BUILTIN): Likewise.
3839         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
3840         definitions to using this define.
3841         * config/darwin-protos.h (darwin_libc_has_function): New.
3842         * config/darwin.c (darwin_libc_has_function): Ditto.
3843         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
3844         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
3845         * config/darwin.h: Ditto.
3846         * config/elfos.h: Ditto.
3847         * config/freebsd.h: Ditto.
3848         * config/i386/cygming.h: Ditto.
3849         * config/i386/djgpp.h: Ditto.
3850         * config/i386/i386-interix.h: Ditto.
3851         * config/microblaze/microblaze.h: Ditto.
3852         * config/mmix/mmix.h: Ditto.
3853         * config/gnu-user.h: Ditto.
3854         * config/ia64/hpux.h: Ditto.
3855         * config/pa/pa-hpux.h: Ditto.
3856         * config/pdp11/pdp11.h: Ditto.
3857         * config/picochip/picochip.h: Ditto.
3858         * config/linux.h: Ditto.
3859         * config/netbsd.h: Ditto.
3860         * config/openbsd.h: Ditto.
3861         * config/rs6000/aix43.h: Ditto.
3862         * config/rs6000/aix51.h: Ditto.
3863         * config/rs6000/aix52.h: Ditto.
3864         * config/rs6000/aix53.h: Ditto.
3865         * config/rs6000/aix61.h: Ditto.
3866         * config/rs6000/darwin.h: Ditto.
3867         * config/rs6000/linux.h: Ditto.
3868         * config/rs6000/linux64.h: Ditto.
3869         * config/s390/tpf.h: Ditto.
3870         * config/sol2-10.h: Ditto.
3871         * config/sol2.h: Ditto.
3872         * config/vms/vms.h: Ditto.
3873         * config/vxworks.h: Ditto.
3874         * config/linux-android.c (linux_android_libc_has_function):
3875         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
3876         * config/linux-protos.h (linux_android_libc_has_function):
3877         New declaration.
3878         * config/i386/i386.c (ix86_libc_has_function): New.
3879         * config/i386/i386-protos.h
3880         (ix86_libc_has_function): New declaration.
3881         * config/i386/i386.md
3882         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
3883         ("isinf<mode>2): Likewise.
3884         * convert.c (convert_to_integer): Using new target hook
3885         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
3886         TARGET_C99_FUNCTIONS.
3887         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
3888         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
3889         * coretypes.h (function_class): New enum for different
3890         classes of functions.
3891         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
3892         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
3893         (TARGET_HAS_SINCOS): Likewise.
3894         (TARGET_LIBC_HAS_FUNCTION): New.
3895         * doc/tm.texi: Regenerated.
3896         * targhooks.h (default_libc_has_function): New declaration.
3897         (no_c99_libc_has_function): Ditto.
3898         (gnu_libc_has_function): Ditto.
3899         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
3900         and TARGET_HAS_SINCOS.
3902 2013-08-18  Jan Hubicka  <jh@suse.cz>
3904         * Makeifle-in (ipa-devirt.o): New.
3905         (GTFILES): Add ipa-utils.h and ipa-devirt.c
3906         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
3907         (analyze_functions): Look into possible targets of polymorphic call.
3908         * dumpfile.c (dump_files): Add type-inheritance dump.
3909         * dumpfile.h (TDI_inheritance): New.
3910         * ipa-devirt.c: New file.
3911         * ipa-utils.h (odr_type_d): Forward declare.
3912         (odr_type): New type.
3913         (build_type_inheritance_graph): Declare.
3914         (possible_polymorphic_call_targets): Declare and introduce inline
3915         variant when only edge is pased.
3916         (dump_possible_polymorphic_call_targets): Likewise.
3917         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
3918         * tree.c (type_in_anonymous_namespace_p): Break out from ...
3919         (types_same_for_odr): ... here.
3920         * tree.h (type_in_anonymous_namespace_p): Declare.
3922 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
3924         PR tree-optimization/58006
3925         * tree-parloops.c (take_address_of): Don't ICE if get_name
3926         returns NULL.
3927         (eliminate_local_variables_stmt): Remove clobber stmts.
3929 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
3931         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
3932         error message is issued for an alias to undefined symbol.
3934 2013-08-18  Jan Hubicka  <jh@suse.cz>
3936         * cgraph.c (cgraph_create_indirect_edge): Discover
3937         polymorphic calls and record basic info into indirect_info.
3938         * gimple-fold.c (gimple_fold_call): When doing BINFO based
3939         devirtualization, ignore objc function calls.
3940         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
3941         call with no parm index info.
3942         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
3943         * tree.c (virtual_method_call_p): New function.
3944         * tree.h (virtual_method_call_p): Declare.
3946 2013-08-16  Jan Hubicka  <jh@suse.cz>
3948         PR middle-end/58179
3949         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
3951 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
3953         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
3954         attribute.
3956 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
3958         * gengtype.c (type_for_name): Add special-case support for
3959         locating types within the "gcc::" namespace.
3960         (open_base_files): Emit a "using namespace gcc" directive.
3962 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
3964         PR target/58160
3965         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
3966         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
3968         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
3969         array instead of each individual operand as a separate argument.
3970         (emit_fusion_gpr_load): Likewise.
3971         (expand_fusion_gpr_load): Add new function declaration.
3973         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
3974         signature to have the operands passed as an array, instead of as
3975         separate arguments.  Allow ZERO_EXTEND to be in the memory
3976         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
3977         depend on the register live/dead flags when peepholes are run.
3978         (expand_fusion_gpr_load): New function to be called from the
3979         peephole2 pass, to change the register that addis sets to be the
3980         target register.
3981         (emit_fusion_gpr_load): Change the calling signature to have the
3982         operands passed as an array, instead of as separate arguments.
3983         Allow ZERO_EXTEND to be in the memory address, and also
3984         SIGN_EXTEND if -mpower8-fusion-sign.
3986         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
3987         unspec enumeration.
3988         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
3989         adjust the register addis loads up in the peephole2 pass.  Do not
3990         depend on the register live/dead state when the peephole pass is done.
3992 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
3994         * gengtype.c (create_user_defined_type): Ensure that the kind
3995         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
3996         declaration is seen before the GTY((user)) marking.
3998 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4000         PR target/58105
4001         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
4003 2013-08-16  Jan Hubicka  <jh@suse.cz>
4005         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
4006         arugment expected_type.
4007         (gimple_fold_call): Use it.
4008         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
4009         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
4010         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
4011         (try_make_edge_direct_virtual_call): Likewise.
4012         * tree.c (obj_type_ref_class): New.
4013         * tree.h (obj_type_ref_class): Use it.
4015 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
4017         * sched-vis.c (rtl_slim_pp_initialized): Remove.
4018         (rtl_slim_pp): Likewise.
4019         (init_rtl_slim_pretty_print): Likewise.
4020         (dump_value_slim):  Don't call it.  Use local pretty printer.
4021         (dump_insn_slim): Likewise.
4022         (dump_rtl_slim): Likewise.
4023         (str_pattern_slim): Likewise.
4024         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
4025         Simplify.
4027 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
4029         PR tree-optimization/58164
4030         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
4031         walk gimple_goto_dest of GIMPLE_GOTO.
4033         PR tree-optimization/58165
4034         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
4035         bi_call must be the last stmt in a bb, don't split_block, instead
4036         use fallthru edge from it and give up if there is none.
4037         Release conds vector when returning early.
4039 2013-08-14  Xinliang David Li  <davidxl@google.com>
4041         * config/i386/i386.c (ix86_option_override_internal):
4042         Remove unused variable and field.
4044 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4046         PR target/57949
4047         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
4048         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
4049         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
4050         and Linux, correct BLKmode alignment when 128-bit alignment is
4051         required and compatibility flag is not set.
4052         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
4053         for zero-size arguments when compatibility flag is not set.
4055 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
4057         PR tree-optimization/58145
4058         * tree-sra.c (build_ref_for_offset): If prev_base has
4059         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
4061 2013-08-14  Xinliang David Li  <davidxl@google.com>
4063         * config/i386/i386.c (ix86_option_override_internal):
4064         Fix uninitialized variable error.
4066 2013-08-14  Xinliang David Li  <davidxl@google.com>
4068         * config/i386/i386.opt: Define two new options.
4069         * config/i386/x86-tune.def: Add arch selector field in macros.
4070         * config/i386/i386.h: Adjust macro definition.
4071         * config/i386/i386.c (ix86_option_override_internal):
4072         Refactor the code.
4073         (parse_mtune_ctrl_str): New function.
4074         (set_ix86_tune_features): New function.
4075         (ix86_function_specific_restore): Call the new helper function.
4077 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
4079         PR rtl-optimization/57662
4080         * sel-sched.c (code_motion_process_successors): When the current insn
4081         is removed after the recursive traversal, break from the loop.
4082         Add comments and debug printouts.
4084 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
4085             Alexandre Oliva  <aoliva@redhat.com>
4087         PR target/58067
4088         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
4089         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
4090         there also UNSPEC_PLTOFF.
4092 2013-08-14  Marek Polacek  <polacek@redhat.com>
4094         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
4095         NUM_CONDITIONS bit positions.
4097 2013-08-13  Cary Coutant  <ccoutant@google.com>
4099         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
4100         (attr_checksum): Hash vector contents instead of pointer.
4101         (attr_checksum_ordered): Likewise.
4103 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
4105         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
4106         when Pmode != word_mode.  Add length_address attribute.
4107         (sse3_monitor_<mode>): Merge from sse3_monitor and
4108         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
4109         Pmode != word_mode.  Update insn length attribute.
4110         * config/i386/i386.c (ix86_option_override_internal): Update
4111         ix86_gen_monitor selection for merged sse3_monitor insn.
4113 2013-08-13  Julian Brown  <julian@codesourcery.com>
4115         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
4116         perform invalid legitimization on greater-than-word-size modes for
4117         TARGET_E500_DOUBLE.
4119 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
4121         * ira.c (setup_class_translate_array): Use aclass instead of cl
4122         for classes not fully covered by allocno classes.
4124 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
4126         PR tree-optimization/57661
4127         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
4128         * tree-inline.c (tree_function_versioning): Initialize it.
4129         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
4130         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
4131         that is not being copied.
4133         PR sanitizer/56417
4134         * asan.c (instrument_strlen_call): Fix typo in comment.
4135         Use char * type even for the lhs of POINTER_PLUS_EXPR.
4137 2013-08-13  Steve Ellcey  <sellcey@mips.com>
4139         * config/mips/mips.md (prefetch): Use lw instead of ld on
4140         loongson in 32bit mode.
4142 2013-08-13  Nick Clifton  <nickc@redhat.com>
4144         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
4146 2013-08-13  Jan Hubicka  <jh@suse.cz>
4148         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
4149         introduced edge; fix typo in sanity check.
4150         (cgraph_resolve_speculation): Export; improve diagnostic.
4151         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
4152         speculation at type mismatch.
4153         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
4154         (cgraph_resolve_speculation): Declare.
4155         (symtab_can_be_discarded): New function.
4156         * value-prof.c (gimple_ic_transform): Remove actual transform code.
4157         * ipa-inline-transform.c (speculation_removed): New global var.
4158         (clone_inlined_nodes): See if speculation can be removed.
4159         (inline_call): If speculations was removed, we growths may not match.
4160         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
4161         (speculation_useful_p): New function.
4162         (resolve_noninline_speculation): New function.
4163         (inline_small_functions): Resolve useless speculations.
4164         * ipa-inline.h (speculation_useful_p): Declare
4165         * ipa.c (can_replace_by_local_alias): Simplify.
4166         (ipa_profile): Produce speculative calls in non-lto, too;
4167         add simple cost model; produce local aliases.
4169 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
4171         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
4172         PASS_MANAGER_H.
4174 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
4176         * config/i386/i386.c (ix86_function_versions): Use error + inform.
4178 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
4180         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
4181         iterator instead of X87MODEF.
4183 2013-08-12  Perez Read  <netfirewall@gmail.com>
4185         PR target/58132
4186         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
4187         operand 0 for intel asm alternative.
4188         (*movabs<mode>_2): Ditto for operand 1.
4190 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
4192         * config/aarch64/arm_none.h
4193         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
4195 2013-08-12  Nick Clifton  <nickc@redhat.com>
4197         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
4199 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
4201         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
4202         expand for QI/HImode operand to produce more effictive code for
4203         unsigned char(short) --> float(double) conversion.
4205 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
4207         * doc/invoke.texi: Mention that -ftls-model does not force the final
4208         model.
4210 2013-08-12  Marek Polacek  <polacek@redhat.com>
4211             Marc Glisse  <marc.glisse@inria.fr>
4213         PR tree-optimization/57980
4214         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
4215         when creating -1 constant.
4217 2013-08-10  Jan Hubicka  <jh@suse.cz>
4219         Workaround binutils PR14342.
4220         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
4221         (gimple_init_edge_profiler): Likewise.
4222         (gimple_gen_ic_func_profiler): Likewise.
4224 2013-08-09  Jan Hubicka  <jh@suse.cz>
4226         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
4228 2013-08-09  Xinliang David Li  <davidxl@google.com>
4230         * config/i386/stringop.def: New file.
4231         * config/i386/stringop.opt: New file.
4232         * config/i386/i386-opts.h: Include stringopt.def.
4233         * config/i386/i386.opt: Include stringopt.opt.
4234         * config/i386/i386.c (ix86_option_override_internal):
4235         Override default size based stringop inline strategies with options.
4236         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
4237         New function.
4239 2013-08-09  Jan Hubicka  <jh@suse.cz>
4241         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
4243 2013-08-09  Jan Hubicka  <jh@suse.cz>
4245         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
4246         CGRAPH_FREQ_MAX.
4247         (dump_cgraph_node): Dump profile-id.
4248         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
4249         and common_target_probability.
4250         * lto-cgraph.c (lto_output_edge): Stream common targets.
4251         (lto_output_node): Stream profile ids.
4252         (input_node): Stream profile ids.
4253         (input_edge): Stream common targets.
4254         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
4255         * ipa.c: Include value-prof.h
4256         (ipa_profile_generate_summary): Turn indirect call statement histograms
4257         into common targets.
4258         (ipa_profile): Turn common targets into speculative edges.
4260 2013-08-09  Jan Hubicka  <jh@suse.cz>
4262         * cgraph.h (cgraph_node): Add profile_id.
4263         * value-prof.c (cgraph_node_map): Turn into pointer_map.
4264         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
4265         (del_node_map): Update.
4266         (find_func_by_funcdef_no): Replace by ...
4267         (find_func_by_profile_id): ... this one.
4268         (gimple_ic_transform): Do not remove useful histograms when
4269         speculation is not done; dump info when indirect call removal
4270         can happen at LTO.
4271         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
4272         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
4273         (__gcov_indirect_call_profiler_v2): .. this one.
4274         * profile.h (init_node_map): Update.
4275         * coverage.c (coverage_compute_profile_id): New function.
4276         * coverage.h (coverage_compute_profile_id): Declare.
4277         * tree-profile.c (init_ic_make_global_vars): Make
4278         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
4279         (gimple_init_edge_profiler): Update prototype of
4280         __gcov_indirect_call_profiler.
4281         (gimple_gen_ic_func_profiler): Simplify.
4282         (tree_profiling): Use init_node_map
4284 2013-08-09  Jan Hubicka  <jh@suse.cz>
4286         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
4287         non-speculative refs.
4288         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
4289         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
4290         (cgraph_set_call_stmt): Likewise.
4291         (cgraph_create_edge_1): Fix release checking compilatoin;
4292         clear lto_stmt_uid.
4293         (cgraph_free_edge): Free indirect info.
4294         (cgraph_turn_edge_to_speculative): New function.
4295         (cgraph_speculative_call_info): New function.
4296         (cgraph_make_edge_direct): Return direct edge; handle speculation.
4297         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
4298         (dump_cgraph_node): Dump speculation.
4299         (verify_edge_count_and_frequency): Accept speculative edges.
4300         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
4301         (verify_cgraph_node): Handle speculation.
4302         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
4303         (cgraph_set_call_stmt): Update prototype.
4304         (cgraph_make_edge_direct): Update prototype.
4305         (cgraph_speculative_call_info): Declare.
4306         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
4307         to change; update call of ipa_find_references.
4308         * ipa-ref.c (ipa_record_reference): Fix return value; clear
4309         lto_stmt_uid and speculative flags.
4310         (ipa_dump_references): Dump speculation.
4311         (ipa_clone_references): Clone speculative flag.
4312         (ipa_clone_referring): Likewise.
4313         (ipa_clone_ref): New function.
4314         (ipa_find_reference): Look into lto_stmt_uids
4315         (ipa_clear_stmts_in_references): Do not clear speculative calls.
4316         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
4317         (ipa_find_reference): Update declaration.
4318         (ipa_clone_ref): Declare.
4319         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
4320         stream speculative flag.
4321         (lto_output_ref): Stream statements uids and speculation.
4322         (input_ref): Likewise.
4323         (input_edge): Stream speuclation.
4324         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
4325         (cgraph_set_call_stmt_including_clones): Handle speculation.
4326         * ipa-inline.c (heap_edge_removal_hook): New function.
4327         (inline_small_functions): Register it.
4328         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
4329         also initialize refs.
4330         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
4331         edge to change.
4332         (try_make_edge_direct_simple_call): Likewise.
4333         (try_make_edge_direct_simple_call): Likewise.
4334         (update_indirect_edges_after_inlining): Likewise.
4335         (remove_described_reference): Look proper lto_stmt_uid.
4336         (propagate_controlled_uses): Likewise.
4337         (propagate_controlled_uses): Liekwise.
4338         * tree-inline.c (copy_bb): Copy speculative edges.
4339         (redirect_all_calls): New function.
4340         (copy_cfg_body): Do redirection after loop info is updated.
4341         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
4343 2013-08-09  Jan Hubicka  <jh@suse.cz>
4345         * lto-streamer-out.c (output_function): Renumber PHIs.
4346         * lto-streamer-in.c (input_function): Likewise.
4348 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
4350         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
4351         (get_lane_unsigned): Likewise.
4352         (dup_lane_scalar): Likewise.
4353         (get_lane): enable for VALL.
4354         * config/aarch64/aarch64-simd.md
4355         (aarch64_dup_lane_scalar<mode>): Remove.
4356         (aarch64_get_lane_signed<mode>): Likewise.
4357         (aarch64_get_lane_unsigned<mode>): Likewise.
4358         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
4359         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
4360         (aarch64_get_lane<mode>): Enable for all vector modes.
4361         (aarch64_get_lanedi): Remove misleading constraints.
4362         * config/aarch64/arm_neon.h
4363         (__aarch64_vget_lane_any): Define.
4364         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
4365         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
4366         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
4367         * config/aarch64/iterators.md (VDQQH): New.
4368         (VDQQHS): Likewise.
4369         (vwcore): Likewise.
4371 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
4373         * configure.ac: Add GAS check for LEON instructions on SPARC.
4374         * configure: Regenerate.
4375         * config.in: Likewise.
4376         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
4377         sparc*-*-* block.
4378         * config/sparc/sparc.opt (LEON, LEON3): New masks.
4379         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
4380         for LEON or LEON3.
4381         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
4382         (AS_LEON_FLAG): New macro.
4383         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
4384         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
4385         Deal with LEON and LEON3 for the memory model.
4386         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
4387         (atomic_compare_and_swap<mode>_1): Likewise.
4388         (*atomic_compare_and_swap<mode>_1): Likewise.
4390 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
4392         * config/arm/neon.md (vcond): Fix floating-point vector
4393         comparisons against 0.
4395 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
4397         * lra-constraints.c (emit_spill_move): Remove assert.
4398         (process_alt_operands): Add more debugging
4399         output.  Increase reject for spilling into memory.  Decrease
4400         reject for reloading scratch.
4401         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
4403 2013-08-08  Steve Ellcey  <sellcey@mips.com>
4405         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
4406         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
4407         micromips incompatible.  Add nan2008.
4408         (MULTILIB_DIRNAMES): Add nan2008.
4409         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
4410         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
4411         and micromips incompatible.  Add nan2008.
4412         (MULTILIB_DIRNAMES): Add nan2008.
4413         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
4415 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
4417         PR rtl-optimization/58079
4418         * combine.c (combine_simplify_rtx): Avoid using SUBST if
4419         simplify_comparison has widened a comparison with an integer.
4421 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4423         * config/arm/neon.md (movmisalign<mode>): Disable when we
4424         don't allow unaligned accesses.
4425         (*movmisalign<mode>_neon_store): Likewise.
4426         (*movmisalign<mode>_neon_load): Likewise.
4427         (*movmisalign<mode>_neon_store): Likewise.
4428         (*movmisalign<mode>_neon_load): Likewise.
4430 2013-08-08  Jan Hubicka  <jh@suse.cz>
4432         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
4433         (make_pass_rebuild_cgraph_edges): Also clear references.
4434         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
4435         * ipa-inline-transform.c (inline_transform): Remove all references
4436         after inlining.
4437         * cgraphunit.c (expand_function): Remove all references after
4438         expansion.
4439         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
4440         (ipa_find_reference): Rewrite to iterator.
4441         (remove_stmt_references): Likewise.
4442         (ipa_clear_stmts_in_references): New function.
4443         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
4444         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
4445         clear references.
4446         * ipa-split.c (split_function): Remove references in split function.
4448 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
4450         PR target/57431
4451         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
4452         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
4454 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
4456         PR target/56979
4457         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
4458         suggested mode for the assignment isn't compatible with the
4459         registers required.
4461 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4463         PR target/58065
4464         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
4466 2013-08-07  Xinliang David Li  <davidxl@google.com>
4468         * config/i386/i386.opt: New option -mtune-ctrl=.
4469         * config/i386/x86-tune.def: New file.
4470         * config/i386/i386.h: include x86-tune.def.
4471         * config/i386/i386.c (ix86_option_override_internal):
4472         Parsing -mtune-ctrl= option and set tune features.
4474 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
4476         PR other/12081
4477         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
4478         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
4479         to gen_2arg_fn_t.
4481 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
4483         * rtl.h (update_alignments): Declare.
4484         * final.c (grow_label_align): New function extracted from...
4485         (shorten_branches): ...here.  Call it.
4486         (update_alignments): New function.
4487         * reorg.c (sibling_labels): New variable.
4488         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
4489         the new label along with it onto the sibling_labels vector.
4490         (fill_simple_delay_slots): Adjust call to get_label_before.
4491         (fill_slots_from_thread): Likewise.
4492         (relax_delay_slots): Likewise.
4493         (make_return_insns): Likewise.
4494         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
4496 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
4498         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
4499         document its semantics.
4500         (diagnostic_report_diagnostic): Adjust accordingly.
4502 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
4504         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
4505         (sparc_option_override): ...and port to new C++ pass API.
4506         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
4508 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
4510         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
4512 2013-08-06  Caroline Tice  <cmtice@google.com>
4514         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
4515         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
4516         * tree-pass.h: Add pass_vtable_verify.
4517         * varasm.c (assemble_variable): Add code to properly set the comdat
4518         section and name for the .vtable_map_vars section.
4519         (assemble_vtyv_preinit_initializer): New function.
4520         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
4521         LINK_ONCE flag.
4522         * output.h: Add function decl for assemble_vtv_preinit_initializer.
4523         * vtable-verify.c: New file.
4524         * vtable-verify.h: New file.
4525         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
4526         initialiation levels.
4527         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
4528         * passes.def: Insert pass_vtable_verify.
4529         * aclocal.m4: Reorder includes.
4530         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
4531         -fvtv-counts options.
4532         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
4533         as appropriate, if -fvtable-verify=... is used.
4534         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
4535         -fvtable-verify=... is used.
4536         * Makefile.in (OBJS):  Add vtable-verify.o to list.
4537         (vtable-verify.o): Add new build rule.
4538         (GTFILES): Add vtable-verify.c to list.
4539         * common.opt (fvtable-verify=): New flag.
4540         (vtv_priority): Values for fvtable-verify= flag.
4541         (fvtv-counts): New flag.
4542         (fvtv-debug): New flag.
4543         * tree.h (save_vtable_map_decl): New extern function decl.
4545 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
4547         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
4548         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
4549         (pass_data_rl78_devirt): ...new pass_data instance and...
4550         (make_pass_rl78_devirt): ...new function.
4551         (rl78_asm_file_start): Port pass registration to new C++ API.
4553 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
4555         * coretypes.h (rtl_opt_pass): Add.
4556         (gcc::context): Add.
4557         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
4558         (epiphany_init): Port to new C++ pass API.
4559         (epiphany_optimize_mode_switching): Likewise.
4560         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
4561         (pass_manager::get_pass_mode_switching): New.
4562         (pass_manager::get_pass_peephole2): New.
4563         * mode-switching.c (pass_mode_switching): Add clone method.
4564         * recog.c (pass_peephole2): Add clone method.
4565         (pass_split_all_insns): Add clone method.
4567 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
4569         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
4570         (mips_option_override): ...here, porting to new C++ API for passes.
4572 2013-08-06  Jan Hubicka  <jh@suse.cz>
4574         * cgraph.c (cgraph_get_body): New function based on lto.c
4575         implementation.
4576         * cgraph.h (cgraph_get_body): Declare.
4577         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
4578         LTO paths.
4579         * cgraphunit.c (expand_function): Get body prior expanding.
4580         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
4581         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
4582         really need.
4583         * passes.c (do_per_function_toporder): Get body.
4584         * tree-inline.c (expand_call_inline): Get body prior inlining it.
4585         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
4587 2013-08-06  Martin Jambor  <mjambor@suse.cz>
4589         PR fortran/57987
4590         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
4591         is not re-finalized.  Rename second parameter to no_collect.
4593 2013-08-06  Martin Jambor  <mjambor@suse.cz>
4595         PR middle-end/58041
4596         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
4597         MEM_REF has proper alignment information.
4599 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
4601         PR other/12081
4602         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
4603         class insn_gen_fn.
4604         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
4605         rtx (*) (rtx, ...) with insn_gen_fn.
4606         * genoutput.c (output_insn_data): Cast gen_? function pointers to
4607         insn_gen_fn::stored_funcptr.  Add initializer braces.
4609 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
4611         Rewrite how instances of passes are cloned to remove assumptions
4612         about their sizes (thus allowing pass subclasses to have
4613         additional data fields, albeit non-GC-managed ones at this point).
4615         * passes.c (make_pass_instance): Now that passes have clone
4616         methods, rewrite this function to eliminate XNEW and memcpy
4617         calls that used hardcoded sizes.  Since this function no longer
4618         creates pass instances, rename it to...
4619         (add_pass_instance): ...this.  Document the old way that passes were
4620         numbered and flagged, and rework this function to continue using it.
4621         (next_pass_1): Add an initial_pass argument for use by
4622         add_pass_instance.
4623         (position_pass): When adding multiple instances of a pass, use
4624         the pass's clone method, rather than relying on the XNEW/memcpy
4625         within the former make_pass_instance (now add_pass_instance).
4626         (pass_manager::pass_manager): When invoking next_pass_1, also supply
4627         the initial instance of the current pass within the pass manager.
4629 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
4631         This is the automated part of the conversion of passes from C
4632         structs to C++ classes.
4634         Patch autogenerated by refactor_passes.py from
4635         https://github.com/davidmalcolm/gcc-refactoring-scripts
4636         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
4638         * asan.c (pass_asan): Convert from a global struct to a subclass of
4639         gimple_opt_pass along with...
4640         (pass_data_asan): ...new pass_data instance and...
4641         (make_pass_asan): ...new function.
4642         (pass_asan_O0): Convert from a global struct to a subclass of
4643         gimple_opt_pass along with...
4644         (pass_data_asan_O0): ...new pass_data instance and...
4645         (make_pass_asan_O0): ...new function.
4646         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
4647         subclass of rtl_opt_pass along with...
4648         (pass_data_inc_dec): ...new pass_data instance and...
4649         (make_pass_inc_dec): ...new function.
4650         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
4651         a subclass of rtl_opt_pass along with...
4652         (pass_data_reorder_blocks): ...new pass_data instance and...
4653         (make_pass_reorder_blocks): ...new function.
4654         (pass_duplicate_computed_gotos): Convert from a global struct to a
4655         subclass of rtl_opt_pass along with...
4656         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
4657         (make_pass_duplicate_computed_gotos): ...new function.
4658         (pass_partition_blocks): Convert from a global struct to a subclass of
4659         rtl_opt_pass along with...
4660         (pass_data_partition_blocks): ...new pass_data instance and...
4661         (make_pass_partition_blocks): ...new function.
4662         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
4663         struct to a subclass of rtl_opt_pass along with...
4664         (pass_data_branch_target_load_optimize1): ...new pass_data instance
4665         and...
4666         (make_pass_branch_target_load_optimize1): ...new function.
4667         (pass_branch_target_load_optimize2): Convert from a global struct to a
4668         subclass of rtl_opt_pass along with...
4669         (pass_data_branch_target_load_optimize2): ...new pass_data instance
4670         and...
4671         (make_pass_branch_target_load_optimize2): ...new function.
4672         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
4673         of rtl_opt_pass along with...
4674         (pass_data_jump): ...new pass_data instance and...
4675         (make_pass_jump): ...new function.
4676         (pass_jump2): Convert from a global struct to a subclass of
4677         rtl_opt_pass along with...
4678         (pass_data_jump2): ...new pass_data instance and...
4679         (make_pass_jump2): ...new function.
4680         * cfgexpand.c (pass_expand): Convert from a global struct to a
4681         subclass of rtl_opt_pass along with...
4682         (pass_data_expand): ...new pass_data instance and...
4683         (make_pass_expand): ...new function.
4684         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
4685         of rtl_opt_pass along with...
4686         (pass_data_free_cfg): ...new pass_data instance and...
4687         (make_pass_free_cfg): ...new function.
4688         (pass_into_cfg_layout_mode): Convert from a global struct to a
4689         subclass of rtl_opt_pass along with...
4690         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
4691         (make_pass_into_cfg_layout_mode): ...new function.
4692         (pass_outof_cfg_layout_mode): Convert from a global struct to a
4693         subclass of rtl_opt_pass along with...
4694         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
4695         (make_pass_outof_cfg_layout_mode): ...new function.
4696         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
4697         struct to a subclass of gimple_opt_pass along with...
4698         (pass_data_build_cgraph_edges): ...new pass_data instance and...
4699         (make_pass_build_cgraph_edges): ...new function.
4700         (pass_rebuild_cgraph_edges): Convert from a global struct to a
4701         subclass of gimple_opt_pass along with...
4702         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
4703         (make_pass_rebuild_cgraph_edges): ...new function.
4704         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
4705         subclass of gimple_opt_pass along with...
4706         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
4707         and...
4708         (make_pass_remove_cgraph_callee_edges): ...new function.
4709         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
4710         struct to a subclass of rtl_opt_pass along with...
4711         (pass_data_stack_adjustments): ...new pass_data instance and...
4712         (make_pass_stack_adjustments): ...new function.
4713         * combine.c (pass_combine): Convert from a global struct to a subclass
4714         of rtl_opt_pass along with...
4715         (pass_data_combine): ...new pass_data instance and...
4716         (make_pass_combine): ...new function.
4717         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
4718         global struct to a subclass of rtl_opt_pass along with...
4719         (pass_data_compare_elim_after_reload): ...new pass_data instance
4720         and...
4721         (make_pass_compare_elim_after_reload): ...new function.
4722         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
4723         of rtl_opt_pass along with...
4724         (pass_data_rtl_cprop): ...new pass_data instance and...
4725         (make_pass_rtl_cprop): ...new function.
4726         * cse.c (pass_cse): Convert from a global struct to a subclass of
4727         rtl_opt_pass along with...
4728         (pass_data_cse): ...new pass_data instance and...
4729         (make_pass_cse): ...new function.
4730         (pass_cse2): Convert from a global struct to a subclass of
4731         rtl_opt_pass along with...
4732         (pass_data_cse2): ...new pass_data instance and...
4733         (make_pass_cse2): ...new function.
4734         (pass_cse_after_global_opts): Convert from a global struct to a
4735         subclass of rtl_opt_pass along with...
4736         (pass_data_cse_after_global_opts): ...new pass_data instance and...
4737         (make_pass_cse_after_global_opts): ...new function.
4738         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
4739         of rtl_opt_pass along with...
4740         (pass_data_ud_rtl_dce): ...new pass_data instance and...
4741         (make_pass_ud_rtl_dce): ...new function.
4742         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
4743         rtl_opt_pass along with...
4744         (pass_data_fast_rtl_dce): ...new pass_data instance and...
4745         (make_pass_fast_rtl_dce): ...new function.
4746         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
4747         a subclass of rtl_opt_pass along with...
4748         (pass_data_df_initialize_opt): ...new pass_data instance and...
4749         (make_pass_df_initialize_opt): ...new function.
4750         (pass_df_initialize_no_opt): Convert from a global struct to a
4751         subclass of rtl_opt_pass along with...
4752         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
4753         (make_pass_df_initialize_no_opt): ...new function.
4754         (pass_df_finish): Convert from a global struct to a subclass of
4755         rtl_opt_pass along with...
4756         (pass_data_df_finish): ...new pass_data instance and...
4757         (make_pass_df_finish): ...new function.
4758         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
4759         rtl_opt_pass along with...
4760         (pass_data_rtl_dse1): ...new pass_data instance and...
4761         (make_pass_rtl_dse1): ...new function.
4762         (pass_rtl_dse2): Convert from a global struct to a subclass of
4763         rtl_opt_pass along with...
4764         (pass_data_rtl_dse2): ...new pass_data instance and...
4765         (make_pass_rtl_dse2): ...new function.
4766         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
4767         subclass of rtl_opt_pass along with...
4768         (pass_data_dwarf2_frame): ...new pass_data instance and...
4769         (make_pass_dwarf2_frame): ...new function.
4770         * except.c (pass_set_nothrow_function_flags): Convert from a global
4771         struct to a subclass of rtl_opt_pass along with...
4772         (pass_data_set_nothrow_function_flags): ...new pass_data instance
4773         and...
4774         (make_pass_set_nothrow_function_flags): ...new function.
4775         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
4776         subclass of rtl_opt_pass along with...
4777         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
4778         and...
4779         (make_pass_convert_to_eh_region_ranges): ...new function.
4780         * final.c (pass_compute_alignments): Convert from a global struct to a
4781         subclass of rtl_opt_pass along with...
4782         (pass_data_compute_alignments): ...new pass_data instance and...
4783         (make_pass_compute_alignments): ...new function.
4784         (pass_final): Convert from a global struct to a subclass of
4785         rtl_opt_pass along with...
4786         (pass_data_final): ...new pass_data instance and...
4787         (make_pass_final): ...new function.
4788         (pass_shorten_branches): Convert from a global struct to a subclass of
4789         rtl_opt_pass along with...
4790         (pass_data_shorten_branches): ...new pass_data instance and...
4791         (make_pass_shorten_branches): ...new function.
4792         (pass_clean_state): Convert from a global struct to a subclass of
4793         rtl_opt_pass along with...
4794         (pass_data_clean_state): ...new pass_data instance and...
4795         (make_pass_clean_state): ...new function.
4796         * function.c (pass_instantiate_virtual_regs): Convert from a global
4797         struct to a subclass of rtl_opt_pass along with...
4798         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
4799         (make_pass_instantiate_virtual_regs): ...new function.
4800         (pass_leaf_regs): Convert from a global struct to a subclass of
4801         rtl_opt_pass along with...
4802         (pass_data_leaf_regs): ...new pass_data instance and...
4803         (make_pass_leaf_regs): ...new function.
4804         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
4805         subclass of rtl_opt_pass along with...
4806         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
4807         and...
4808         (make_pass_thread_prologue_and_epilogue): ...new function.
4809         (pass_match_asm_constraints): Convert from a global struct to a
4810         subclass of rtl_opt_pass along with...
4811         (pass_data_match_asm_constraints): ...new pass_data instance and...
4812         (make_pass_match_asm_constraints): ...new function.
4813         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
4814         subclass of rtl_opt_pass along with...
4815         (pass_data_rtl_fwprop): ...new pass_data instance and...
4816         (make_pass_rtl_fwprop): ...new function.
4817         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
4818         rtl_opt_pass along with...
4819         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
4820         (make_pass_rtl_fwprop_addr): ...new function.
4821         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
4822         rtl_opt_pass along with...
4823         (pass_data_rtl_pre): ...new pass_data instance and...
4824         (make_pass_rtl_pre): ...new function.
4825         (pass_rtl_hoist): Convert from a global struct to a subclass of
4826         rtl_opt_pass along with...
4827         (pass_data_rtl_hoist): ...new pass_data instance and...
4828         (make_pass_rtl_hoist): ...new function.
4829         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
4830         subclass of gimple_opt_pass along with...
4831         (pass_data_lower_cf): ...new pass_data instance and...
4832         (make_pass_lower_cf): ...new function.
4833         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
4834         from a global struct to a subclass of gimple_opt_pass along with...
4835         (pass_data_strength_reduction): ...new pass_data instance and...
4836         (make_pass_strength_reduction): ...new function.
4837         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
4838         of rtl_opt_pass along with...
4839         (pass_data_rtl_ifcvt): ...new pass_data instance and...
4840         (make_pass_rtl_ifcvt): ...new function.
4841         (pass_if_after_combine): Convert from a global struct to a subclass of
4842         rtl_opt_pass along with...
4843         (pass_data_if_after_combine): ...new pass_data instance and...
4844         (make_pass_if_after_combine): ...new function.
4845         (pass_if_after_reload): Convert from a global struct to a subclass of
4846         rtl_opt_pass along with...
4847         (pass_data_if_after_reload): ...new pass_data instance and...
4848         (make_pass_if_after_reload): ...new function.
4849         * init-regs.c (pass_initialize_regs): Convert from a global struct to
4850         a subclass of rtl_opt_pass along with...
4851         (pass_data_initialize_regs): ...new pass_data instance and...
4852         (make_pass_initialize_regs): ...new function.
4853         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
4854         of ipa_opt_pass_d along with...
4855         (pass_data_ipa_cp): ...new pass_data instance and...
4856         (make_pass_ipa_cp): ...new function.
4857         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
4858         global struct to a subclass of gimple_opt_pass along with...
4859         (pass_data_inline_parameters): ...new pass_data instance and...
4860         (make_pass_inline_parameters): ...new function.
4861         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
4862         subclass of gimple_opt_pass along with...
4863         (pass_data_early_inline): ...new pass_data instance and...
4864         (make_pass_early_inline): ...new function.
4865         (pass_ipa_inline): Convert from a global struct to a subclass of
4866         ipa_opt_pass_d along with...
4867         (pass_data_ipa_inline): ...new pass_data instance and...
4868         (make_pass_ipa_inline): ...new function.
4869         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
4870         struct to a subclass of gimple_opt_pass along with...
4871         (pass_data_local_pure_const): ...new pass_data instance and...
4872         (make_pass_local_pure_const): ...new function.
4873         (pass_ipa_pure_const): Convert from a global struct to a subclass of
4874         ipa_opt_pass_d along with...
4875         (pass_data_ipa_pure_const): ...new pass_data instance and...
4876         (make_pass_ipa_pure_const): ...new function.
4877         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
4878         to a subclass of ipa_opt_pass_d along with...
4879         (pass_data_ipa_reference): ...new pass_data instance and...
4880         (make_pass_ipa_reference): ...new function.
4881         * ipa-split.c (pass_split_functions): Convert from a global struct to
4882         a subclass of gimple_opt_pass along with...
4883         (pass_data_split_functions): ...new pass_data instance and...
4884         (make_pass_split_functions): ...new function.
4885         (pass_feedback_split_functions): Convert from a global struct to a
4886         subclass of gimple_opt_pass along with...
4887         (pass_data_feedback_split_functions): ...new pass_data instance and...
4888         (make_pass_feedback_split_functions): ...new function.
4889         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
4890         global struct to a subclass of simple_ipa_opt_pass along with...
4891         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
4892         instance and...
4893         (make_pass_ipa_function_and_variable_visibility): ...new function.
4894         (pass_ipa_free_inline_summary): Convert from a global struct to a
4895         subclass of simple_ipa_opt_pass along with...
4896         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
4897         (make_pass_ipa_free_inline_summary): ...new function.
4898         (pass_ipa_whole_program_visibility): Convert from a global struct to a
4899         subclass of ipa_opt_pass_d along with...
4900         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
4901         and...
4902         (make_pass_ipa_whole_program_visibility): ...new function.
4903         (pass_ipa_profile): Convert from a global struct to a subclass of
4904         ipa_opt_pass_d along with...
4905         (pass_data_ipa_profile): ...new pass_data instance and...
4906         (make_pass_ipa_profile): ...new function.
4907         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
4908         ipa_opt_pass_d along with...
4909         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
4910         (make_pass_ipa_cdtor_merge): ...new function.
4911         * ira.c (pass_ira): Convert from a global struct to a subclass of
4912         rtl_opt_pass along with...
4913         (pass_data_ira): ...new pass_data instance and...
4914         (make_pass_ira): ...new function.
4915         (pass_reload): Convert from a global struct to a subclass of
4916         rtl_opt_pass along with...
4917         (pass_data_reload): ...new pass_data instance and...
4918         (make_pass_reload): ...new function.
4919         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
4920         subclass of rtl_opt_pass along with...
4921         (pass_data_cleanup_barriers): ...new pass_data instance and...
4922         (make_pass_cleanup_barriers): ...new function.
4923         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
4924         of rtl_opt_pass along with...
4925         (pass_data_loop2): ...new pass_data instance and...
4926         (make_pass_loop2): ...new function.
4927         (pass_rtl_loop_init): Convert from a global struct to a subclass of
4928         rtl_opt_pass along with...
4929         (pass_data_rtl_loop_init): ...new pass_data instance and...
4930         (make_pass_rtl_loop_init): ...new function.
4931         (pass_rtl_loop_done): Convert from a global struct to a subclass of
4932         rtl_opt_pass along with...
4933         (pass_data_rtl_loop_done): ...new pass_data instance and...
4934         (make_pass_rtl_loop_done): ...new function.
4935         (pass_rtl_move_loop_invariants): Convert from a global struct to a
4936         subclass of rtl_opt_pass along with...
4937         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
4938         (make_pass_rtl_move_loop_invariants): ...new function.
4939         (pass_rtl_unswitch): Convert from a global struct to a subclass of
4940         rtl_opt_pass along with...
4941         (pass_data_rtl_unswitch): ...new pass_data instance and...
4942         (make_pass_rtl_unswitch): ...new function.
4943         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
4944         subclass of rtl_opt_pass along with...
4945         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
4946         and...
4947         (make_pass_rtl_unroll_and_peel_loops): ...new function.
4948         (pass_rtl_doloop): Convert from a global struct to a subclass of
4949         rtl_opt_pass along with...
4950         (pass_data_rtl_doloop): ...new pass_data instance and...
4951         (make_pass_rtl_doloop): ...new function.
4952         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
4953         a subclass of rtl_opt_pass along with...
4954         (pass_data_lower_subreg): ...new pass_data instance and...
4955         (make_pass_lower_subreg): ...new function.
4956         (pass_lower_subreg2): Convert from a global struct to a subclass of
4957         rtl_opt_pass along with...
4958         (pass_data_lower_subreg2): ...new pass_data instance and...
4959         (make_pass_lower_subreg2): ...new function.
4960         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
4961         struct to a subclass of ipa_opt_pass_d along with...
4962         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
4963         (make_pass_ipa_lto_gimple_out): ...new function.
4964         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
4965         of ipa_opt_pass_d along with...
4966         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
4967         (make_pass_ipa_lto_finish_out): ...new function.
4968         * mode-switching.c (pass_mode_switching): Convert from a global struct
4969         to a subclass of rtl_opt_pass along with...
4970         (pass_data_mode_switching): ...new pass_data instance and...
4971         (make_pass_mode_switching): ...new function.
4972         * modulo-sched.c (pass_sms): Convert from a global struct to a
4973         subclass of rtl_opt_pass along with...
4974         (pass_data_sms): ...new pass_data instance and...
4975         (make_pass_sms): ...new function.
4976         * omp-low.c (pass_expand_omp): Convert from a global struct to a
4977         subclass of gimple_opt_pass along with...
4978         (pass_data_expand_omp): ...new pass_data instance and...
4979         (make_pass_expand_omp): ...new function.
4980         (pass_lower_omp): Convert from a global struct to a subclass of
4981         gimple_opt_pass along with...
4982         (pass_data_lower_omp): ...new pass_data instance and...
4983         (make_pass_lower_omp): ...new function.
4984         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
4985         of gimple_opt_pass along with...
4986         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
4987         (make_pass_diagnose_omp_blocks): ...new function.
4988         * passes.c (pass_early_local_passes): Convert from a global struct to
4989         a subclass of simple_ipa_opt_pass along with...
4990         (pass_data_early_local_passes): ...new pass_data instance and...
4991         (make_pass_early_local_passes): ...new function.
4992         (pass_all_early_optimizations): Convert from a global struct to a
4993         subclass of gimple_opt_pass along with...
4994         (pass_data_all_early_optimizations): ...new pass_data instance and...
4995         (make_pass_all_early_optimizations): ...new function.
4996         (pass_all_optimizations): Convert from a global struct to a subclass
4997         of gimple_opt_pass along with...
4998         (pass_data_all_optimizations): ...new pass_data instance and...
4999         (make_pass_all_optimizations): ...new function.
5000         (pass_all_optimizations_g): Convert from a global struct to a subclass
5001         of gimple_opt_pass along with...
5002         (pass_data_all_optimizations_g): ...new pass_data instance and...
5003         (make_pass_all_optimizations_g): ...new function.
5004         (pass_rest_of_compilation): Convert from a global struct to a subclass
5005         of rtl_opt_pass along with...
5006         (pass_data_rest_of_compilation): ...new pass_data instance and...
5007         (make_pass_rest_of_compilation): ...new function.
5008         (pass_postreload): Convert from a global struct to a subclass of
5009         rtl_opt_pass along with...
5010         (pass_data_postreload): ...new pass_data instance and...
5011         (make_pass_postreload): ...new function.
5012         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
5013         subclass of rtl_opt_pass along with...
5014         (pass_data_gcse2): ...new pass_data instance and...
5015         (make_pass_gcse2): ...new function.
5016         * postreload.c (pass_postreload_cse): Convert from a global struct to
5017         a subclass of rtl_opt_pass along with...
5018         (pass_data_postreload_cse): ...new pass_data instance and...
5019         (make_pass_postreload_cse): ...new function.
5020         * predict.c (pass_profile): Convert from a global struct to a subclass
5021         of gimple_opt_pass along with...
5022         (pass_data_profile): ...new pass_data instance and...
5023         (make_pass_profile): ...new function.
5024         (pass_strip_predict_hints): Convert from a global struct to a subclass
5025         of gimple_opt_pass along with...
5026         (pass_data_strip_predict_hints): ...new pass_data instance and...
5027         (make_pass_strip_predict_hints): ...new function.
5028         * recog.c (pass_peephole2): Convert from a global struct to a subclass
5029         of rtl_opt_pass along with...
5030         (pass_data_peephole2): ...new pass_data instance and...
5031         (make_pass_peephole2): ...new function.
5032         (pass_split_all_insns): Convert from a global struct to a subclass of
5033         rtl_opt_pass along with...
5034         (pass_data_split_all_insns): ...new pass_data instance and...
5035         (make_pass_split_all_insns): ...new function.
5036         (pass_split_after_reload): Convert from a global struct to a subclass
5037         of rtl_opt_pass along with...
5038         (pass_data_split_after_reload): ...new pass_data instance and...
5039         (make_pass_split_after_reload): ...new function.
5040         (pass_split_before_regstack): Convert from a global struct to a
5041         subclass of rtl_opt_pass along with...
5042         (pass_data_split_before_regstack): ...new pass_data instance and...
5043         (make_pass_split_before_regstack): ...new function.
5044         (pass_split_before_sched2): Convert from a global struct to a subclass
5045         of rtl_opt_pass along with...
5046         (pass_data_split_before_sched2): ...new pass_data instance and...
5047         (make_pass_split_before_sched2): ...new function.
5048         (pass_split_for_shorten_branches): Convert from a global struct to a
5049         subclass of rtl_opt_pass along with...
5050         (pass_data_split_for_shorten_branches): ...new pass_data instance
5051         and...
5052         (make_pass_split_for_shorten_branches): ...new function.
5053         * ree.c (pass_ree): Convert from a global struct to a subclass of
5054         rtl_opt_pass along with...
5055         (pass_data_ree): ...new pass_data instance and...
5056         (make_pass_ree): ...new function.
5057         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
5058         subclass of rtl_opt_pass along with...
5059         (pass_data_stack_regs): ...new pass_data instance and...
5060         (make_pass_stack_regs): ...new function.
5061         (pass_stack_regs_run): Convert from a global struct to a subclass of
5062         rtl_opt_pass along with...
5063         (pass_data_stack_regs_run): ...new pass_data instance and...
5064         (make_pass_stack_regs_run): ...new function.
5065         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
5066         subclass of rtl_opt_pass along with...
5067         (pass_data_cprop_hardreg): ...new pass_data instance and...
5068         (make_pass_cprop_hardreg): ...new function.
5069         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
5070         subclass of rtl_opt_pass along with...
5071         (pass_data_reginfo_init): ...new pass_data instance and...
5072         (make_pass_reginfo_init): ...new function.
5073         * regmove.c (pass_regmove): Convert from a global struct to a subclass
5074         of rtl_opt_pass along with...
5075         (pass_data_regmove): ...new pass_data instance and...
5076         (make_pass_regmove): ...new function.
5077         * regrename.c (pass_regrename): Convert from a global struct to a
5078         subclass of rtl_opt_pass along with...
5079         (pass_data_regrename): ...new pass_data instance and...
5080         (make_pass_regrename): ...new function.
5081         * reorg.c (pass_delay_slots): Convert from a global struct to a
5082         subclass of rtl_opt_pass along with...
5083         (pass_data_delay_slots): ...new pass_data instance and...
5084         (make_pass_delay_slots): ...new function.
5085         (pass_machine_reorg): Convert from a global struct to a subclass of
5086         rtl_opt_pass along with...
5087         (pass_data_machine_reorg): ...new pass_data instance and...
5088         (make_pass_machine_reorg): ...new function.
5089         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
5090         of rtl_opt_pass along with...
5091         (pass_data_sched): ...new pass_data instance and...
5092         (make_pass_sched): ...new function.
5093         (pass_sched2): Convert from a global struct to a subclass of
5094         rtl_opt_pass along with...
5095         (pass_data_sched2): ...new pass_data instance and...
5096         (make_pass_sched2): ...new function.
5097         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
5098         to a subclass of rtl_opt_pass along with...
5099         (pass_data_stack_ptr_mod): ...new pass_data instance and...
5100         (make_pass_stack_ptr_mod): ...new function.
5101         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
5102         to a subclass of rtl_opt_pass along with...
5103         (pass_data_rtl_store_motion): ...new pass_data instance and...
5104         (make_pass_rtl_store_motion): ...new function.
5105         * tracer.c (pass_tracer): Convert from a global struct to a subclass
5106         of gimple_opt_pass along with...
5107         (pass_data_tracer): ...new pass_data instance and...
5108         (make_pass_tracer): ...new function.
5109         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
5110         to a subclass of gimple_opt_pass along with...
5111         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
5112         (make_pass_diagnose_tm_blocks): ...new function.
5113         (pass_lower_tm): Convert from a global struct to a subclass of
5114         gimple_opt_pass along with...
5115         (pass_data_lower_tm): ...new pass_data instance and...
5116         (make_pass_lower_tm): ...new function.
5117         (pass_tm_init): Convert from a global struct to a subclass of
5118         gimple_opt_pass along with...
5119         (pass_data_tm_init): ...new pass_data instance and...
5120         (make_pass_tm_init): ...new function.
5121         (pass_tm_mark): Convert from a global struct to a subclass of
5122         gimple_opt_pass along with...
5123         (pass_data_tm_mark): ...new pass_data instance and...
5124         (make_pass_tm_mark): ...new function.
5125         (pass_tm_edges): Convert from a global struct to a subclass of
5126         gimple_opt_pass along with...
5127         (pass_data_tm_edges): ...new pass_data instance and...
5128         (make_pass_tm_edges): ...new function.
5129         (pass_tm_memopt): Convert from a global struct to a subclass of
5130         gimple_opt_pass along with...
5131         (pass_data_tm_memopt): ...new pass_data instance and...
5132         (make_pass_tm_memopt): ...new function.
5133         (pass_ipa_tm): Convert from a global struct to a subclass of
5134         simple_ipa_opt_pass along with...
5135         (pass_data_ipa_tm): ...new pass_data instance and...
5136         (make_pass_ipa_tm): ...new function.
5137         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
5138         subclass of gimple_opt_pass along with...
5139         (pass_data_call_cdce): ...new pass_data instance and...
5140         (make_pass_call_cdce): ...new function.
5141         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
5142         subclass of gimple_opt_pass along with...
5143         (pass_data_build_cfg): ...new pass_data instance and...
5144         (make_pass_build_cfg): ...new function.
5145         (pass_split_crit_edges): Convert from a global struct to a subclass of
5146         gimple_opt_pass along with...
5147         (pass_data_split_crit_edges): ...new pass_data instance and...
5148         (make_pass_split_crit_edges): ...new function.
5149         (pass_warn_function_return): Convert from a global struct to a
5150         subclass of gimple_opt_pass along with...
5151         (pass_data_warn_function_return): ...new pass_data instance and...
5152         (make_pass_warn_function_return): ...new function.
5153         (pass_warn_function_noreturn): Convert from a global struct to a
5154         subclass of gimple_opt_pass along with...
5155         (pass_data_warn_function_noreturn): ...new pass_data instance and...
5156         (make_pass_warn_function_noreturn): ...new function.
5157         (pass_warn_unused_result): Convert from a global struct to a subclass
5158         of gimple_opt_pass along with...
5159         (pass_data_warn_unused_result): ...new pass_data instance and...
5160         (make_pass_warn_unused_result): ...new function.
5161         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
5162         a subclass of gimple_opt_pass along with...
5163         (pass_data_merge_phi): ...new pass_data instance and...
5164         (make_pass_merge_phi): ...new function.
5165         * tree-complex.c (pass_lower_complex): Convert from a global struct to
5166         a subclass of gimple_opt_pass along with...
5167         (pass_data_lower_complex): ...new pass_data instance and...
5168         (make_pass_lower_complex): ...new function.
5169         (pass_lower_complex_O0): Convert from a global struct to a subclass of
5170         gimple_opt_pass along with...
5171         (pass_data_lower_complex_O0): ...new pass_data instance and...
5172         (make_pass_lower_complex_O0): ...new function.
5173         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
5174         subclass of gimple_opt_pass along with...
5175         (pass_data_lower_eh): ...new pass_data instance and...
5176         (make_pass_lower_eh): ...new function.
5177         (pass_refactor_eh): Convert from a global struct to a subclass of
5178         gimple_opt_pass along with...
5179         (pass_data_refactor_eh): ...new pass_data instance and...
5180         (make_pass_refactor_eh): ...new function.
5181         (pass_lower_resx): Convert from a global struct to a subclass of
5182         gimple_opt_pass along with...
5183         (pass_data_lower_resx): ...new pass_data instance and...
5184         (make_pass_lower_resx): ...new function.
5185         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
5186         of gimple_opt_pass along with...
5187         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
5188         (make_pass_lower_eh_dispatch): ...new function.
5189         (pass_cleanup_eh): Convert from a global struct to a subclass of
5190         gimple_opt_pass along with...
5191         (pass_data_cleanup_eh): ...new pass_data instance and...
5192         (make_pass_cleanup_eh): ...new function.
5193         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
5194         to a subclass of simple_ipa_opt_pass along with...
5195         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
5196         (make_pass_ipa_lower_emutls): ...new function.
5197         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
5198         a subclass of gimple_opt_pass along with...
5199         (pass_data_if_conversion): ...new pass_data instance and...
5200         (make_pass_if_conversion): ...new function.
5201         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
5202         subclass of gimple_opt_pass along with...
5203         (pass_data_build_ssa): ...new pass_data instance and...
5204         (make_pass_build_ssa): ...new function.
5205         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
5206         global struct to a subclass of gimple_opt_pass along with...
5207         (pass_data_loop_distribution): ...new pass_data instance and...
5208         (make_pass_loop_distribution): ...new function.
5209         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
5210         subclass of gimple_opt_pass along with...
5211         (pass_data_mudflap_1): ...new pass_data instance and...
5212         (make_pass_mudflap_1): ...new function.
5213         (pass_mudflap_2): Convert from a global struct to a subclass of
5214         gimple_opt_pass along with...
5215         (pass_data_mudflap_2): ...new pass_data instance and...
5216         (make_pass_mudflap_2): ...new function.
5217         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
5218         subclass of gimple_opt_pass along with...
5219         (pass_data_mudflap_1): ...new pass_data instance and...
5220         (make_pass_mudflap_1): ...new function.
5221         (pass_mudflap_2): Convert from a global struct to a subclass of
5222         gimple_opt_pass along with...
5223         (pass_data_mudflap_2): ...new pass_data instance and...
5224         (make_pass_mudflap_2): ...new function.
5225         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
5226         gimple_opt_pass along with...
5227         (pass_data_nrv): ...new pass_data instance and...
5228         (make_pass_nrv): ...new function.
5229         (pass_return_slot): Convert from a global struct to a subclass of
5230         gimple_opt_pass along with...
5231         (pass_data_return_slot): ...new pass_data instance and...
5232         (make_pass_return_slot): ...new function.
5233         * tree-object-size.c (pass_object_sizes): Convert from a global struct
5234         to a subclass of gimple_opt_pass along with...
5235         (pass_data_object_sizes): ...new pass_data instance and...
5236         (make_pass_object_sizes): ...new function.
5237         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
5238         global struct to a subclass of gimple_opt_pass along with...
5239         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
5240         and...
5241         (make_pass_cleanup_cfg_post_optimizing): ...new function.
5242         (pass_fixup_cfg): Convert from a global struct to a subclass of
5243         gimple_opt_pass along with...
5244         (pass_data_fixup_cfg): ...new pass_data instance and...
5245         (make_pass_fixup_cfg): ...new function.
5246         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
5247         (make_pass_mudflap_1): ...new function.
5248         (pass_mudflap_2): Replace declaration with that of...
5249         (make_pass_mudflap_2): ...new function.
5250         (pass_asan): Replace declaration with that of...
5251         (make_pass_asan): ...new function.
5252         (pass_asan_O0): Replace declaration with that of...
5253         (make_pass_asan_O0): ...new function.
5254         (pass_tsan): Replace declaration with that of...
5255         (make_pass_tsan): ...new function.
5256         (pass_tsan_O0): Replace declaration with that of...
5257         (make_pass_tsan_O0): ...new function.
5258         (pass_lower_cf): Replace declaration with that of...
5259         (make_pass_lower_cf): ...new function.
5260         (pass_refactor_eh): Replace declaration with that of...
5261         (make_pass_refactor_eh): ...new function.
5262         (pass_lower_eh): Replace declaration with that of...
5263         (make_pass_lower_eh): ...new function.
5264         (pass_lower_eh_dispatch): Replace declaration with that of...
5265         (make_pass_lower_eh_dispatch): ...new function.
5266         (pass_lower_resx): Replace declaration with that of...
5267         (make_pass_lower_resx): ...new function.
5268         (pass_build_cfg): Replace declaration with that of...
5269         (make_pass_build_cfg): ...new function.
5270         (pass_early_tree_profile): Replace declaration with that of...
5271         (make_pass_early_tree_profile): ...new function.
5272         (pass_cleanup_eh): Replace declaration with that of...
5273         (make_pass_cleanup_eh): ...new function.
5274         (pass_sra): Replace declaration with that of...
5275         (make_pass_sra): ...new function.
5276         (pass_sra_early): Replace declaration with that of...
5277         (make_pass_sra_early): ...new function.
5278         (pass_early_ipa_sra): Replace declaration with that of...
5279         (make_pass_early_ipa_sra): ...new function.
5280         (pass_tail_recursion): Replace declaration with that of...
5281         (make_pass_tail_recursion): ...new function.
5282         (pass_tail_calls): Replace declaration with that of...
5283         (make_pass_tail_calls): ...new function.
5284         (pass_tree_loop): Replace declaration with that of...
5285         (make_pass_tree_loop): ...new function.
5286         (pass_tree_loop_init): Replace declaration with that of...
5287         (make_pass_tree_loop_init): ...new function.
5288         (pass_lim): Replace declaration with that of...
5289         (make_pass_lim): ...new function.
5290         (pass_tree_unswitch): Replace declaration with that of...
5291         (make_pass_tree_unswitch): ...new function.
5292         (pass_predcom): Replace declaration with that of...
5293         (make_pass_predcom): ...new function.
5294         (pass_iv_canon): Replace declaration with that of...
5295         (make_pass_iv_canon): ...new function.
5296         (pass_scev_cprop): Replace declaration with that of...
5297         (make_pass_scev_cprop): ...new function.
5298         (pass_empty_loop): Replace declaration with that of...
5299         (make_pass_empty_loop): ...new function.
5300         (pass_record_bounds): Replace declaration with that of...
5301         (make_pass_record_bounds): ...new function.
5302         (pass_graphite): Replace declaration with that of...
5303         (make_pass_graphite): ...new function.
5304         (pass_graphite_transforms): Replace declaration with that of...
5305         (make_pass_graphite_transforms): ...new function.
5306         (pass_if_conversion): Replace declaration with that of...
5307         (make_pass_if_conversion): ...new function.
5308         (pass_loop_distribution): Replace declaration with that of...
5309         (make_pass_loop_distribution): ...new function.
5310         (pass_vectorize): Replace declaration with that of...
5311         (make_pass_vectorize): ...new function.
5312         (pass_slp_vectorize): Replace declaration with that of...
5313         (make_pass_slp_vectorize): ...new function.
5314         (pass_complete_unroll): Replace declaration with that of...
5315         (make_pass_complete_unroll): ...new function.
5316         (pass_complete_unrolli): Replace declaration with that of...
5317         (make_pass_complete_unrolli): ...new function.
5318         (pass_parallelize_loops): Replace declaration with that of...
5319         (make_pass_parallelize_loops): ...new function.
5320         (pass_loop_prefetch): Replace declaration with that of...
5321         (make_pass_loop_prefetch): ...new function.
5322         (pass_iv_optimize): Replace declaration with that of...
5323         (make_pass_iv_optimize): ...new function.
5324         (pass_tree_loop_done): Replace declaration with that of...
5325         (make_pass_tree_loop_done): ...new function.
5326         (pass_ch): Replace declaration with that of...
5327         (make_pass_ch): ...new function.
5328         (pass_ccp): Replace declaration with that of...
5329         (make_pass_ccp): ...new function.
5330         (pass_phi_only_cprop): Replace declaration with that of...
5331         (make_pass_phi_only_cprop): ...new function.
5332         (pass_build_ssa): Replace declaration with that of...
5333         (make_pass_build_ssa): ...new function.
5334         (pass_build_alias): Replace declaration with that of...
5335         (make_pass_build_alias): ...new function.
5336         (pass_build_ealias): Replace declaration with that of...
5337         (make_pass_build_ealias): ...new function.
5338         (pass_dominator): Replace declaration with that of...
5339         (make_pass_dominator): ...new function.
5340         (pass_dce): Replace declaration with that of...
5341         (make_pass_dce): ...new function.
5342         (pass_dce_loop): Replace declaration with that of...
5343         (make_pass_dce_loop): ...new function.
5344         (pass_cd_dce): Replace declaration with that of...
5345         (make_pass_cd_dce): ...new function.
5346         (pass_call_cdce): Replace declaration with that of...
5347         (make_pass_call_cdce): ...new function.
5348         (pass_merge_phi): Replace declaration with that of...
5349         (make_pass_merge_phi): ...new function.
5350         (pass_split_crit_edges): Replace declaration with that of...
5351         (make_pass_split_crit_edges): ...new function.
5352         (pass_pre): Replace declaration with that of...
5353         (make_pass_pre): ...new function.
5354         (pass_profile): Replace declaration with that of...
5355         (make_pass_profile): ...new function.
5356         (pass_strip_predict_hints): Replace declaration with that of...
5357         (make_pass_strip_predict_hints): ...new function.
5358         (pass_lower_complex_O0): Replace declaration with that of...
5359         (make_pass_lower_complex_O0): ...new function.
5360         (pass_lower_complex): Replace declaration with that of...
5361         (make_pass_lower_complex): ...new function.
5362         (pass_lower_vector): Replace declaration with that of...
5363         (make_pass_lower_vector): ...new function.
5364         (pass_lower_vector_ssa): Replace declaration with that of...
5365         (make_pass_lower_vector_ssa): ...new function.
5366         (pass_lower_omp): Replace declaration with that of...
5367         (make_pass_lower_omp): ...new function.
5368         (pass_diagnose_omp_blocks): Replace declaration with that of...
5369         (make_pass_diagnose_omp_blocks): ...new function.
5370         (pass_expand_omp): Replace declaration with that of...
5371         (make_pass_expand_omp): ...new function.
5372         (pass_expand_omp_ssa): Replace declaration with that of...
5373         (make_pass_expand_omp_ssa): ...new function.
5374         (pass_object_sizes): Replace declaration with that of...
5375         (make_pass_object_sizes): ...new function.
5376         (pass_strlen): Replace declaration with that of...
5377         (make_pass_strlen): ...new function.
5378         (pass_fold_builtins): Replace declaration with that of...
5379         (make_pass_fold_builtins): ...new function.
5380         (pass_stdarg): Replace declaration with that of...
5381         (make_pass_stdarg): ...new function.
5382         (pass_early_warn_uninitialized): Replace declaration with that of...
5383         (make_pass_early_warn_uninitialized): ...new function.
5384         (pass_late_warn_uninitialized): Replace declaration with that of...
5385         (make_pass_late_warn_uninitialized): ...new function.
5386         (pass_cse_reciprocals): Replace declaration with that of...
5387         (make_pass_cse_reciprocals): ...new function.
5388         (pass_cse_sincos): Replace declaration with that of...
5389         (make_pass_cse_sincos): ...new function.
5390         (pass_optimize_bswap): Replace declaration with that of...
5391         (make_pass_optimize_bswap): ...new function.
5392         (pass_optimize_widening_mul): Replace declaration with that of...
5393         (make_pass_optimize_widening_mul): ...new function.
5394         (pass_warn_function_return): Replace declaration with that of...
5395         (make_pass_warn_function_return): ...new function.
5396         (pass_warn_function_noreturn): Replace declaration with that of...
5397         (make_pass_warn_function_noreturn): ...new function.
5398         (pass_cselim): Replace declaration with that of...
5399         (make_pass_cselim): ...new function.
5400         (pass_phiopt): Replace declaration with that of...
5401         (make_pass_phiopt): ...new function.
5402         (pass_forwprop): Replace declaration with that of...
5403         (make_pass_forwprop): ...new function.
5404         (pass_phiprop): Replace declaration with that of...
5405         (make_pass_phiprop): ...new function.
5406         (pass_tree_ifcombine): Replace declaration with that of...
5407         (make_pass_tree_ifcombine): ...new function.
5408         (pass_dse): Replace declaration with that of...
5409         (make_pass_dse): ...new function.
5410         (pass_nrv): Replace declaration with that of...
5411         (make_pass_nrv): ...new function.
5412         (pass_rename_ssa_copies): Replace declaration with that of...
5413         (make_pass_rename_ssa_copies): ...new function.
5414         (pass_sink_code): Replace declaration with that of...
5415         (make_pass_sink_code): ...new function.
5416         (pass_fre): Replace declaration with that of...
5417         (make_pass_fre): ...new function.
5418         (pass_check_data_deps): Replace declaration with that of...
5419         (make_pass_check_data_deps): ...new function.
5420         (pass_copy_prop): Replace declaration with that of...
5421         (make_pass_copy_prop): ...new function.
5422         (pass_vrp): Replace declaration with that of...
5423         (make_pass_vrp): ...new function.
5424         (pass_uncprop): Replace declaration with that of...
5425         (make_pass_uncprop): ...new function.
5426         (pass_return_slot): Replace declaration with that of...
5427         (make_pass_return_slot): ...new function.
5428         (pass_reassoc): Replace declaration with that of...
5429         (make_pass_reassoc): ...new function.
5430         (pass_rebuild_cgraph_edges): Replace declaration with that of...
5431         (make_pass_rebuild_cgraph_edges): ...new function.
5432         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
5433         (make_pass_remove_cgraph_callee_edges): ...new function.
5434         (pass_build_cgraph_edges): Replace declaration with that of...
5435         (make_pass_build_cgraph_edges): ...new function.
5436         (pass_local_pure_const): Replace declaration with that of...
5437         (make_pass_local_pure_const): ...new function.
5438         (pass_tracer): Replace declaration with that of...
5439         (make_pass_tracer): ...new function.
5440         (pass_warn_unused_result): Replace declaration with that of...
5441         (make_pass_warn_unused_result): ...new function.
5442         (pass_diagnose_tm_blocks): Replace declaration with that of...
5443         (make_pass_diagnose_tm_blocks): ...new function.
5444         (pass_lower_tm): Replace declaration with that of...
5445         (make_pass_lower_tm): ...new function.
5446         (pass_tm_init): Replace declaration with that of...
5447         (make_pass_tm_init): ...new function.
5448         (pass_tm_mark): Replace declaration with that of...
5449         (make_pass_tm_mark): ...new function.
5450         (pass_tm_memopt): Replace declaration with that of...
5451         (make_pass_tm_memopt): ...new function.
5452         (pass_tm_edges): Replace declaration with that of...
5453         (make_pass_tm_edges): ...new function.
5454         (pass_split_functions): Replace declaration with that of...
5455         (make_pass_split_functions): ...new function.
5456         (pass_feedback_split_functions): Replace declaration with that of...
5457         (make_pass_feedback_split_functions): ...new function.
5458         (pass_strength_reduction): Replace declaration with that of...
5459         (make_pass_strength_reduction): ...new function.
5460         (pass_ipa_lower_emutls): Replace declaration with that of...
5461         (make_pass_ipa_lower_emutls): ...new function.
5462         (pass_ipa_function_and_variable_visibility): Replace declaration with
5463         that of...
5464         (make_pass_ipa_function_and_variable_visibility): ...new function.
5465         (pass_ipa_tree_profile): Replace declaration with that of...
5466         (make_pass_ipa_tree_profile): ...new function.
5467         (pass_early_local_passes): Replace declaration with that of...
5468         (make_pass_early_local_passes): ...new function.
5469         (pass_ipa_whole_program_visibility): Replace declaration with that
5470         of...
5471         (make_pass_ipa_whole_program_visibility): ...new function.
5472         (pass_ipa_lto_gimple_out): Replace declaration with that of...
5473         (make_pass_ipa_lto_gimple_out): ...new function.
5474         (pass_ipa_increase_alignment): Replace declaration with that of...
5475         (make_pass_ipa_increase_alignment): ...new function.
5476         (pass_ipa_inline): Replace declaration with that of...
5477         (make_pass_ipa_inline): ...new function.
5478         (pass_ipa_free_lang_data): Replace declaration with that of...
5479         (make_pass_ipa_free_lang_data): ...new function.
5480         (pass_ipa_free_inline_summary): Replace declaration with that of...
5481         (make_pass_ipa_free_inline_summary): ...new function.
5482         (pass_ipa_cp): Replace declaration with that of...
5483         (make_pass_ipa_cp): ...new function.
5484         (pass_ipa_reference): Replace declaration with that of...
5485         (make_pass_ipa_reference): ...new function.
5486         (pass_ipa_pure_const): Replace declaration with that of...
5487         (make_pass_ipa_pure_const): ...new function.
5488         (pass_ipa_pta): Replace declaration with that of...
5489         (make_pass_ipa_pta): ...new function.
5490         (pass_ipa_lto_finish_out): Replace declaration with that of...
5491         (make_pass_ipa_lto_finish_out): ...new function.
5492         (pass_ipa_tm): Replace declaration with that of...
5493         (make_pass_ipa_tm): ...new function.
5494         (pass_ipa_profile): Replace declaration with that of...
5495         (make_pass_ipa_profile): ...new function.
5496         (pass_ipa_cdtor_merge): Replace declaration with that of...
5497         (make_pass_ipa_cdtor_merge): ...new function.
5498         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
5499         of...
5500         (make_pass_cleanup_cfg_post_optimizing): ...new function.
5501         (pass_init_datastructures): Replace declaration with that of...
5502         (make_pass_init_datastructures): ...new function.
5503         (pass_fixup_cfg): Replace declaration with that of...
5504         (make_pass_fixup_cfg): ...new function.
5505         (pass_expand): Replace declaration with that of...
5506         (make_pass_expand): ...new function.
5507         (pass_instantiate_virtual_regs): Replace declaration with that of...
5508         (make_pass_instantiate_virtual_regs): ...new function.
5509         (pass_rtl_fwprop): Replace declaration with that of...
5510         (make_pass_rtl_fwprop): ...new function.
5511         (pass_rtl_fwprop_addr): Replace declaration with that of...
5512         (make_pass_rtl_fwprop_addr): ...new function.
5513         (pass_jump): Replace declaration with that of...
5514         (make_pass_jump): ...new function.
5515         (pass_jump2): Replace declaration with that of...
5516         (make_pass_jump2): ...new function.
5517         (pass_lower_subreg): Replace declaration with that of...
5518         (make_pass_lower_subreg): ...new function.
5519         (pass_cse): Replace declaration with that of...
5520         (make_pass_cse): ...new function.
5521         (pass_fast_rtl_dce): Replace declaration with that of...
5522         (make_pass_fast_rtl_dce): ...new function.
5523         (pass_ud_rtl_dce): Replace declaration with that of...
5524         (make_pass_ud_rtl_dce): ...new function.
5525         (pass_rtl_dce): Replace declaration with that of...
5526         (make_pass_rtl_dce): ...new function.
5527         (pass_rtl_dse1): Replace declaration with that of...
5528         (make_pass_rtl_dse1): ...new function.
5529         (pass_rtl_dse2): Replace declaration with that of...
5530         (make_pass_rtl_dse2): ...new function.
5531         (pass_rtl_dse3): Replace declaration with that of...
5532         (make_pass_rtl_dse3): ...new function.
5533         (pass_rtl_cprop): Replace declaration with that of...
5534         (make_pass_rtl_cprop): ...new function.
5535         (pass_rtl_pre): Replace declaration with that of...
5536         (make_pass_rtl_pre): ...new function.
5537         (pass_rtl_hoist): Replace declaration with that of...
5538         (make_pass_rtl_hoist): ...new function.
5539         (pass_rtl_store_motion): Replace declaration with that of...
5540         (make_pass_rtl_store_motion): ...new function.
5541         (pass_cse_after_global_opts): Replace declaration with that of...
5542         (make_pass_cse_after_global_opts): ...new function.
5543         (pass_rtl_ifcvt): Replace declaration with that of...
5544         (make_pass_rtl_ifcvt): ...new function.
5545         (pass_into_cfg_layout_mode): Replace declaration with that of...
5546         (make_pass_into_cfg_layout_mode): ...new function.
5547         (pass_outof_cfg_layout_mode): Replace declaration with that of...
5548         (make_pass_outof_cfg_layout_mode): ...new function.
5549         (pass_loop2): Replace declaration with that of...
5550         (make_pass_loop2): ...new function.
5551         (pass_rtl_loop_init): Replace declaration with that of...
5552         (make_pass_rtl_loop_init): ...new function.
5553         (pass_rtl_move_loop_invariants): Replace declaration with that of...
5554         (make_pass_rtl_move_loop_invariants): ...new function.
5555         (pass_rtl_unswitch): Replace declaration with that of...
5556         (make_pass_rtl_unswitch): ...new function.
5557         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
5558         (make_pass_rtl_unroll_and_peel_loops): ...new function.
5559         (pass_rtl_doloop): Replace declaration with that of...
5560         (make_pass_rtl_doloop): ...new function.
5561         (pass_rtl_loop_done): Replace declaration with that of...
5562         (make_pass_rtl_loop_done): ...new function.
5563         (pass_web): Replace declaration with that of...
5564         (make_pass_web): ...new function.
5565         (pass_cse2): Replace declaration with that of...
5566         (make_pass_cse2): ...new function.
5567         (pass_df_initialize_opt): Replace declaration with that of...
5568         (make_pass_df_initialize_opt): ...new function.
5569         (pass_df_initialize_no_opt): Replace declaration with that of...
5570         (make_pass_df_initialize_no_opt): ...new function.
5571         (pass_reginfo_init): Replace declaration with that of...
5572         (make_pass_reginfo_init): ...new function.
5573         (pass_inc_dec): Replace declaration with that of...
5574         (make_pass_inc_dec): ...new function.
5575         (pass_stack_ptr_mod): Replace declaration with that of...
5576         (make_pass_stack_ptr_mod): ...new function.
5577         (pass_initialize_regs): Replace declaration with that of...
5578         (make_pass_initialize_regs): ...new function.
5579         (pass_combine): Replace declaration with that of...
5580         (make_pass_combine): ...new function.
5581         (pass_if_after_combine): Replace declaration with that of...
5582         (make_pass_if_after_combine): ...new function.
5583         (pass_ree): Replace declaration with that of...
5584         (make_pass_ree): ...new function.
5585         (pass_partition_blocks): Replace declaration with that of...
5586         (make_pass_partition_blocks): ...new function.
5587         (pass_match_asm_constraints): Replace declaration with that of...
5588         (make_pass_match_asm_constraints): ...new function.
5589         (pass_regmove): Replace declaration with that of...
5590         (make_pass_regmove): ...new function.
5591         (pass_split_all_insns): Replace declaration with that of...
5592         (make_pass_split_all_insns): ...new function.
5593         (pass_fast_rtl_byte_dce): Replace declaration with that of...
5594         (make_pass_fast_rtl_byte_dce): ...new function.
5595         (pass_lower_subreg2): Replace declaration with that of...
5596         (make_pass_lower_subreg2): ...new function.
5597         (pass_mode_switching): Replace declaration with that of...
5598         (make_pass_mode_switching): ...new function.
5599         (pass_sms): Replace declaration with that of...
5600         (make_pass_sms): ...new function.
5601         (pass_sched): Replace declaration with that of...
5602         (make_pass_sched): ...new function.
5603         (pass_ira): Replace declaration with that of...
5604         (make_pass_ira): ...new function.
5605         (pass_reload): Replace declaration with that of...
5606         (make_pass_reload): ...new function.
5607         (pass_clean_state): Replace declaration with that of...
5608         (make_pass_clean_state): ...new function.
5609         (pass_branch_prob): Replace declaration with that of...
5610         (make_pass_branch_prob): ...new function.
5611         (pass_value_profile_transformations): Replace declaration with that
5612         of...
5613         (make_pass_value_profile_transformations): ...new function.
5614         (pass_postreload_cse): Replace declaration with that of...
5615         (make_pass_postreload_cse): ...new function.
5616         (pass_gcse2): Replace declaration with that of...
5617         (make_pass_gcse2): ...new function.
5618         (pass_split_after_reload): Replace declaration with that of...
5619         (make_pass_split_after_reload): ...new function.
5620         (pass_branch_target_load_optimize1): Replace declaration with that
5621         of...
5622         (make_pass_branch_target_load_optimize1): ...new function.
5623         (pass_thread_prologue_and_epilogue): Replace declaration with that
5624         of...
5625         (make_pass_thread_prologue_and_epilogue): ...new function.
5626         (pass_stack_adjustments): Replace declaration with that of...
5627         (make_pass_stack_adjustments): ...new function.
5628         (pass_peephole2): Replace declaration with that of...
5629         (make_pass_peephole2): ...new function.
5630         (pass_if_after_reload): Replace declaration with that of...
5631         (make_pass_if_after_reload): ...new function.
5632         (pass_regrename): Replace declaration with that of...
5633         (make_pass_regrename): ...new function.
5634         (pass_cprop_hardreg): Replace declaration with that of...
5635         (make_pass_cprop_hardreg): ...new function.
5636         (pass_reorder_blocks): Replace declaration with that of...
5637         (make_pass_reorder_blocks): ...new function.
5638         (pass_branch_target_load_optimize2): Replace declaration with that
5639         of...
5640         (make_pass_branch_target_load_optimize2): ...new function.
5641         (pass_leaf_regs): Replace declaration with that of...
5642         (make_pass_leaf_regs): ...new function.
5643         (pass_split_before_sched2): Replace declaration with that of...
5644         (make_pass_split_before_sched2): ...new function.
5645         (pass_compare_elim_after_reload): Replace declaration with that of...
5646         (make_pass_compare_elim_after_reload): ...new function.
5647         (pass_sched2): Replace declaration with that of...
5648         (make_pass_sched2): ...new function.
5649         (pass_stack_regs): Replace declaration with that of...
5650         (make_pass_stack_regs): ...new function.
5651         (pass_stack_regs_run): Replace declaration with that of...
5652         (make_pass_stack_regs_run): ...new function.
5653         (pass_df_finish): Replace declaration with that of...
5654         (make_pass_df_finish): ...new function.
5655         (pass_compute_alignments): Replace declaration with that of...
5656         (make_pass_compute_alignments): ...new function.
5657         (pass_duplicate_computed_gotos): Replace declaration with that of...
5658         (make_pass_duplicate_computed_gotos): ...new function.
5659         (pass_variable_tracking): Replace declaration with that of...
5660         (make_pass_variable_tracking): ...new function.
5661         (pass_free_cfg): Replace declaration with that of...
5662         (make_pass_free_cfg): ...new function.
5663         (pass_machine_reorg): Replace declaration with that of...
5664         (make_pass_machine_reorg): ...new function.
5665         (pass_cleanup_barriers): Replace declaration with that of...
5666         (make_pass_cleanup_barriers): ...new function.
5667         (pass_delay_slots): Replace declaration with that of...
5668         (make_pass_delay_slots): ...new function.
5669         (pass_split_for_shorten_branches): Replace declaration with that of...
5670         (make_pass_split_for_shorten_branches): ...new function.
5671         (pass_split_before_regstack): Replace declaration with that of...
5672         (make_pass_split_before_regstack): ...new function.
5673         (pass_convert_to_eh_region_ranges): Replace declaration with that
5674         of...
5675         (make_pass_convert_to_eh_region_ranges): ...new function.
5676         (pass_shorten_branches): Replace declaration with that of...
5677         (make_pass_shorten_branches): ...new function.
5678         (pass_set_nothrow_function_flags): Replace declaration with that of...
5679         (make_pass_set_nothrow_function_flags): ...new function.
5680         (pass_dwarf2_frame): Replace declaration with that of...
5681         (make_pass_dwarf2_frame): ...new function.
5682         (pass_final): Replace declaration with that of...
5683         (make_pass_final): ...new function.
5684         (pass_rtl_seqabstr): Replace declaration with that of...
5685         (make_pass_rtl_seqabstr): ...new function.
5686         (pass_release_ssa_names): Replace declaration with that of...
5687         (make_pass_release_ssa_names): ...new function.
5688         (pass_early_inline): Replace declaration with that of...
5689         (make_pass_early_inline): ...new function.
5690         (pass_inline_parameters): Replace declaration with that of...
5691         (make_pass_inline_parameters): ...new function.
5692         (pass_update_address_taken): Replace declaration with that of...
5693         (make_pass_update_address_taken): ...new function.
5694         (pass_convert_switch): Replace declaration with that of...
5695         (make_pass_convert_switch): ...new function.
5696         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
5697         to a subclass of simple_ipa_opt_pass along with...
5698         (pass_data_ipa_tree_profile): ...new pass_data instance and...
5699         (make_pass_ipa_tree_profile): ...new function.
5700         * tree-sra.c (pass_sra_early): Convert from a global struct to a
5701         subclass of gimple_opt_pass along with...
5702         (pass_data_sra_early): ...new pass_data instance and...
5703         (make_pass_sra_early): ...new function.
5704         (pass_sra): Convert from a global struct to a subclass of
5705         gimple_opt_pass along with...
5706         (pass_data_sra): ...new pass_data instance and...
5707         (make_pass_sra): ...new function.
5708         (pass_early_ipa_sra): Convert from a global struct to a subclass of
5709         gimple_opt_pass along with...
5710         (pass_data_early_ipa_sra): ...new pass_data instance and...
5711         (make_pass_early_ipa_sra): ...new function.
5712         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
5713         subclass of gimple_opt_pass along with...
5714         (pass_data_ccp): ...new pass_data instance and...
5715         (make_pass_ccp): ...new function.
5716         (pass_fold_builtins): Convert from a global struct to a subclass of
5717         gimple_opt_pass along with...
5718         (pass_data_fold_builtins): ...new pass_data instance and...
5719         (make_pass_fold_builtins): ...new function.
5720         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
5721         subclass of gimple_opt_pass along with...
5722         (pass_data_copy_prop): ...new pass_data instance and...
5723         (make_pass_copy_prop): ...new function.
5724         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
5725         global struct to a subclass of gimple_opt_pass along with...
5726         (pass_data_rename_ssa_copies): ...new pass_data instance and...
5727         (make_pass_rename_ssa_copies): ...new function.
5728         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
5729         subclass of gimple_opt_pass along with...
5730         (pass_data_dce): ...new pass_data instance and...
5731         (make_pass_dce): ...new function.
5732         (pass_dce_loop): Convert from a global struct to a subclass of
5733         gimple_opt_pass along with...
5734         (pass_data_dce_loop): ...new pass_data instance and...
5735         (make_pass_dce_loop): ...new function.
5736         (pass_cd_dce): Convert from a global struct to a subclass of
5737         gimple_opt_pass along with...
5738         (pass_data_cd_dce): ...new pass_data instance and...
5739         (make_pass_cd_dce): ...new function.
5740         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
5741         subclass of gimple_opt_pass along with...
5742         (pass_data_dominator): ...new pass_data instance and...
5743         (make_pass_dominator): ...new function.
5744         (pass_phi_only_cprop): Convert from a global struct to a subclass of
5745         gimple_opt_pass along with...
5746         (pass_data_phi_only_cprop): ...new pass_data instance and...
5747         (make_pass_phi_only_cprop): ...new function.
5748         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
5749         subclass of gimple_opt_pass along with...
5750         (pass_data_dse): ...new pass_data instance and...
5751         (make_pass_dse): ...new function.
5752         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
5753         a subclass of gimple_opt_pass along with...
5754         (pass_data_forwprop): ...new pass_data instance and...
5755         (make_pass_forwprop): ...new function.
5756         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
5757         struct to a subclass of gimple_opt_pass along with...
5758         (pass_data_tree_ifcombine): ...new pass_data instance and...
5759         (make_pass_tree_ifcombine): ...new function.
5760         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
5761         subclass of gimple_opt_pass along with...
5762         (pass_data_ch): ...new pass_data instance and...
5763         (make_pass_ch): ...new function.
5764         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
5765         subclass of gimple_opt_pass along with...
5766         (pass_data_tree_loop): ...new pass_data instance and...
5767         (make_pass_tree_loop): ...new function.
5768         (pass_tree_loop_init): Convert from a global struct to a subclass of
5769         gimple_opt_pass along with...
5770         (pass_data_tree_loop_init): ...new pass_data instance and...
5771         (make_pass_tree_loop_init): ...new function.
5772         (pass_lim): Convert from a global struct to a subclass of
5773         gimple_opt_pass along with...
5774         (pass_data_lim): ...new pass_data instance and...
5775         (make_pass_lim): ...new function.
5776         (pass_tree_unswitch): Convert from a global struct to a subclass of
5777         gimple_opt_pass along with...
5778         (pass_data_tree_unswitch): ...new pass_data instance and...
5779         (make_pass_tree_unswitch): ...new function.
5780         (pass_predcom): Convert from a global struct to a subclass of
5781         gimple_opt_pass along with...
5782         (pass_data_predcom): ...new pass_data instance and...
5783         (make_pass_predcom): ...new function.
5784         (pass_vectorize): Convert from a global struct to a subclass of
5785         gimple_opt_pass along with...
5786         (pass_data_vectorize): ...new pass_data instance and...
5787         (make_pass_vectorize): ...new function.
5788         (pass_graphite): Convert from a global struct to a subclass of
5789         gimple_opt_pass along with...
5790         (pass_data_graphite): ...new pass_data instance and...
5791         (make_pass_graphite): ...new function.
5792         (pass_graphite_transforms): Convert from a global struct to a subclass
5793         of gimple_opt_pass along with...
5794         (pass_data_graphite_transforms): ...new pass_data instance and...
5795         (make_pass_graphite_transforms): ...new function.
5796         (pass_check_data_deps): Convert from a global struct to a subclass of
5797         gimple_opt_pass along with...
5798         (pass_data_check_data_deps): ...new pass_data instance and...
5799         (make_pass_check_data_deps): ...new function.
5800         (pass_iv_canon): Convert from a global struct to a subclass of
5801         gimple_opt_pass along with...
5802         (pass_data_iv_canon): ...new pass_data instance and...
5803         (make_pass_iv_canon): ...new function.
5804         (pass_scev_cprop): Convert from a global struct to a subclass of
5805         gimple_opt_pass along with...
5806         (pass_data_scev_cprop): ...new pass_data instance and...
5807         (make_pass_scev_cprop): ...new function.
5808         (pass_record_bounds): Convert from a global struct to a subclass of
5809         gimple_opt_pass along with...
5810         (pass_data_record_bounds): ...new pass_data instance and...
5811         (make_pass_record_bounds): ...new function.
5812         (pass_complete_unroll): Convert from a global struct to a subclass of
5813         gimple_opt_pass along with...
5814         (pass_data_complete_unroll): ...new pass_data instance and...
5815         (make_pass_complete_unroll): ...new function.
5816         (pass_complete_unrolli): Convert from a global struct to a subclass of
5817         gimple_opt_pass along with...
5818         (pass_data_complete_unrolli): ...new pass_data instance and...
5819         (make_pass_complete_unrolli): ...new function.
5820         (pass_parallelize_loops): Convert from a global struct to a subclass
5821         of gimple_opt_pass along with...
5822         (pass_data_parallelize_loops): ...new pass_data instance and...
5823         (make_pass_parallelize_loops): ...new function.
5824         (pass_loop_prefetch): Convert from a global struct to a subclass of
5825         gimple_opt_pass along with...
5826         (pass_data_loop_prefetch): ...new pass_data instance and...
5827         (make_pass_loop_prefetch): ...new function.
5828         (pass_iv_optimize): Convert from a global struct to a subclass of
5829         gimple_opt_pass along with...
5830         (pass_data_iv_optimize): ...new pass_data instance and...
5831         (make_pass_iv_optimize): ...new function.
5832         (pass_tree_loop_done): Convert from a global struct to a subclass of
5833         gimple_opt_pass along with...
5834         (pass_data_tree_loop_done): ...new pass_data instance and...
5835         (make_pass_tree_loop_done): ...new function.
5836         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
5837         struct to a subclass of gimple_opt_pass along with...
5838         (pass_data_cse_reciprocals): ...new pass_data instance and...
5839         (make_pass_cse_reciprocals): ...new function.
5840         (pass_cse_sincos): Convert from a global struct to a subclass of
5841         gimple_opt_pass along with...
5842         (pass_data_cse_sincos): ...new pass_data instance and...
5843         (make_pass_cse_sincos): ...new function.
5844         (pass_optimize_bswap): Convert from a global struct to a subclass of
5845         gimple_opt_pass along with...
5846         (pass_data_optimize_bswap): ...new pass_data instance and...
5847         (make_pass_optimize_bswap): ...new function.
5848         (pass_optimize_widening_mul): Convert from a global struct to a
5849         subclass of gimple_opt_pass along with...
5850         (pass_data_optimize_widening_mul): ...new pass_data instance and...
5851         (make_pass_optimize_widening_mul): ...new function.
5852         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
5853         subclass of gimple_opt_pass along with...
5854         (pass_data_phiopt): ...new pass_data instance and...
5855         (make_pass_phiopt): ...new function.
5856         (pass_cselim): Convert from a global struct to a subclass of
5857         gimple_opt_pass along with...
5858         (pass_data_cselim): ...new pass_data instance and...
5859         (make_pass_cselim): ...new function.
5860         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
5861         subclass of gimple_opt_pass along with...
5862         (pass_data_phiprop): ...new pass_data instance and...
5863         (make_pass_phiprop): ...new function.
5864         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
5865         subclass of gimple_opt_pass along with...
5866         (pass_data_pre): ...new pass_data instance and...
5867         (make_pass_pre): ...new function.
5868         (pass_fre): Convert from a global struct to a subclass of
5869         gimple_opt_pass along with...
5870         (pass_data_fre): ...new pass_data instance and...
5871         (make_pass_fre): ...new function.
5872         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
5873         subclass of gimple_opt_pass along with...
5874         (pass_data_reassoc): ...new pass_data instance and...
5875         (make_pass_reassoc): ...new function.
5876         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
5877         subclass of gimple_opt_pass along with...
5878         (pass_data_sink_code): ...new pass_data instance and...
5879         (make_pass_sink_code): ...new function.
5880         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
5881         subclass of gimple_opt_pass along with...
5882         (pass_data_strlen): ...new pass_data instance and...
5883         (make_pass_strlen): ...new function.
5884         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
5885         struct to a subclass of gimple_opt_pass along with...
5886         (pass_data_build_alias): ...new pass_data instance and...
5887         (make_pass_build_alias): ...new function.
5888         (pass_build_ealias): Convert from a global struct to a subclass of
5889         gimple_opt_pass along with...
5890         (pass_data_build_ealias): ...new pass_data instance and...
5891         (make_pass_build_ealias): ...new function.
5892         (pass_ipa_pta): Convert from a global struct to a subclass of
5893         simple_ipa_opt_pass along with...
5894         (pass_data_ipa_pta): ...new pass_data instance and...
5895         (make_pass_ipa_pta): ...new function.
5896         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
5897         subclass of gimple_opt_pass along with...
5898         (pass_data_uncprop): ...new pass_data instance and...
5899         (make_pass_uncprop): ...new function.
5900         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
5901         global struct to a subclass of gimple_opt_pass along with...
5902         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
5903         (make_pass_late_warn_uninitialized): ...new function.
5904         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
5905         to a subclass of gimple_opt_pass along with...
5906         (pass_data_init_datastructures): ...new pass_data instance and...
5907         (make_pass_init_datastructures): ...new function.
5908         (pass_early_warn_uninitialized): Convert from a global struct to a
5909         subclass of gimple_opt_pass along with...
5910         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
5911         (make_pass_early_warn_uninitialized): ...new function.
5912         (pass_update_address_taken): Convert from a global struct to a
5913         subclass of gimple_opt_pass along with...
5914         (pass_data_update_address_taken): ...new pass_data instance and...
5915         (make_pass_update_address_taken): ...new function.
5916         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
5917         struct to a subclass of gimple_opt_pass along with...
5918         (pass_data_release_ssa_names): ...new pass_data instance and...
5919         (make_pass_release_ssa_names): ...new function.
5920         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
5921         subclass of gimple_opt_pass along with...
5922         (pass_data_stdarg): ...new pass_data instance and...
5923         (make_pass_stdarg): ...new function.
5924         * tree-switch-conversion.c (pass_convert_switch): Convert from a
5925         global struct to a subclass of gimple_opt_pass along with...
5926         (pass_data_convert_switch): ...new pass_data instance and...
5927         (make_pass_convert_switch): ...new function.
5928         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
5929         to a subclass of gimple_opt_pass along with...
5930         (pass_data_tail_recursion): ...new pass_data instance and...
5931         (make_pass_tail_recursion): ...new function.
5932         (pass_tail_calls): Convert from a global struct to a subclass of
5933         gimple_opt_pass along with...
5934         (pass_data_tail_calls): ...new pass_data instance and...
5935         (make_pass_tail_calls): ...new function.
5936         * tree-vect-generic.c (pass_lower_vector): Convert from a global
5937         struct to a subclass of gimple_opt_pass along with...
5938         (pass_data_lower_vector): ...new pass_data instance and...
5939         (make_pass_lower_vector): ...new function.
5940         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
5941         gimple_opt_pass along with...
5942         (pass_data_lower_vector_ssa): ...new pass_data instance and...
5943         (make_pass_lower_vector_ssa): ...new function.
5944         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
5945         to a subclass of gimple_opt_pass along with...
5946         (pass_data_slp_vectorize): ...new pass_data instance and...
5947         (make_pass_slp_vectorize): ...new function.
5948         (pass_ipa_increase_alignment): Convert from a global struct to a
5949         subclass of simple_ipa_opt_pass along with...
5950         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
5951         (make_pass_ipa_increase_alignment): ...new function.
5952         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
5953         gimple_opt_pass along with...
5954         (pass_data_vrp): ...new pass_data instance and...
5955         (make_pass_vrp): ...new function.
5956         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
5957         subclass of simple_ipa_opt_pass along with...
5958         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
5959         (make_pass_ipa_free_lang_data): ...new function.
5960         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
5961         gimple_opt_pass along with...
5962         (pass_data_tsan): ...new pass_data instance and...
5963         (make_pass_tsan): ...new function.
5964         (pass_tsan_O0): Convert from a global struct to a subclass of
5965         gimple_opt_pass along with...
5966         (pass_data_tsan_O0): ...new pass_data instance and...
5967         (make_pass_tsan_O0): ...new function.
5968         * var-tracking.c (pass_variable_tracking): Convert from a global
5969         struct to a subclass of rtl_opt_pass along with...
5970         (pass_data_variable_tracking): ...new pass_data instance and...
5971         (make_pass_variable_tracking): ...new function.
5972         * web.c (pass_web): Convert from a global struct to a subclass of
5973         rtl_opt_pass along with...
5974         (pass_data_web): ...new pass_data instance and...
5975         (make_pass_web): ...new function.
5976         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
5977         declaration with that of...
5978         (make_pass_mode_switch_use): ...new function.
5979         (pass_resolve_sw_modes): Replace declaration with that of...
5980         (make_pass_resolve_sw_modes): ...new function.
5981         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
5982         from a global struct to a subclass of rtl_opt_pass along with...
5983         (pass_data_mode_switch_use): ...new pass_data instance and...
5984         (make_pass_mode_switch_use): ...new function.
5985         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
5986         from a global struct to a subclass of rtl_opt_pass along with...
5987         (pass_data_resolve_sw_modes): ...new pass_data instance and...
5988         (make_pass_resolve_sw_modes): ...new function.
5989         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
5990         struct to a subclass of rtl_opt_pass along with...
5991         (pass_data_insert_vzeroupper): ...new pass_data instance and...
5992         (make_pass_insert_vzeroupper): ...new function.
5993         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
5994         global struct to a subclass of rtl_opt_pass along with...
5995         (pass_data_work_around_errata): ...new pass_data instance and...
5996         (make_pass_work_around_errata): ...new function.
5997         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
5998         struct to a subclass of rtl_opt_pass along with...
5999         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
6000         (make_pass_mips_machine_reorg2): ...new function.
6002 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
6004         * passes.c (pass_manager::operator new): New.
6006 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
6008         Handwritten part of conversion of passes to C++ classes.
6010         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
6011         (toplev.o): Add dep on PASS_MANAGER_H.
6012         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
6013         of early local pases to reflect this moving from a global to a
6014         member of gcc::pass_manager.
6015         (cgraph_add_new_function): Likewise.
6016         * lto-cgraph.c (lto_output_node): Update for conversion of
6017         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
6018         * passes.c (opt_pass::clone): New.
6019         (opt_pass::gate): New.
6020         (opt_pass::execute): New.
6021         (opt_pass::opt_pass): New.
6022         (pass_manager::execute_early_local_passes): New.
6023         (pass_manager::execute_pass_mode_switching): new.
6024         (finish_optimization_passes): Convert to...
6025         (pass_manager::finish_optimization_passes): ...this.
6026         (finish_optimization_passes): Update for conversion of passes to
6027         C++ classes.
6028         (register_dump_files_1): Use has_gate since we cannot portably
6029         check a vtable entry against NULL.
6030         (dump_one_pass): Likewise.
6031         (ipa_write_summaries_2): Likewise.
6032         (ipa_write_optimization_summaries_1): Likewise.
6033         (ipa_read_summaries_1): Likewise.
6034         (ipa_read_optimization_summaries_1): Likewise.
6035         (execute_ipa_stmt_fixups): Likewise.
6036         (pass_manager::pass_manager): Rewrite pass-creation, invoking
6037         pass-creation functions rather than wiring up globals, and
6038         storing the results in fields of pass_manager generated using
6039         pass-instances.def.
6040         (pass_manager::dump_profile_report): Update for conversion of
6041         passes to C++ classes.
6042         (pass_manager::execute_ipa_summary_passes): Likewise.
6043         (execute_one_ipa_transform_pass): Likewise.
6044         (execute_one_pass): Use has_gate and has_execute since we cannot
6045         portably check a vtable entry against NULL.
6046         * pass_manager.h (pass_manager::finish_optimization_passes): New.
6047         (pass_manager): Use pass-instances.def to add fields for the
6048         various pass instances.
6049         * toplev.c (finalize): Update for move of
6050         finish_optimization_passes to a method of gcc::pass_manager.
6051         * toplev.h (finish_optimization_passes): Move to method of class
6052         pass_manager.
6053         * tree-pass.h (struct pass_data): New.
6054         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
6055         (opt_pass::gate): Convert to virtual function.
6056         (opt_pass::~opt_pass): New.
6057         (opt_pass::clone): New.
6058         (opt_pass::execute): Convert to virtual function.
6059         (opt_pass::opt_pass): New.
6060         (opt_pass::ctxt_): new.
6061         (gimple_opt_pass): Convert to subclass of opt_pass.
6062         (gimple_opt_pass::gimple_opt_pass): New.
6063         (rtl_opt_pass): Convert to subclass of opt_pass.
6064         (rtl_opt_pass::rtl_opt_pass): New.
6065         (ipa_opt_pass_d): Convert to subclass of opt_pass.
6066         (ipa_opt_pass_d::ipa_opt_pass_d): New.
6067         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
6068         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
6069         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
6070         invocation of pass_mode_switching to reflect this moving from a
6071         global to a member of gcc::pass_manager.
6072         (ix86_option_override): Rework how pass_insert_vzeroupper is
6073         added to the pass_manager to reflect autogenerated changes.
6074         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
6076 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
6078         PR rtl-optimization/57708
6079         * recog.c (peep2_find_free_register): Validate all regs in a
6080         multi-reg mode.
6082 2013-08-05  Jan Hubicka  <jh@suse.cz>
6084         PR lto/57602
6085         * cgraph.c (verify_cgraph_node): Accept local flags from other
6086         partitions.
6087         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
6088         (function_and_variable_visibility): Likewise.
6089         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
6091 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
6093         * graph.c (init_graph_slim_pretty_print): Remove.
6094         (print_graph_cfg): Do not call it.  Use local pretty printer.
6095         (start_graph_dump): Likewise.
6097 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
6099         * gimple-pretty-print.c (buffer): Remove.
6100         (initialized): Likewise.
6101         (maybe_init_pretty_print): Likewise.
6102         (print_gimple_stmt): Do not call it.  Use non-static local
6103         pretty_printer variable.
6104         (print_gimple_expr): Likewise.
6105         (print_gimple_seq): Likewise.
6106         (gimple_dump_bb): Likewise.
6108 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
6110         * asan.c (asan_pp): Remove.
6111         (asan_pp_initialized): Likewise.
6112         (asan_pp_initialize): Likewise.
6113         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
6114         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
6115         (asan_add_global): Likewise.
6117 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
6119         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
6120         * diagnostic.h (diagnostic_flush_buffer): Adjust.
6121         * pretty-print.c (pp_formatted_text_data): Likewise.
6122         (pp_indent): Rename from pp_base_indent.
6123         (pp_format): Rename from pp_base_format.
6124         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
6125         (pp_format_verbatim): Rename from pp_base_format_verbatim.
6126         (pp_flush): Rename from pp_base_flush.
6127         (pp_set_line_maximum_length): Rename from
6128         pp_base_set_line_maximum_length.
6129         (pp_clear_output_area): Rename from pp_base_clear_output_area.
6130         (pp_set_prefix): Rename from pp_base_set_prefix.
6131         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
6132         (pp_emit_prefix): Rename from pp_base_emit_prefix.
6133         (pp_append_text): Rename from pp_base_append_text.
6134         (pp_formatted_text): Rename from pp_base_formatted_text.
6135         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
6136         (pp_remaining_character_count_for_line): Rename from
6137         pp_base_remaining_character_count_for_line.
6138         (pp_newline): Rename from pp_base_newline.
6139         (pp_character): Rename from pp_base_character.
6140         (pp_string): Rename from pp_base_string.
6141         (pp_maybe_space): Rename from pp_base_maybe_space.
6142         * asan.c (asan_pp_string): Adjust.
6143         (asan_emit_stack_protection): Likewise.
6144         (asan_add_global): Likewise.
6145         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
6146         * tree-mudflap.c (mf_varname_tree): Likewise.
6147         * tree-pretty-print.c (pp_tree_identifier): Rename from
6148         pp_base_tree_identifier.
6149         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
6150         Declare as function.
6152 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
6154         * pretty-print.h (pp_bar_bar): New.
6155         (pp_ampersand_ampersand): Likewise.
6156         (pp_less_equal): Likewise.
6157         (pp_greater_equal): Likewise.
6158         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
6159         printer functions instead of pp_string or operators and punctuators.
6160         (dump_gimple_call): Likewise.
6161         (dump_gimple_omp_for): Likewise.
6162         (dump_gimple_transaction): Likewise.
6163         (dump_gimple_phi): Likewise.
6164         (pp_gimple_stmt_1): Likewise.
6165         * sched-vis.c (print_insn): Likewise.
6166         * tree-mudflap.c (mf_varname_tree): Likewise.
6167         * tree-pretty-print.c (dump_block_node): Likewise.
6168         (dump_generic_node): Likewise.
6170 2013-08-02  Jan Hubicka  <jh@suse.cz>
6172         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
6173         boundaries.
6174         * lto-streamer-out.c (tree_is_indexable): Results decls and
6175         parm decls are not indexable.
6176         (DFS_write_tree_body): Do not follow args and results.
6177         (hash_tree): Likewise.
6178         (output_functions): Rearrange so struct function is needed
6179         only when real body is output; be able to also ouptut abstract
6180         functions; output DECL_ARGUMENTS and DECL_RESULT.
6181         (lto_output): When not in WPA, ale store abstract functions.
6182         (write_symbol): Do not care about RESULT_DECL.
6183         (output_symbol_p): Handle correctly sbtract decls.
6184         * lto-streamer-in.c (input_function): Rearrange so struct
6185         function can be NULL at entry; allow streaming of
6186         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
6187         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
6188         sanity check during LTO.
6189         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
6190         RESULT_DECl and DECL_ARGUMENTS.
6191         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
6192         Likewise.
6194 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
6196         * pretty-print.h (pp_underscore): New.
6197         (pp_comma): Tidy.
6198         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
6199         printer functions instead of pp_character.
6200         (dump_binary_rhs): Likewise.
6201         (dump_ternary_rhs): Likewise.
6202         (dump_gimple_call_args): Likewise.
6203         (pp_points_to_solution): Likewise.
6204         (dump_gimple_call): Likewise.
6205         (dump_gimple_switch): Likewise.
6206         (dump_gimple_cond): Likewise.
6207         (dump_gimple_bind): Likewise.
6208         (dump_gimple_try): Likewise.
6209         (dump_gimple_omp_for): Likewise.
6210         (dump_gimple_omp_continue): Likewise.
6211         (dump_gimple_omp_single): Likewise.
6212         (dump_gimple_omp_sections): Likewise.
6213         (dump_gimple_omp_block): Likewise.
6214         (dump_gimple_omp_critical): Likewise.
6215         (dump_gimple_transaction): Likewise.
6216         (dump_gimple_asm): Likewise.
6217         (dump_gimple_phi): Likewise.
6218         (dump_gimple_omp_parallel): Likewise.
6219         (dump_gimple_omp_task): Likewise.
6220         (dump_gimple_omp_atomic_load): Likewise.
6221         (dump_gimple_omp_atomic_store): Likewise.
6222         (dump_gimple_mem_ops): Likewise.
6223         (pp_gimple_stmt_1): Likewise.
6224         (pp_cfg_jump): Likewise.
6225         (dump_implicit_edges): Likewise.
6226         (gimple_dump_bb_for_graph): Likewise.
6227         * graph.c (draw_cfg_node): Likewise.
6228         * langhooks.c (lhd_print_error_function): Likewise.
6229         * sched-vis.c (print_exp): Likewise.
6230         (print_value): Likewise.
6231         (print_pattern): Likewise.
6232         (print_insn): Likewise.
6233         (rtl_dump_bb_for_graph): Likewise.
6234         * tree-pretty-print.c (dump_function_declaration): Likewise.
6235         (dump_array_domain): Likewise.
6236         (dump_omp_clause): Likewise.
6237         (dump_location): Likewise.
6238         (dump_generic_node): Likewise.
6239         (print_struct_decl): Likewise.
6240         * diagnostic.c (diagnostic_show_locus): Use pp_space.
6242 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
6244         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
6245         candidate table when replacing a candidate statement.
6246         (replace_rhs_if_not_dup): Likewise.
6247         (replace_one_candidate): Likewise.
6249 2013-08-02  Jan Hubicka  <jh@suse.cz>
6250             Martin Liska  <marxin.liska@gmail.com>
6252         * cgraphunit.c (add_new_function): Fix logic when adding from
6253         late IPA pass.
6254         (assemble_thunk): Rename to ...
6255         (expand_thunk); .. this one; export; get it working with
6256         general functions; make produced gimple valid.
6257         * cgraph.h (expand_thunk): Declare.
6259 2013-08-02  Jan Hubicka  <jh@suse.cz>
6261         * ipa-cp.c (gather_context_independent_values): Use
6262         ipa_get_param_move_cost.
6263         (get_replacement_map): Remove PARAM; move parameter folding
6264         into tree-inline.c
6265         (create_specialized_node): Update.
6266         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
6267         assert that we have gimple body; update move_cost.
6268         (count_formal_params): Assert that we have gimple body.
6269         (ipa_dump_param): New function.
6270         (ipa_alloc_node_params): Break out from ...
6271         (ipa_initialize_node_params): ... here.
6272         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
6273         (ipa_write_node_info): Stream move costs.
6274         (ipa_read_node_info): Read move costs.
6275         (ipa_update_after_lto_read): Do not recompute node params.
6276         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
6277         (ipa_get_param): Check we are not in WPA.
6278         (ipa_get_param_move_cost): New.
6279         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
6280         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
6281         parm numbers to be present.
6283 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
6285         PR rtl-optimization/58048
6286         * lra-constraints.c (process_alt_operands): Don't check asm
6287         operand on register.
6289 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
6291         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
6292         the implied StoreLoad barrier for atomic operations if before.
6294 2013-08-02  Jan Hubicka  <jh@suse.cz>
6295             Martin Liska  <marxin.liska@gmail.com>
6297         * cgraph.c (cgraph_function_body_availability): Do not check
6298         cgraph flags.
6299         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
6300         symtab_node_availability): Declare.
6301         * ipa.c (can_replace_by_local_alias): New.
6302         (function_and_variable_visibility): Use it.
6303         * symtab.c (symtab_for_node_and_aliases,
6304         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
6306 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
6308         PR rtl-optimization/57963
6309         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
6310         (lra_constraints): Use them.
6312 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
6314         * config/arm/types.md (define_attr "type"): Add "load_acq"
6315         and "store_rel".
6316         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
6317         changes.
6318         (cortex_a53_store1): Likewise.
6320 2013-08-01  Jan Hubicka  <jh@suse.cz>
6322         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
6323         partitions are not needed.
6325 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
6327         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
6328         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
6329         MAYBE_NON_Q_CLASS_P where appropriate.
6331 2013-08-01  Jan Hubicka  <jh@suse.cz>
6333         * cgraph.h (release_function_body): Declare.
6334         * tree.c (free_lang_data_in_decl): Free, parameters and return values
6335         of unused delcarations.
6337 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6339         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
6340         RTL form when subtracting a constant.
6342 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6344         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
6345         Generate canonical plus rtx with negated immediate instead of minus
6346         where appropriate.
6347         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
6349 2013-08-01  Jan Hubicka  <jh@suse.cz>
6351         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
6352         (cgraph_release_function_body): Likewise.
6353         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
6354         * cgraph.h (cgrpah_node): Rename abstract_and_needed
6355         to used_as_abstract_origin.
6356         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
6357         symbols used as abstract origins.
6358         * cgraphunit.c (analyze_functions): Update.
6359         * ipa.c (symtab_remove_unreachable_nodes): Recompute
6360         used_as_abstract_origin.
6361         * tree-inline.c (tree_function_versioning): Update
6362         used_as_abstract_origin; be ready for DECL_RESULT and
6363         DECL_ARGUMENTS to be NULL.
6365         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
6366         for abstract functions.
6367         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
6368         real symbols.
6370 2013-08-01  Jan Hubicka  <jh@suse.cz>
6372         * profile.c (compute_value_histograms): Fix thinko.
6374 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
6376         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
6377         aarch-common-protos.h to extra_headers.
6378         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
6379         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
6380         * config/aarch64/t-aarch64 (aarch-common.o): Define.
6382 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
6384         * config/aarch64/aarch64.md (define_attr "type"): Delete.
6385         Include "../arm/types.md".  Define "type" attribute for all patterns.
6386         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
6387         attribute changes.
6389 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
6391         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
6392         to support power8 load fusion.
6393         (fusion_gpr_mem_load): Likewise.
6395         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
6397         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
6398         declarations for power8 load fusion.
6399         (emit_fusion_gpr_load): Likewise.
6401         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
6402         tuning for power8, turn on fusion mode by default.  Turn on sign
6403         extending fusion mode if normal fusion mode is on, and we are at
6404         -O2 or -O3.
6405         (fusion_gpr_load_p): New function, return true if we can fuse an
6406         addis instruction with a dependent load to a GPR.
6407         (emit_fusion_gpr_load): Emit the instructions for power8 load
6408         fusion to GPRs.
6410         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
6411         (VSX load fusion peepholes): New peepholes to fuse together an
6412         addi instruction with a VSX load instruction.
6414         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
6415         peepholes to fuse an addis instruction with a load to a GPR base
6416         register.  If we are supporting sign extending fusions, convert
6417         sign extending loads to zero extending loads and add an explicit
6418         sign extension.
6420 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
6422         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
6423         aarch-common-protos.h to extra_headers.
6424         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
6425         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
6426         (arm_early_store_addr_dep): Likewise.
6427         (arm_no_early_alu_shift_dep): Likewise.
6428         (arm_no_early_alu_shift_value_dep): Likewise.
6429         (arm_no_early_mul_dep): Likewise.
6430         (arm_no_early_store_addr_dep): Likewise.
6431         (arm_mac_accumulator_is_mul_result): Likewise.
6432         (arm_mac_accumulator_is_result): Likewise.
6433         * config/arm/aarch-common.c: ... here.  New file.
6434         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
6435         here to ...
6436         (arm_early_store_addr_dep): Likewise.
6437         (arm_no_early_alu_shift_dep): Likewise.
6438         (arm_no_early_alu_shift_value_dep): Likewise.
6439         (arm_no_early_mul_dep): Likewise.
6440         (arm_no_early_store_addr_dep): Likewise.
6441         (arm_mac_accumulator_is_mul_result): Likewise.
6442         (arm_mac_accumulator_is_result): Likewise.
6443         * config/arm/aarch-common-protos.h: ... here.  New file.
6444         * config/arm/t-arm (aarch-common.o): Define.
6446 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
6448         * config/arm/arm.md: Include new file "types.md".
6449         (define_attr "type"): Move from here to ...
6450         (define_attr "mul32"): Likewise.
6451         (define_attr "mul64"): Likewise.
6452         * config/arm/types.md: ... here.  New file.
6454 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
6456         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
6457         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
6459 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
6461         * gen-pass-instances.awk: Fix offset of substr().
6463 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
6465         * Makefile.in (pass-instances.def): New.
6466         (passes.o): Replace dependency on passes.def with one on
6467         pass-instances.def
6469         * gen-pass-instances.awk: New.
6471         * passes.c (pass_manager::pass_manager): Use pass-instances.def
6472         rather than passes.def, updating local definition of NEXT_PASS
6473         macro to add an extra NUM parameter (currently unused).
6475 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
6477         * Makefile.in (PASS_MANAGER_H): New.
6478         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
6479         (passes.o): Likewise.
6480         (statistics.o): Likewise.
6481         (cgraphunit.o): Likewise.
6482         (context.o): Depend on PASS_MANAGER_H.
6484         * pass_manager.h: New.
6486         * cgraphunit.c (cgraph_add_new_function): Update for moves
6487         of globals to fields of pass_manager.
6488         (analyze_function): Likewise.
6489         (expand_function): Likewise.
6490         (ipa_passes): Likewise.
6491         (compile): Likewise.
6493         * context.c (context::context): New.
6494         * context.h  (context::context): New.
6495         (context::get_passes): New.
6496         (context::passes_): New.
6498         * lto-cgraph.c (input_node): Update for moves of globals to
6499         fields of pass_manager.
6501         * passes.c (all_passes): Remove, in favor of a field of the
6502         same name within the new class pass_manager.
6503         (all_small_ipa_passes): Likewise.
6504         (all_lowering_passes): Likewise.
6505         (all_regular_ipa_passes): Likewise.
6506         (all_late_ipa_passes): Likewise.
6507         (all_lto_gen_passes): Likewise.
6508         (passes_by_id): Likewise.
6509         (passes_by_id_size): Likewise.
6510         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
6511         the new class pass_manager.
6512         (set_pass_for_id): Convert to...
6513         (pass_manager::set_pass_for_id): ...method.
6514         (get_pass_for_id): Convert to...
6515         (pass_manager::get_pass_for_id): ...method.
6516         (register_one_dump_file): Move body of implementation into...
6517         (pass_manager::register_one_dump_file): ...here.
6518         (register_dump_files_1): Convert to...
6519         (pass_manager::register_dump_files_1): ...method.
6520         (register_dump_files): Convert to...
6521         (pass_manager::register_dump_files): ...method.
6522         (create_pass_tab): Update for moves of globals to fields of
6523         pass_manager.
6524         (dump_passes): Move body of implementation into...
6525         (pass_manager::dump_passes): ...here.
6526         (register_pass): Move body of implementation into...
6527         (pass_manager::register_pass): ...here.
6528         (init_optimization_passes): Convert into...
6529         (pass_manager::pass_manager): ...constructor for new
6530         pass_manager class, and initialize the pass_lists array.
6531         (check_profile_consistency): Update for moves of globals to
6532         fields of pass_manager.
6533         (dump_profile_report): Move body of implementation into...
6534         (pass_manager::dump_profile_report): ...here.
6535         (ipa_write_summaries_1): Update for moves of pass lists from
6536         being globals to fields of pass_manager.
6537         (ipa_write_optimization_summaries): Likewise.
6538         (ipa_read_summaries):  Likewise.
6539         (ipa_read_optimization_summaries): Likewise.
6540         (execute_all_ipa_stmt_fixups): Likewise.
6542         * statistics.c (statistics_fini): Update for moves of globals to
6543         fields of pass_manager.
6545         * toplev.c (general_init): Replace call to
6546         init_optimization_passes with construction of the pass_manager
6547         instance.
6549         * tree-pass.h (all_passes): Remove, in favor of a field of the
6550         same name within the new class pass_manager.
6551         (all_small_ipa_passes): Likewise.
6552         (all_lowering_passes): Likewise.
6553         (all_regular_ipa_passes): Likewise.
6554         (all_lto_gen_passes): Likewise.
6555         (all_late_ipa_passes): Likewise.
6556         (passes_by_id): Likewise.
6557         (passes_by_id_size): Likewise.
6558         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
6559         the new class pass_manager.
6560         (get_pass_for_id): Remove.
6562 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
6564         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
6565         configs.
6567 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
6569         * arm.md (mulhi3): New expand pattern.
6571 2013-07-30  Jan Hubicka  <jh@suse.cz>
6572             Martin Liska  <marxin.liska@gmail.com>
6574         * profile.c (compute_value_histograms): Do not ICE when
6575         there is mismatch only on some counters.
6577 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
6579         PR rtl-optimization/57637
6580         * function.c (move_insn_for_shrink_wrap): Also check the
6581         GEN set of the LIVE problem for the liveness analysis
6582         if it exists, otherwise give up.
6584 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
6586         PR tree-optimization/57993
6587         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
6588         replaced statement in the candidate table.
6589         (phi_add_costs): Return infinite cost when the hidden basis does
6590         not dominate all phis on which the candidate is dependent.
6591         (replace_one_candidate): Record replaced statement in the
6592         candidate table.
6594 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
6596         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
6597         (ashlv2si3): New expander.
6598         (*ashlv2si3_i): New define_insn_and_split.
6599         * predicates.md (float_operation): Allow patterns with three
6600         basic sub-patterns.
6602         PR rtl-optimization/58021
6603         * mode-switching.c (create_pre_exit): Always split off preceding
6604         insns if we are not at the basic block head.
6606 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
6608         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
6609         (UCLIBC_DYNAMIC_LINKER): New macro.
6610         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
6611         `-mnan=2008'.
6612         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
6613         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
6614         `-mnan=2008'.
6615         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
6616         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
6617         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
6618         for SF and DF modes.  Use ieee_quad_format for TF mode.
6619         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
6620         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
6621         (mips_option_override): Handle `-mnan=legacy'.
6622         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
6623         `-mabs=2008' and `-mnan=2008'.
6624         (OPTION_DEFAULT_SPECS): Add "nan" default.
6625         (ASM_SPEC): Handle `-mnan='.
6626         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
6627         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
6628         comment accordingly.
6629         (neg<mode>2): Likewise.
6630         * config/mips/mips.opt (mabs, mnan): New options.
6631         * doc/install.texi (Configuration): Document `--with-nan=' option.
6632         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
6633         `-mnan=' options.
6634         (MIPS Options): Document them.
6635         * config.gcc <mips*-*-*>: Handle `--with-nan='.
6636         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
6637         * configure: Regenerate.
6638         * config.in: Regenerate.
6640 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
6642         * config/i386/i386.md (float post-reload splitters): Do not check
6643         for subregs of SSE registers.
6645 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
6646             H.J. Lu  <hongjiu.lu@intel.com>
6648         PR target/57954
6649         PR target/57988
6650         * config/i386/i386.md (post-reload splitter
6651         to avoid partial SSE reg dependency stalls): New pattern.
6653 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
6655         * config/s390/s390.md ("movcc"): Swap load and store instructions.
6657 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
6659         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
6660         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
6662 2013-07-26  Cary Coutant  <ccoutant@google.com>
6664         * dwarf2out.c (die_checksum_ordered): Don't include template
6665         instantiations in signature.
6666         (is_template_parameter): New function.
6667         (is_template_instantiation): New function.
6668         (generate_skeleton_bottom_up): Don't include template instantiations
6669         in type unit DIE.
6670         (generate_skeleton): Likewise.
6671         (break_out_comdat_types): Move recursive call to break out nested
6672         types earlier.
6673         (prune_unused_types_mark_generic_parms_dies): Call
6674         is_template_parameter.
6676 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
6678         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
6679         uses vector registers.
6680         * config/aarch64/iterators.md: Add attributes rtn and vas.
6682 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6683             Richard Earnshaw  <richard.earnshaw@arm.com>
6685         * combine.c (simplify_comparison): Re-canonicalize operands
6686         where appropriate.
6687         * config/arm/arm.md (movcond_addsi): New splitter.
6689 2013-07-25  Sterling Augustine  <saugustine@google.com>
6691         * dwarf2out.c (size_of_pubnames): Move code to...
6692         (include_pubname_in_output): ...here.  New.
6693         (want_pubnames): Rearrange.
6694         (output_pubnames): Call include_pubname_in_output.  Move assertion.
6696 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
6698         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
6700 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
6702         PR target/38836
6703         * doc/extend.texi: Remove obsolete builtins. Fix
6704         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
6706 2013-07-25  Jan Hubicka  <jh@suse.cz>
6708         * cgraph.c (release_function_body): Break out from ...
6709         (cgraph_release_function_body): ... this one; also release DECL_RESULT
6710         and DECL_ARGUMENTS.
6711         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
6712         old_tree in the map.
6713         (create_specialized_node): Update.
6714         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
6715         into index.
6716         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
6717         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
6718         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
6719         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
6720         DECL_RESULT.
6722 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6724         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
6725         addsi3_carryin_alt2_<optab>): Correct output template.
6727 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6729         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
6730         Adjust for arm_restrict_it.
6731         Remove trailing whitespace.
6733 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
6735         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
6736         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
6738         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
6740 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
6742         PR rtl-optimization/57960
6743         * lra-constraints.c (process_alt_operands): Use the right mode
6744         when checking strict_low.
6746 2013-07-25  Jan Hubicka  <jh@suse.cz>
6748         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
6749         * cgraph.c (cgraph_remove_node): Do not release function body
6750         when in cgraph streaming.
6751         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
6752         in other partitions are not considered reachable; fix handling of
6753         clones.
6755 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
6757         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
6759 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
6761         PR target/19599
6762         PR target/57731
6763         PR target/57837
6764         * config/arm/arm.md ("*sibcall_insn): Replace use of
6765         Ss with US. Adjust output for v5 and v4t.
6766         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
6768         * config/arm/constraints.md ("Ss"): Rename to US.
6770 2013-07-25  Terry Guo  <terry.guo@arm.com>
6772         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
6773         shift_add/shift_sub0/shift_sub1 RTXs.
6775 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
6776             Anton Blanchard  <anton@au1.ibm.com>
6778         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
6779         (altivec_vpks<VI_char>ss): Likewise.
6780         (altivec_vpks<VI_char>us): Likewise.
6781         (altivec_vpku<VI_char>us): Likewise.
6782         (altivec_vpku<VI_char>um): Likewise.
6784 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
6786         Introduce context class.
6788         * Makefile.in (CONTEXT_H): New.
6789         (OBJS): Add context.o.
6790         (toplev.o): Add CONTEXT_H to dependencies.
6791         (context.o): New.
6793         * toplev.c (general_init): Create the singleton gcc::context instance.
6795         * context.c: New.
6797         * context.h: New.
6799 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
6801         PR rtl-optimization/57968
6802         * mode-switching.c (create_pre_exit): Allow instructions that
6803         don't set a return register to need a non-exit mode.
6805 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
6806             Anton Blanchard  <anton@au1.ibm.com>
6808         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
6809         operands to vperm for little endian.
6810         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
6811         of lvsl to create the control mask for a vperm for little endian.
6813 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6814             Anton Blanchard  <anton@au1.ibm.com>
6816         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
6817         two operands for little-endian.
6819 2013-07-23  Steve Ellcey  <sellcey@mips.com>
6821         * config/mips/mips.c (mips_case_values_threshold): New.
6822         (TARGET_CASE_VALUES_THRESHOLD): Define.
6824 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6825             Anton Blanchard  <anton@au1.ibm.com>
6827         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
6828         selection of field for vector splat in little endian mode.
6830 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
6832         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
6833         expanders to rs6000.md.
6834         (ior<mode>3): Likewise.
6835         (and<mode>3): Likewise.
6836         (one_cmpl<mode>2): Likewise.
6837         (nor<mode>3): Likewise.
6838         (andc<mode>3): Likewise.
6839         (eqv<mode>3): Likewise.
6840         (nand<mode>3): Likewise.
6841         (orc<mode>3): Likewise.
6843         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
6844         declaration.
6846         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
6847         to split multi-word logical operations.
6848         (rs6000_split_logical_di): Likewise.
6849         (rs6000_split_logical): Likewise.
6851         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
6852         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
6853         and allow TImode operations in 32-bit.
6854         (vsx_and<mode>3_64bit): Likewise.
6855         (vsx_ior<mode>3_32bit): Likewise.
6856         (vsx_ior<mode>3_64bit): Likewise.
6857         (vsx_xor<mode>3_32bit): Likewise.
6858         (vsx_xor<mode>3_64bit): Likewise.
6859         (vsx_one_cmpl<mode>2_32bit): Likewise.
6860         (vsx_one_cmpl<mode>2_64bit): Likewise.
6861         (vsx_nor<mode>3_32bit): Likewise.
6862         (vsx_nor<mode>3_64bit): Likewise.
6863         (vsx_andc<mode>3_32bit): Likewise.
6864         (vsx_andc<mode>3_64bit): Likewise.
6865         (vsx_eqv<mode>3_32bit): Likewise.
6866         (vsx_eqv<mode>3_64bit): Likewise.
6867         (vsx_nand<mode>3_32bit): Likewise.
6868         (vsx_nand<mode>3_64bit): Likewise.
6869         (vsx_orc<mode>3_32bit): Likewise.
6870         (vsx_orc<mode>3_64bit): Likewise.
6872         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
6873         logical types in GPRs.
6875         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
6876         logical insns to rs6000.md, and allow TImode operations in
6877         32-bit.
6878         (altivec_ior<mode>3): Likewise.
6879         (altivec_xor<mode>3): Likewise.
6880         (altivec_one_cmpl<mode>2): Likewise.
6881         (altivec_nor<mode>3): Likewise.
6882         (altivec_andc<mode>3): Likewise.
6884         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
6885         attributes for moving the 128-bit logical operations into
6886         rs6000.md.
6887         (BOOL_REGS_OUTPUT): Likewise.
6888         (BOOL_REGS_OP1): Likewise.
6889         (BOOL_REGS_OP2): Likewise.
6890         (BOOL_REGS_UNARY): Likewise.
6891         (BOOL_REGS_AND_CR0): Likewise.
6892         (one_cmpl<mode>2): Add support for DI logical operations on
6893         32-bit, splitting the operations to 32-bit.
6894         (anddi3): Likewise.
6895         (iordi3): Likewise.
6896         (xordi3): Likewise.
6897         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
6898         changes to combine the 32/64-bit code, allow logical operations on
6899         TI mode in 32-bit, and to use similar match_operator patterns like
6900         scalar mode uses.  Combine the Altivec and VSX code for logical
6901         operations, and move it here.
6902         (ior<mode>3, 128-bit types): Likewise.
6903         (xor<mode>3, 128-bit types): Likewise.
6904         (one_cmpl<mode>3, 128-bit types): Likewise.
6905         (nor<mode>3, 128-bit types): Likewise.
6906         (andc<mode>3, 128-bit types): Likewise.
6907         (eqv<mode>3, 128-bit types): Likewise.
6908         (nand<mode>3, 128-bit types): Likewise.
6909         (orc<mode>3, 128-bit types): Likewise.
6910         (and<mode>3_internal): Likewise.
6911         (bool<mode>3_internal): Likewise.
6912         (boolc<mode>3_internal1): Likewise.
6913         (boolc<mode>3_internal2): Likewise.
6914         (boolcc<mode>3_internal1): Likewise.
6915         (boolcc<mode>3_internal2): Likewise.
6916         (eqv<mode>3_internal1): Likewise.
6917         (eqv<mode>3_internal2): Likewise.
6918         (one_cmpl1<mode>3_internal): Likewise.
6920 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
6922         * config/microblaze/microblaze.c (microblaze_expand_prologue):
6923         Rename flag_stack_usage to flag_stack_usage_info.
6925 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
6927         * config/microblaze/sync.md: New file.
6928         * config/microblaze/microblaze.md: Include sync.md
6929         * config/microblaze/microblaze.c: Add print_operand 'y'.
6930         * config/microblaze/constraints.md: Add memory_contraint
6931         'Q' which is a single register.
6933 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
6935         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
6937 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
6939         * reload.c (find_reloads): Exit loop once we find this operand
6940         cannot be reloaded somehow for this alternative.
6942         * reload.c (find_reloads): Exit loop once we find a hard register.
6944         * rtlanal.c (computed_jump_p): Exit loop once we find label
6945         reference is used.
6947         * i386.c (ix86_pad_returns): Exit loop after setting replace.
6949         * cfgloopmanip.c (remove_path): Exit loop after setting
6950         irred_invalidated.
6952         * gensupport.c (subst_dup): Avoid loop if code is not
6953         MATCH_DUP nor MATCH_OP_DUP.
6955 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
6957         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
6959 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6961         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
6962         true for SP_REGNUM if mode == ptr_mode.
6963         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
6964         with value R0_REGNUM + 31.
6966 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6968         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
6969         pad pointer-typed argument downward.
6971 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6973         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
6974         and __ILP32__ when the ILP32 model is in use.
6976 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
6978         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
6979         (aarch64_load_symref_appropriately): In the case of
6980         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
6981         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
6982         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
6983         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
6984         if 'mode' doesn't equal to 'ptr_mode'.
6985         (aarch64_output_mi_thunk): Add an assertion on the alignment of
6986         'vcall_offset'; change to call aarch64_emit_move differently depending
6987         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
6988         to calculate the upper bound of 'vcall_offset'.
6989         (aarch64_cannot_force_const_mem): Change to also return true if
6990         mode != ptr_mode.
6991         (aarch64_legitimize_reload_address): In the case of large
6992         displacements, add new local variable 'xmode' and an assertion
6993         based on it; change to use 'xmode' to generate the new rtx and
6994         reload.
6995         (aarch64_asm_trampoline_template): Change to generate the template
6996         differently depending on TARGET_ILP32 or not; change to use
6997         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
6998         (aarch64_trampoline_size): Removed.
6999         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
7000         and replace immediate literals with it.  Change to use 'ptr_mode'
7001         instead of 'DImode' and call convert_memory_address if the mode
7002         of 'fnaddr' doesn't equal to 'ptr_mode'.
7003         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
7004         to output symbol.
7005         (aarch64_elf_asm_destructor): Likewise.
7006         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
7007         on TARGET_ILP32 instead of aarch64_trampoline_size.
7008         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
7009         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
7010         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
7011         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
7012         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
7013         (storewb_pair<GPI:mode>_<P:mode>): ... this.
7014         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
7015         depending on the value of 'mode'.
7016         (add_losym_<mode>): New.
7017         (ldr_got_small_<mode>): New, based on ldr_got_small.
7018         (ldr_got_small): Remove.
7019         (ldr_got_small_sidi): New.
7020         * config/aarch64/iterators.md (P): New.
7021         (PTR): Change to 'ptr_mode' in the condition.
7023 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
7025         * config.gcc (aarch64*-*-*): Support --with-abi.
7026         (aarch64*-*-elf): Support --with-multilib-list.
7027         (aarch64*-*-linux*): Likewise.
7028         (supported_defaults): Add abi to aarch64*-*-*.
7029         * configure.ac: Mention AArch64 for --with-multilib-list.
7030         * configure: Re-generated.
7031         * config/aarch64/biarchilp32.h: New file.
7032         * config/aarch64/biarchlp64.h: New file.
7033         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
7034         (ABI_SPEC): Ditto.
7035         (MULTILIB_DEFAULTS): Ditto.
7036         (DRIVER_SELF_SPECS): Ditto.
7037         (ASM_SPEC): Update to also substitute -mabi.
7038         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
7039         file whose name depends on -mabi= and -mbig-endian.
7040         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
7041         TARGET_ILP32.
7042         (POINTER_SIZE): New define.
7043         (POINTERS_EXTEND_UNSIGNED): Ditto.
7044         (enum aarch64_abi_type): New enumeration tag.
7045         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
7046         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
7047         (TARGET_ILP32): New define.
7048         * config/aarch64/aarch64.opt (mabi): New.
7049         (aarch64_abi): New.
7050         (ilp32, lp64): New values for -mabi.
7051         * config/aarch64/t-aarch64 (comma): New define.
7052         (MULTILIB_OPTIONS): Ditto.
7053         (MULTILIB_DIRNAMES): Ditto.
7054         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
7055         * doc/invoke.texi: Document -mabi for AArch64.
7057 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
7059         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
7061 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
7062             Anton Blanchard  <anton@au1.ibm.com>
7064         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
7065         endianness when selecting field to splat.
7067 2013-07-22  Eric Christopher  <echristo@gmail.com>
7069         * dwarf2out.c (die_odr_checksum): New function to use
7070         CHECKSUM_ macros and ULEB128 for DIE tag.
7071         (generate_type_signature): Use.
7073 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
7075         * config.gcc (sparc*-*-*): Accept leon3 processor.
7076         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
7077         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
7078         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
7079         * config/sparc/sparc.opt (enum processor_type): Add leon3.
7080         (mfix-ut699): Adjust comment.
7081         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
7082         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
7083         (CPP_CPU_SPEC): Likewise.
7084         (ASM_CPU_SPEC): Likewise.
7085         * config/sparc/sparc.c (leon3_cost): New constant.
7086         (sparc_option_override): Add leon3 support.
7087         (mem_ref): New function.
7088         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
7089         (sparc_do_work_around_errata): Look into the instruction in the delay
7090         slot and adjust accordingly.  Add fix for the data cache nullify issues
7091         of the UT699.  Change insertion position for the NOP.
7092         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
7093         (leon3_load): New reservation.
7094         (leon_store): Bump latency to 2.
7095         (grfpu): New automaton.
7096         (grfpu_alu): New unit.
7097         (grfpu_ds): Likewise.
7098         (leon_fp_alu): Adjust.
7099         (leon_fp_mult): Delete.
7100         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
7101         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
7102         * config/sparc/sparc.md (cpu): Add leon3.
7103         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
7104         (swapsi): Likewise.
7105         (atomic_test_and_set): Likewise.
7106         (ldstub): Likewise.
7108 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
7110         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
7111         default for R5900 targets.
7112         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
7113         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
7114         * config/mips/mips.c (mips_option_override): Report an error for
7115         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
7116         for -march=r5900 -mhard-float.
7118 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
7120         * df-problems.c (can_move_insns_across): Exit loop once we
7121         find a non-fixed, non-global register.
7123         * ipa-pure-const.c (propagate_nothrow): Exit loop after
7124         setting can_throw.
7126         * omega.c (omega_eliminate_red): Break after setting red_found.
7127         (omega_problem_has_red_equations): Similarly after setting found.
7128         (omega_query_variable): Similarly after setting coupled.
7130 2013-07-22  Marek Polacek  <polacek@redhat.com>
7132         * gimplify.c: Don't include gimple.h twice.
7134 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7136         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
7137         instead of TARGET_THUMB1.
7138         (Pz): New constraint.
7139         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
7140         encodings.
7141         (compare_negsi_si): Likewise.
7142         (compare_addsi2_op0): Likewise.
7143         (compare_addsi2_op1): Likewise.
7144         (addsi3_carryin_<optab>): Likewise.
7145         (addsi3_carryin_alt2_<optab>): Likewise.
7146         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
7147         for arm_restrict_it.
7148         (subsi3_carryin): Likewise.
7149         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
7150         (minmax_arithsi): Disable for arm_restrict_it.
7151         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
7152         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
7153         (satsi_<SAT:code>_shift): Likewise.
7154         (arm_shiftsi3): Add alternative for 16-bit encoding.
7155         (arm32_movhf): Disable for arm_restrict_it.
7156         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
7157         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
7159 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
7161         * config/arm/arm.md (attribute "insn"): Delete.
7162         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
7163         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
7164         (not_shiftsi): Update for attribute change.
7165         (not_shiftsi_compare0): Likewise.
7166         (not_shiftsi_compare0_scratch): Likewise.
7167         (arm_one_cmplsi2): Likewise.
7168         (thumb1_one_cmplsi2): Likewise.
7169         (notsi_compare0): Likewise.
7170         (notsi_compare0_scratch): Likewise.
7171         (thumb1_movdi_insn): Likewise.
7172         (arm_movsi_insn): Likewise.
7173         (movhi_insn_arch4): Likewise.
7174         (movhi_bytes): Likewise.
7175         (arm_movqi_insn): Likewise.
7176         (thumb1_movqi_insn): Likewise.
7177         (arm32_movhf): Likewise.
7178         (thumb1_movhf): Likewise.
7179         (arm_movsf_soft_insn): Likewise.
7180         (thumb1_movsf_insn): Likewise.
7181         (thumb_movdf_insn): Likewise.
7182         (movsicc_insn): Likewise.
7183         (movsfcc_soft_insn): Likewise.
7184         (and_scc): Likewise.
7185         (cond_move): Likewise.
7186         (if_move_not): Likewise.
7187         (if_not_move): Likewise.
7188         (if_shift_move): Likewise.
7189         (if_move_shift): Likewise.
7190         (if_shift_shift): Likewise.
7191         (if_not_arith): Likewise.
7192         (if_arith_not): Likewise.
7193         (cond_move_not): Likewise.
7194         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
7195         (neon_mov<mode>): Likewise.
7196         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
7197         (thumb2_movsi_vfp): Likewise.
7198         (movsf_vfp): Likewise.
7199         (thumb2_movsf_vfp): Likewise.
7200         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
7201         change.
7202         (cortexa7_older_only): Likewise.
7203         (cortexa7_younger): Likewise.
7204         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
7205         (1020alu_shift_op): Likewise.
7206         (1020alu_shift_reg_op): Likewise.
7207         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
7208         (alu_shift_op): Likewise.
7209         (alu_shift_reg_op): Likewise.
7210         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
7211         (11_alu_shift_op): Likewise.
7212         (11_alu_shift_reg_op): Likewise.
7213         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
7214         (9_alu_shift_reg_op): Likewise.
7215         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7216         change.
7217         (cortex_a15_alu_shift): Likewise.
7218         (cortex_a15_alu_shift_reg): Likewise.
7219         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
7220         (cortex_a5_alu_shift): Likewise.
7221         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
7222         change.
7223         (cortex_a53_alu_shift): Likewise.
7224         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
7225         change.
7226         (cortex_a7_alu_reg): Likewise.
7227         (cortex_a7_alu_shift): Likewise.
7228         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
7229         (cortex_a8_alu_shift): Likewise.
7230         (cortex_a8_alu_shift_reg): Likewise.
7231         (cortex_a8_mov): Likewise.
7232         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
7233         (cortex_a9_dp_shift): Likewise.
7234         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
7235         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
7236         (cortex_r4_mov): Likewise.
7237         (cortex_r4_alu_shift): Likewise.
7238         (cortex_r4_alu_shift_reg): Likewise.
7239         * config/arm/fa526.md (526_alu_op): Update for attribute change.
7240         (526_alu_shift_op): Likewise.
7241         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
7242         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
7243         (626te_alu_shift_op): Likewise.
7244         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
7245         (726te_alu_op): Likewise.
7246         (726te_alu_shift_op): Likewise.
7247         (726te_alu_shift_reg_op): Likewise.
7248         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
7249         (mp626_alu_shift_op): Likewise.
7250         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
7251         (pj4_alu_e1_conds): Likewise.
7252         (pj4_alu): Likewise.
7253         (pj4_alu_conds): Likewise.
7254         (pj4_shift): Likewise.
7255         (pj4_shift_conds): Likewise.
7256         (pj4_alu_shift): Likewise.
7257         (pj4_alu_shift_conds): Likewise.
7259 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7261         * config/arm/predicates.md (shiftable_operator_strict_it):
7262         New predicate.
7263         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
7264         Disable cond_exec version for arm_restrict_it.
7265         (thumb2_smaxsi3): Convert to generate cond_exec.
7266         (thumb2_sminsi3): Likewise.
7267         (thumb32_umaxsi3): Likewise.
7268         (thumb2_uminsi3): Likewise.
7269         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
7270         (thumb2_neg_abssi2): Likewise.
7271         (thumb2_mov_scc): Add alternative for 16-bit encoding.
7272         (thumb2_movsicc_insn): Adjust alternatives.
7273         (thumb2_mov_negscc): Disable for arm_restrict_it.
7274         (thumb2_mov_negscc_strict_it): New pattern.
7275         (thumb2_mov_notscc_strict_it): New pattern.
7276         (thumb2_mov_notscc): Disable for arm_restrict_it.
7277         (thumb2_ior_scc): Likewise.
7278         (thumb2_ior_scc_strict_it): New pattern.
7279         (thumb2_cond_move): Adjust for arm_restrict_it.
7280         (thumb2_cond_arith): Disable for arm_restrict_it.
7281         (thumb2_cond_arith_strict_it): New pattern.
7282         (thumb2_cond_sub): Adjust for arm_restrict_it.
7283         (thumb2_movcond): Likewise.
7284         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
7285         (thumb2_zero_extendhisi2_v6): Likewise.
7286         (thumb2_zero_extendqisi2_v6): Likewise.
7287         (orsi_notsi_si): Likewise.
7288         (orsi_not_shiftsi_si): Likewise.
7290 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
7292         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
7293         instruction sequence is 1 byte shorter.
7295 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
7297         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
7298         it is not needed after split.
7300 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
7302         PR target/51784
7303         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
7304         second label for nonlocal goto receivers. Don't output pic base labels
7305         unless we're producing PIC; mark that action unreachable().
7306         (ix86_save_reg): If the function contains a nonlocal label, save the
7307         PIC base reg.
7308         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
7309         * config/darwin.c (emitted_pic_label_num): New GTY.
7310         (update_pic_label_number_if_needed): New.
7311         (machopic_output_function_base_name): Adjust for nonlocal receiver
7312         case.
7313         (machopic_should_output_picbase_label): New.
7314         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
7315         (nonlocal_goto_receiver): New insn and split.
7317 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
7319         * config/aarch64/aarch64-builtins.c
7320         (aarch64_fold_builtin): Fold abs in all modes.
7321         * config/aarch64/aarch64-simd-builtins.def
7322         (abs): Enable for all modes.
7323         * config/aarch64/arm_neon.h
7324         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
7325         (vabs_f64): Add missing intrinsic.
7327 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
7329         * config/aarch64/arm_neon.h (vabs_s64): New function
7331 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
7333         PR target/57516
7334         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
7335         * config/avr/avr.md (adjust_len): Add `round'.
7336         * config/avr/avr-protos.h (avr_out_round): New prototype.
7337         (avr_out_plus): Add `out_label' argument.
7338         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
7339         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
7340         Handle the case where `insn' is just a pattern.
7341         (avr_out_bitop): Handle the case where `insn' is just a pattern.
7342         (avr_out_round): New function.
7343         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
7345 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
7347         * config/microblaze/microblaze.c (microblaze_expand_prologue):
7348         Add check for flag_stack_usage to handle -fstack-usage support
7350 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
7352         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
7353         interaction for new Power8 flags and VSX.
7355 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
7357         PR middle-end/57698
7358         * tree-inline.c (expand_call_inline): Emit errors during
7359         early_inlining only if optimization is not turned on.
7361 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
7363         * passes.def: New.
7365         * passes.c (init_optimization_passes): Move the construction of
7366         the pass hierarchy into a new passes.def file.
7368         * Makefile.in (passes.o): Add dependency on passes.def.
7370 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
7372         * passes.c (init_optimization_passes): Introduce macros for
7373         constructing the tree of passes (INSERT_PASSES_AFTER,
7374         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
7375         TERMINATE_PASS_LIST).
7377 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
7378             Wei Mi  <wmi@google.com>
7380         PR rtl-optimization/57878
7381         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
7382         top.
7383         (reload_pseudo_compare_func): Check nregs first for reload
7384         pseudos.
7386 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
7388         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
7390 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
7392         * read-rtl.c (validate_const_int): Once an invalid character is
7393         seen, quit the loop.
7395         * gengtype.c (write_roots): Similarly once we find the "deletable"
7396         or "if_marked" option.
7398 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
7400         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
7401         "xtab" and "sat".  Move value "clz" from here to ...
7402         (attriubte "type"): ... here.
7403         (satsi_<SAT:code>): Delete "insn" attribute.
7404         (satsi_<SAT:code>_shift): Likewise.
7405         (arm_zero_extendqisi2addsi): Likewise.
7406         (arm_extendqisi2addsi): Likewise.
7407         (clzsi2): Update for attribute changes.
7408         (rbitsi2): Likewise.
7409         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
7410         attribute.
7411         (arm_usatsihi): Likewise.
7412         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
7414 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
7416         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
7417         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
7418         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
7419         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
7420         in alphabetical order.
7421         (attribute "core_cycles"): Update for attribute changes.
7422         (arm_addsi3): Likewise.
7423         (addsi3_compare0): Likewise.
7424         (addsi3_compare0_scratch): Likewise.
7425         (addsi3_compare_op1): Likewise.
7426         (addsi3_compare_op2): Likewise.
7427         (compare_addsi2_op0): Likewise.
7428         (compare_addsi2_op1): Likewise.
7429         (addsi3_carryin_shift_<optab>): Likewise.
7430         (subsi3_carryin_shift): Likewise.
7431         (rsbsi3_carryin_shift): Likewise.
7432         (arm_subsi3_insn): Likewise.
7433         (subsi3_compare0): Likewise.
7434         (subsi3_compare): Likewise.
7435         (arm_andsi3_insn): Likewise.
7436         (thumb1_andsi3_insn): Likewise.
7437         (andsi3_compare0): Likewise.
7438         (andsi3_compare0_scratch): Likewise.
7439         (zeroextractsi_compare0_scratch
7440         (andsi_not_shiftsi_si): Likewise.
7441         (iorsi3_insn): Likewise.
7442         (iorsi3_compare0): Likewise.
7443         (iorsi3_compare0_scratch): Likewise.
7444         (arm_xorsi3): Likewise.
7445         (thumb1_xorsi3_insn): Likewise.
7446         (xorsi3_compare0): Likewise.
7447         (xorsi3_compare0_scratch): Likewise.
7448         (satsi_<SAT:code>_shift): Likewise.
7449         (rrx): Likewise.
7450         (arm_shiftsi3): Likewise.
7451         (shiftsi3_compare0): Likewise.
7452         (not_shiftsi): Likewise.
7453         (not_shiftsi_compare0): Likewise.
7454         (not_shiftsi_compare0_scratch): Likewise.
7455         (arm_one_cmplsi2): Likewise.
7456         (thumb_one_complsi2): Likewise.
7457         (notsi_compare0): Likewise.
7458         (notsi_compare0_scratch): Likewise.
7459         (thumb1_zero_extendhisi2): Likewise.
7460         (arm_zero_extendhisi2): Likewise.
7461         (arm_zero_extendhisi2_v6): Likewise.
7462         (arm_zero_extendhisi2addsi): Likewise.
7463         (thumb1_zero_extendqisi2): Likewise.
7464         (thumb1_zero_extendqisi2_v6): Likewise.
7465         (arm_zero_extendqisi2): Likewise.
7466         (arm_zero_extendqisi2_v6): Likewise.
7467         (arm_zero_extendqisi2addsi): Likewise.
7468         (thumb1_extendhisi2): Likewise.
7469         (arm_extendhisi2): Likewise.
7470         (arm_extendhisi2_v6): Likewise.
7471         (arm_extendqisi): Likewise.
7472         (arm_extendqisi_v6): Likewise.
7473         (arm_extendqisi2addsi): Likewise.
7474         (thumb1_extendqisi2): Likewise.
7475         (thumb1_movdi_insn): Likewise.
7476         (arm_movsi_insn): Likewise.
7477         (movsi_compare0): Likewise.
7478         (movhi_insn_arch4): Likewise.
7479         (movhi_bytes): Likewise.
7480         (arm_movqi_insn): Likewise.
7481         (thumb1_movqi_insn): Likewise.
7482         (arm32_movhf): Likewise.
7483         (thumb1_movhf): Likewise.
7484         (arm_movsf_soft_insn): Likewise.
7485         (thumb1_movsf_insn): Likewise.
7486         (movdf_soft_insn): Likewise.
7487         (thumb_movdf_insn): Likewise.
7488         (arm_cmpsi_insn): Likewise.
7489         (cmpsi_shiftsi): Likewise.
7490         (cmpsi_shiftsi_swp): Likewise.
7491         (arm_cmpsi_negshiftsi_si): Likewise.
7492         (movsicc_insn): Likewise.
7493         (movsfcc_soft_insn): Likewise.
7494         (arith_shiftsi): Likewise.
7495         (arith_shiftsi_compare0
7496         (arith_shiftsi_compare0_scratch
7497         (sub_shiftsi): Likewise.
7498         (sub_shiftsi_compare0
7499         (sub_shiftsi_compare0_scratch
7500         (and_scc): Likewise.
7501         (cond_move): Likewise.
7502         (if_plus_move): Likewise.
7503         (if_move_plus): Likewise.
7504         (if_move_not): Likewise.
7505         (if_not_move): Likewise.
7506         (if_shift_move): Likewise.
7507         (if_move_shift): Likewise.
7508         (if_shift_shift): Likewise.
7509         (if_not_arith): Likewise.
7510         (if_arith_not): Likewise.
7511         (cond_move_not): Likewise.
7512         (thumb1_ashlsi3): Set type attribute.
7513         (thumb1_ashrsi3): Likewise.
7514         (thumb1_lshrsi3): Likewise.
7515         (thumb1_rotrsi3): Likewise.
7516         (shiftsi3_compare0_scratch): Likewise.
7517         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
7518         (neon_mov<mode>): Likewise.
7519         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
7520         attribute changes.
7521         (thumb2_movsi_insn): Likewise.
7522         (thumb2_cmpsi_neg_shiftsi): Likewise.
7523         (thumb2_extendqisi_v6): Likewise.
7524         (thumb2_zero_extendhisi2_v6): Likewise.
7525         (thumb2_zero_extendqisi2_v6): Likewise.
7526         (thumb2_shiftsi3_short): Likewise.
7527         (thumb2_addsi3_compare0_scratch): Likewise.
7528         (orsi_not_shiftsi_si): Likewise.
7529         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
7530         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
7531         changes.
7532         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
7533         (1020alu_shift_op): Likewise.
7534         (1020alu_shift_reg_op): Likewise.
7535         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
7536         (alu_shift_op): Likewise.
7537         (alu_shift_reg_op): Likewise.
7538         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
7539         (11_alu_shift_op): Likewise.
7540         (11_alu_shift_reg_op): Likewise.
7541         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
7542         (9_alu_shift_reg_op): Likewise.
7543         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7544         changes.
7545         (cortex_a15_alu_shift): Likewise.
7546         (cortex_a15_alu_shift_reg): Likewise.
7547         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
7548         changes.
7549         (cortex_a5_alu_shift): Likewise.
7550         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
7551         changes.
7552         (cortex_a53_alu_shift): Likewise.
7553         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
7554         changes.
7555         (cortex_a7_alu_reg): Likewise.
7556         (cortex_a7_alu_shift): Likewise.
7557         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
7558         changes.
7559         (cortex_a8_alu_shift): Likewise.
7560         (cortex_a8_alu_shift_reg): Likewise.
7561         (cortex_a8_mov): Likewise.
7562         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
7563         (cortex_a9_dp_shift): Likewise.
7564         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
7565         changes.
7566         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
7567         changes.
7568         (cortex_r4_mov): Likewise.
7569         (cortex_r4_alu_shift): Likewise.
7570         (cortex_r4_alu_shift_reg): Likewise.
7571         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
7572         (526_alu_shift_op): Likewise.
7573         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
7574         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
7575         (626te_alu_shift_op): Likewise.
7576         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
7577         (726te_alu_op): Likewise.
7578         (726te_alu_shift_op): Likewise.
7579         (726te_alu_shift_reg_op): Likewise.
7580         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
7581         (mp626_alu_shift_op): Likewise.
7582         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
7583         (pj4_alu_e1_conds): Likewise.
7584         (pj4_alu): Likewise.
7585         (pj4_alu_conds): Likewise.
7586         (pj4_shift): Likewise.
7587         (pj4_shift_conds): Likewise.
7588         (pj4_alu_shift): Likewise.
7589         (pj4_alu_shift_conds): Likewise.
7590         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
7591         changes.
7592         (cortexa7_older_only): Likewise.
7593         (cortexa7_younger): Likewise.
7595 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
7597         * ipa-pure-const.c (generate_summary): Rename to...
7598         (pure_const_generate_summary): ... this.
7600 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
7602         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
7604 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
7606         PR target/57909
7607         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
7608         usage in HI mode.
7610 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7612         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
7613         enabled without -march=zEC12.
7614         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
7615         flags to be set.
7617 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
7619         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
7620         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
7621         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
7622         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
7623         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
7624         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
7625         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
7626         ISA_HAS_FP4.
7627         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
7628         and ISA_HAS_NMADD3_NMSUB3.
7629         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
7630         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
7631         (nmsub4<mode>, nmsub3<mode>): Likewise.
7632         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
7634 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
7636         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
7637         TARGET_MIPS5400 checking.
7639 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
7640             Peter Bergner  <bergner@vnet.ibm.com>
7642         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
7643         registers in the comment.
7644         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
7645         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
7646         rather than FIRST_PSEUDO_REGISTERS.
7648 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
7650         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
7651         enable extra ISA flags with TARGET_HTM.
7653 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
7655         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
7656         Fix comment typos.
7658 2013-07-15  Cong Hou  <congh@google.com>
7660         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
7661         in compare function for sorting.
7663 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
7665         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
7666         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
7667         * config/rs6000/rs6000.opt: Add -mhtm option.
7668         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
7669         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
7670         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
7671         __HTM__ if the HTM instructions are available.
7672         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
7673         htm_spr_reg_operand): New define_predicates.
7674         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
7675         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
7676         Include htm.md.
7677         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
7678         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
7679         HTM builtin functions.
7680         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
7681         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
7682         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
7683         (rs6000_builtin_mask_calculate): Likewise.
7684         (rs6000_option_override_internal): Likewise.
7685         (bdesc_htm): Add new HTM builtin support.
7686         (htm_spr_num): New function.
7687         (htm_spr_regno): Likewise.
7688         (rs6000_htm_spr_icode): Likewise.
7689         (htm_expand_builtin): Likewise.
7690         (htm_init_builtins): Likewise.
7691         (rs6000_expand_builtin): Add support for HTM builtin functions.
7692         (rs6000_init_builtins): Likewise.
7693         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
7694         option.
7695         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
7696         (TARGET_HTM, MASK_HTM): Define macros.
7697         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
7698         (FIXED_REGISTERS): Likewise.
7699         (CALL_USED_REGISTERS): Likewise.
7700         (CALL_REALLY_USED_REGISTERS): Likewise.
7701         (REG_ALLOC_ORDER): Likewise.
7702         (enum reg_class): Likewise.
7703         (REG_CLASS_NAMES): Likewise.
7704         (REG_CLASS_CONTENTS): Likewise.
7705         (REGISTER_NAMES): Likewise.
7706         (ADDITIONAL_REGISTER_NAMES): Likewise.
7707         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
7708         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
7709         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
7710         * config/rs6000/htm.md: New file.
7711         * config/rs6000/htmintrin.h: New file.
7712         * config/rs6000/htmxlintrin.h: New file.
7714 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7716         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
7717         Define SYMBOL_TINY_GOT, update comment.
7718         * config/aarch64/aarch64.c
7719         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
7720         (aarch64_expand_mov_immediate): Likewise.
7721         (aarch64_print_operand): Likewise.
7722         (aarch64_classify_symbol): Likewise.
7723         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
7724         (ldr_got_tiny): Define.
7726 2013-07-13  Tobias Grosser  <tobias@grosser.es>
7728         PR tree-optimization/54094
7729         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
7730           scheduling dimension for the parallelism check from the polyhedral
7731           information in the AST.
7732         * graphite-dependences.c (carries_deps): Do not assume the schedule is
7733           in 2D + 1 form.
7735 2013-07-13  Jason Merrill  <jason@redhat.com>
7737         * print-tree.c (debug_vec_tree): Use debug_raw.
7738         (debug_raw (vec<tree, va_gc> &)): New.
7739         (debug_raw (vec<tree, va_gc> *)): New.
7740         * tree.h: Declare them.
7742 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
7744         * ifcvt.c (ifcvt_after_combine): New static variable.
7745         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
7746         for size.
7747         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
7748         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
7749         rest_of_handle_if_after_reload): Pass new argument for if_convert.
7751 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
7753         * config/mips/mips.c (mips_expand_call): Remove empty statement.
7755 2013-07-12  Michael Matz  <matz@suse.de>
7757         PR middle-end/55771
7758         * convert.c (convert_to_real): Reject non-float inner types.
7760 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
7762         * config/aarch64/aarch64-protos.h
7763         (aarch64_simd_immediate_valid_for_move): Remove.
7764         * config/aarch64/aarch64.c (simd_immediate_info): New member.
7765         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
7766         cases.
7767         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
7769 2013-07-11  Steve Ellcey  <sellcey@mips.com>
7771         * config/mips/mips.c (mips_conditional_register_usage): Do not
7772         use t[0-7] registers in MIPS16 mode when optimizing for size.
7774 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
7776         * config/i386/i386.c (dispatch_function_versions): Fix array
7777         indexing of function_version_info to match actual_versions.
7779 2013-07-11  Teresa Johnson  <tejohnson@google.com>
7781         * vec.h (struct va_gc): Move release out-of-line.
7782         (va_gc::release): Call ggc_free on released vec.
7784 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
7786         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
7787         Require GOT register as additional operand in UNSPEC.
7788         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
7789         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
7790         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
7791         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
7792         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
7793         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
7794         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
7796 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
7798         PR target/57631
7799         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
7800         name seen by assembler/linker rather if available.
7802 2013-07-11  Andreas Schwab  <schwab@suse.de>
7804         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
7806 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
7808         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
7810 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
7812         * doc/tm.texi.in: Move hook documentation to ....
7813         * target.def: ... here.
7815         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
7816         text on @hook line.
7817         * doc/tm.texi: Regenerate.
7819 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
7821         PR c++/57869
7822         * doc/invoke.texi: Document -Wconditionally-supported.
7824 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
7826         PR target/57844
7827         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
7828         of my_fp.
7830 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
7832         PR target/57506
7833         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
7834         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
7835         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
7836         Remove duplicate devices.
7837         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
7838         * config/avr/t-multilib: Regenerate.
7839         * config/avr/avr-tables.opt: Regenerate.
7840         * doc/avr-mmcu.texi: Regenerate.
7842 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
7844         PR target/56987
7845         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
7847 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
7849         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
7850         the cost of MULT when optimizing for size.
7852 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
7854         * config/cr16/cr16-protos.h: Don't include target.h.
7856 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
7858         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
7859         adjust register size for TDmode and TFmode for VSX registers.
7861 2013-07-08  Kai Tietz  <ktietz@redhat.com>
7863         PR target/56892
7864         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
7865         hook_bool_const_tree_true.
7867 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7869         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
7870         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
7871         * config/s390/s390.md: Define FPR*_REGNUM constants.
7872         Fix FPR2_REGNUM constant (18 -> 17).
7873         ("*trunc<BFP:mode><DFP_ALL:mode>2")
7874         ("*trunc<DFP_ALL:mode><BFP:mode>2")
7875         ("trunc<BFP:mode><DFP_ALL:mode>2")
7876         ("trunc<DFP_ALL:mode><BFP:mode>2")
7877         ("*extend<BFP:mode><DFP_ALL:mode>2")
7878         ("*extend<DFP_ALL:mode><BFP:mode>2")
7879         ("extend<BFP:mode><DFP_ALL:mode>2")
7880         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
7881         FPR4_REGNUM.
7883 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
7885         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
7887 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7889         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
7890         and cfun_fpr_bit_p to cfun_fpr_save_p.
7891         (s390_frame_area, s390_register_info, s390_frame_info)
7892         (s390_emit_prologue, s390_emit_epilogue)
7893         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
7894         register numbers.
7895         * config/s390/s390.h: Define *_REGNUM macros for floating point
7896         register numbers.
7898 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
7900         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
7902 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
7904         PR rtl-optimization/57786
7905         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
7906         and break out of the loop when it is set to false.
7908 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
7910         PR target/57819
7911         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
7912         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
7913         (const_int 63)) 0)).
7914         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
7915         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
7916         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
7918         PR rtl-optimization/57829
7919         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
7920         mask bits outside of mode are just sign-extension from mode to HWI.
7922 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
7924         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
7925         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
7926         adjust_address instead of change_address to keep info about alignment.
7927         (emit_strmov): Remove.
7928         (emit_memmov): New function.
7929         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
7930         (expand_movmem_epilogue): Likewise and return updated rtx for
7931         destination.
7932         (expand_constant_movmem_prologue): Likewise and return updated rtx for
7933         destination and source.
7934         (decide_alignment): Refactor, handle vector_loop.
7935         (ix86_expand_movmem): Likewise.
7936         (ix86_expand_setmem): Likewise.
7937         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
7939 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
7941         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
7942         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
7944 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
7946         * config/i386/sse.md (sse_movlhps): Change alternative 3
7947         of operand 2 to "m".
7949 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
7951         PR target/57807
7952         * config/i386/sse.md (iptr): New mode attribute.
7953         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
7954         (<sse>_vm<plusminus_insn><mode>3): Ditto.
7955         (<sse>_vmmul<mode>3): Ditto.
7956         (<sse>_vmdiv<mode>3): Ditto.
7957         (sse_vmrcpv4sf2): Ditto.
7958         (<sse>_vmsqrt<mode>2): Ditto.
7959         (sse_vmrsqrtv4sf2): Ditto.
7960         (<sse>_vm<code><mode>3): Ditto.
7961         (avx_vmcmp<mode>3): Ditto.
7962         (<sse>_vmmaskcmp<mode>3): Ditto.
7963         (<sse>_comi): Ditto.
7964         (<sse>_ucomi): Ditto.
7965         (*xop_vmfrcz_<mode>): Ditto.
7966         (*fmai_fmadd_<mode>): Ditto.
7967         (*fmai_fmsub_<mode>): Ditto.
7968         (*fmai_fnmadd_<mode>): Ditto.
7969         (*fmai_fnmsub_<mode>): Ditto.
7970         (*fma4i_vmfmadd_<mode>): Ditto.
7971         (*fma4i_vmfmsub_<mode>): Ditto.
7972         (*fma4i_vmfnmadd_<mode>): Ditto.
7973         (*fma4i_vmfnmsub_<mode>): Ditto.
7974         (*xop_vmfrcz_<mode>): Ditto.
7975         (sse_cvtps2pi): Ditto.
7976         (sse_cvttps2pi): Ditto.
7977         (sse_cvtss2si): Ditto.
7978         (sse_cvtss2si_2): Ditto.
7979         (sse_cvtss2siq_2): Ditto.
7980         (sse_cvttss2si): Ditto.
7981         (sse_cvttss2siq): Ditto.
7982         (sse_cvtsd2si): Ditto.
7983         (sse_cvtsd2si_2): Ditto.
7984         (sse_cvtsd2siq_2): Ditto.
7985         (sse_cvttsd2si): Ditto.
7986         (sse_cvttsd2siq): Ditto.
7987         (sse_cvtsd2ss): Ditto.
7988         (sse_cvtss2sd): Ditto.
7989         (avx2_pbroadcast<mode>): Ditto.
7990         (avx2_pbroadcast<mode>_1): Ditto.
7991         (*avx_vperm_broadcast_v4sf): Ditto.
7993         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
7994         (sse_movlhps): Ditto.
7995         (sse_storehps): Ditto.
7996         (sse_loadhps): Ditto.
7997         (sse_storelps): Ditto.
7998         (sse_loadlps): Ditto.
7999         (*vec_concatv4sf): Ditto.
8000         (*vec_interleave_highv2df): Ditto.
8001         (*vec_interleave_lowv2df): Ditto.
8002         (*vec_extractv2df_1_sse): Ditto.
8003         (*vec_extractv2df_0_sse): Ditto.
8004         (sse2_storelpd): Ditto.
8005         (sse2_loadlpd): Ditto.
8006         (sse2_movsd): Ditto.
8007         (*vec_concatv4si): Ditto.
8008         (vec_concatv2di): Ditto.
8010         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
8011         for Intel asm dialect.
8012         (mmx_punpcklwd): Ditto.
8013         (mmx_punpckldq): Ditto.
8015         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
8016         for intel assembler dialect.
8018 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
8020         PR target/29776
8021         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
8022         for BUILT_IN_C{LZ,LRSB}*.
8023         * tree.h (CASE_INT_FN): Add FN##IMAX case.
8024         * tree-vrp.c (extract_range_basic): Handle
8025         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
8026         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
8027         fall thru to code calling set_value*.
8028         * builtins.c (expand_builtin): Remove *IMAX cases.
8029         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
8030         if width is bigger than 2*HWI.
8032 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
8034         PR rtl-optimization/55342
8035         * lra-int.h (lra_subreg_reload_pseudos): New.
8036         * lra.c: Add undoing optional reloads to the block diagram.
8037         (lra_subreg_reload_pseudos): New.
8038         (lra_optional_reload_pseudos): Change comments.
8039         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
8040         lra_optional_reload_pseudos after undo transformations.
8041         * lra-assigns.c (pseudo_prefix_title): New.
8042         (lra_setup_reg_renumber): Use it.
8043         (spill_for): Ditto.  Check subreg reload pseudos too.
8044         (assign_by_spills): Consider subreg reload pseudos too.
8045         * lra-constraints.c (simplify_operand_subreg): Use
8046         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
8047         (curr_insn_transform): Recognize and do optional reloads.
8048         (undo_optional_reloads): New.
8049         (lra_undo_inheritance): Call undo_optional_reloads.
8051 2013-07-05  Thomas Quinot  <quinot@adacore.com>
8053         * tree-complex.c (expand_complex_operations_1): Fix typo.
8055 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
8057         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
8058         (tune_params): New member 'const vec_costs'.
8059         * config/aarch64/aarch64.c (generic_vector_cost): New.
8060         (generic_tunings): New member 'generic_vector_cost'.
8061         (aarch64_builtin_vectorization_cost): New.
8062         (aarch64_add_stmt_cost): New.
8063         (TARGET_VECTORIZE_ADD_STMT_COST): New.
8064         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
8066 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
8068         PR target/57777
8069         * config/i386/predicates.md (vsib_address_operand): Disallow
8070         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
8072 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
8074         PR middle-end/55030
8075         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
8076         expand_builtin_setjmp_receiver.
8077         (expand_label): Adjust, call expand_builtin_setjmp_receiver
8078         with NULL for the label parameter.
8079         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
8080         the frame-pointer.  Adjust comments.
8081         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
8082         only if LABEL is non-NULL.
8084 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
8086         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
8087         (ARM_ABI_AAPCS64): Ditto.
8088         (arm_abi): Ditto.
8089         (ARM_DEFAULT_ABI): Ditto.
8091 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
8093         * config/aarch64/aarch64-builtins.c
8094         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
8095         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
8096         (st1): Likewise.
8097         * config/aarch64/aarch64-simd.md
8098         (aarch64_ld1<VALL:mode>): New.
8099         (aarch64_st1<VALL:mode>): Likewise.
8100         * config/aarch64/arm_neon.h
8101         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
8103 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
8105         * config/i386/i386.c (gate_insert_vzeroupper): Check if
8106         target ISA is AVX.
8107         (ix86_option_override_internal):Turn on all -mavx target flags by
8108         default as they are dependent on AVX anyway.
8110 2013-07-02  Cary Coutant  <ccoutant@google.com>
8112         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
8113         deterministic hash.
8114         (loc_checksum_ordered): Likewise.
8115         (hash_loc_operands): Remove inline keyword.
8117 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
8119         PR tree-optimization/57741
8120         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
8121         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
8122         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
8123         Allow REAL_CST step_exprs if flag_associative_math.
8124         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
8126 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
8128         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
8130 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
8132         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
8134 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8136         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
8137         encoding.
8138         (iorsi3_insn): Likewise.
8139         (arm_xorsi3): Likewise.
8141 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
8143         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
8144         to ...
8145         (attribute "type"): ... here, and prefix with "wmmx_".
8146         (attribute "core_cycles"): Update for attribute changes.
8147         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
8148         (tbcstv4hi): Likewise.
8149         (tbcstv2si): Likewise.
8150         (iwmmxt_iordi3): Likewise.
8151         (iwmmxt_xordi3): Likewise.
8152         (iwmmxt_anddi3): Likewise.
8153         (iwmmxt_nanddi3): Likewise.
8154         (iwmmxt_arm_movdi): Likewise.
8155         (iwmmxt_movsi_insn): Likewise.
8156         (mov<mode>_internal): Likewise.
8157         (and<mode>3_iwmmxt): Likewise.
8158         (ior<mode>3_iwmmxt): Likewise.
8159         (xor<mode>3_iwmmxt): Likewise.
8160         (add<mode>3_iwmmxt): Likewise.
8161         (ssaddv8qi3): Likewise.
8162         (ssaddv4hi3): Likewise.
8163         (ssaddv2si3): Likewise.
8164         (usaddv8qi3): Likewise.
8165         (usaddv4hi3): Likewise.
8166         (usaddv2si3): Likewise.
8167         (sub<mode>3_iwmmxt): Likewise.
8168         (sssubv8qi3): Likewise.
8169         (sssubv4hi3): Likewise.
8170         (sssubv2si3): Likewise.
8171         (ussubv8qi3): Likewise.
8172         (ussubv4hi3): Likewise.
8173         (ussubv2si3): Likewise.
8174         (mulv4hi3_iwmmxt): Likewise.
8175         (smulv4hi3_highpart): Likewise.
8176         (umulv4hi3_highpart): Likewise.
8177         (iwmmxt_wmacs): Likewise.
8178         (iwmmxt_wmacsz): Likewise.
8179         (iwmmxt_wmacu): Likewise.
8180         (iwmmxt_wmacuz): Likewise.
8181         (iwmmxt_clrdi): Likewise.
8182         (iwmmxt_clrv8qi): Likewise.
8183         (iwmmxt_clr4hi): Likewise.
8184         (iwmmxt_clr2si): Likewise.
8185         (iwmmxt_uavgrndv8qi3): Likewise.
8186         (iwmmxt_uavgrndv4hi3): Likewise.
8187         (iwmmxt_uavgv8qi3): Likewise.
8188         (iwmmxt_uavgv4hi3): Likewise.
8189         (iwmmxt_tinsrb): Likewise.
8190         (iwmmxt_tinsrh): Likewise.
8191         (iwmmxt_tinsrw): Likewise.
8192         (iwmmxt_textrmub): Likewise.
8193         (iwmmxt_textrmsb): Likewise.
8194         (iwmmxt_textrmuh): Likewise.
8195         (iwmmxt_textrmsh): Likewise.
8196         (iwmmxt_textrmw): Likewise.
8197         (iwmxxt_wshufh): Likewise.
8198         (eqv8qi3): Likewise.
8199         (eqv4hi3): Likewise.
8200         (eqv2si3): Likewise.
8201         (gtuv8qi3): Likewise.
8202         (gtuv4hi3): Likewise.
8203         (gtuv2si3): Likewise.
8204         (gtv8qi3): Likewise.
8205         (gtv4hi3): Likewise.
8206         (gtv2si3): Likewise.
8207         (smax<mode>3_iwmmxt): Likewise.
8208         (umax<mode>3_iwmmxt): Likewise.
8209         (smin<mode>3_iwmmxt): Likewise.
8210         (umin<mode>3_iwmmxt): Likewise.
8211         (iwmmxt_wpackhss): Likewise.
8212         (iwmmxt_wpackwss): Likewise.
8213         (iwmmxt_wpackdss): Likewise.
8214         (iwmmxt_wpackhus): Likewise.
8215         (iwmmxt_wpackwus): Likewise.
8216         (iwmmxt_wpackdus): Likewise.
8217         (iwmmxt_wunpckihb): Likewise.
8218         (iwmmxt_wunpckihh): Likewise.
8219         (iwmmxt_wunpckihw): Likewise.
8220         (iwmmxt_wunpckilb): Likewise.
8221         (iwmmxt_wunpckilh): Likewise.
8222         (iwmmxt_wunpckilw): Likewise.
8223         (iwmmxt_wunpckehub): Likewise.
8224         (iwmmxt_wunpckehuh): Likewise.
8225         (iwmmxt_wunpckehuw): Likewise.
8226         (iwmmxt_wunpckehsb): Likewise.
8227         (iwmmxt_wunpckehsh): Likewise.
8228         (iwmmxt_wunpckehsw): Likewise.
8229         (iwmmxt_wunpckelub): Likewise.
8230         (iwmmxt_wunpckeluh): Likewise.
8231         (iwmmxt_wunpckeluw): Likewise.
8232         (iwmmxt_wunpckelsb): Likewise.
8233         (iwmmxt_wunpckelsh): Likewise.
8234         (iwmmxt_wunpckelsw): Likewise.
8235         (ror<mode>3): Likewise.
8236         (ashr<mode>3_iwmmxt): Likewise.
8237         (lshr<mode>3_iwmmxt): Likewise.
8238         (ashl<mode>3_iwmmxt): Likewise.
8239         (ror<mode>3_di): Likewise.
8240         (ashr<mode>3_di): Likewise.
8241         (lshr<mode>3_di): Likewise.
8242         (ashl<mode>3_di): Likewise.
8243         (iwmmxt_wmadds): Likewise.
8244         (iwmmxt_wmaddu): Likewise.
8245         (iwmmxt_tmia): Likewise.
8246         (iwmmxt_tmiaph): Likewise.
8247         (iwmmxt_tmiabb): Likewise.
8248         (iwmmxt_tmiatb): Likewise.
8249         (iwmmxt_tmiabt): Likewise.
8250         (iwmmxt_tmiatt): Likewise.
8251         (iwmmxt_tmovmskb): Likewise.
8252         (iwmmxt_tmovmskh): Likewise.
8253         (iwmmxt_tmovmskw): Likewise.
8254         (iwmmxt_waccb): Likewise.
8255         (iwmmxt_wacch): Likewise.
8256         (iwmmxt_waccw): Likewise.
8257         (iwmmxt_waligni): Likewise.
8258         (iwmmxt_walignr): Likewise.
8259         (iwmmxt_walignr0): Likewise.
8260         (iwmmxt_walignr1): Likewise.
8261         (iwmmxt_walignr2): Likewise.
8262         (iwmmxt_walignr3): Likewise.
8263         (iwmmxt_wsadb): Likewise.
8264         (iwmmxt_wsadh): Likewise.
8265         (iwmmxt_wsadbz): Likewise.
8266         (iwmmxt_wsadhz): Likewise.
8267         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
8268         (iwmmxt_wabsdiffb): Likewise.
8269         (iwmmxt_wabsdiffh): Likewise.
8270         (iwmmxt_wabsdiffw): Likewise.
8271         (iwmmxt_waddsubhx): Likewise
8272         (iwmmxt_wsubaddhx): Likewise.
8273         (addc<mode>3): Likewise.
8274         (iwmmxt_avg4): Likewise.
8275         (iwmmxt_avg4r): Likewise.
8276         (iwmmxt_wmaddsx): Likewise.
8277         (iwmmxt_wmaddux): Likewise.
8278         (iwmmxt_wmaddsn): Likewise.
8279         (iwmmxt_wmaddun): Likewise.
8280         (iwmmxt_wmulwsm): Likewise.
8281         (iwmmxt_wmulwum): Likewise.
8282         (iwmmxt_wmulsmr): Likewise.
8283         (iwmmxt_wmulumr): Likewise.
8284         (iwmmxt_wmulwsmr): Likewise.
8285         (iwmmxt_wmulwumr): Likewise.
8286         (iwmmxt_wmulwl): Likewise.
8287         (iwmmxt_wqmulm): Likewise.
8288         (iwmmxt_wqmulwm): Likewise.
8289         (iwmmxt_wqmulmr): Likewise.
8290         (iwmmxt_wqmulwmr): Likewise.
8291         (iwmmxt_waddbhusm): Likewise.
8292         (iwmmxt_waddbhusl): Likewise.
8293         (iwmmxt_wqmiabb): Likewise.
8294         (iwmmxt_wqmiabt): Likewise.
8295         (iwmmxt_wqmiatb): Likewise.
8296         (iwmmxt_wqmiatt): Likewise.
8297         (iwmmxt_wqmiabbn): Likewise.
8298         (iwmmxt_wqmiabtn): Likewise.
8299         (iwmmxt_wqmiatbn): Likewise.
8300         (iwmmxt_wqmiattn): Likewise.
8301         (iwmmxt_wmiabb): Likewise.
8302         (iwmmxt_wmiabt): Likewise.
8303         (iwmmxt_wmiatb): Likewise.
8304         (iwmmxt_wmiatt): Likewise.
8305         (iwmmxt_wmiabbn): Likewise.
8306         (iwmmxt_wmiabtn): Likewise.
8307         (iwmmxt_wmiatbn): Likewise.
8308         (iwmmxt_wmiattn): Likewise.
8309         (iwmmxt_wmiawbb): Likewise.
8310         (iwmmxt_wmiawbt): Likewise.
8311         (iwmmxt_wmiawtb): Likewise.
8312         (iwmmxt_wmiawtt): Likewise.
8313         (iwmmxt_wmiawbbn): Likewise.
8314         (iwmmxt_wmiawbtn): Likewise.
8315         (iwmmxt_wmiawtbn): Likewise.
8316         (iwmmxt_wmiawttn): Likewise.
8317         (iwmmxt_wmerge): Likewise.
8318         (iwmmxt_tandc<mode>3): Likewise.
8319         (iwmmxt_torc<mode>3): Likewise.
8320         (iwmmxt_torvsc<mode>3): Likewise.
8321         (iwmmxt_textrc<mode>3): Likewise.
8322         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
8323         (wmmxt_pack): Likewise.
8324         (wmmxt_mult_c1): Likewise.
8325         (wmmxt_mult_c2): Likewise.
8326         (wmmxt_alu_c1): Likewise.
8327         (wmmxt_alu_c2): Likewise.
8328         (wmmxt_alu_c3): Likewise.
8329         (wmmxt_transfer_c1): Likewise.
8330         (wmmxt_transfer_c2): Likewise.
8331         (wmmxt_transfer_c3): Likewise.
8332         (marvell_f_iwmmxt_wstr): Likewise.
8333         (marvell_f_iwmmxt_wldr): Likewise.
8335 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
8337         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
8339 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
8341         Revert:
8342         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
8343         * lra-constraints.c (need_for_split_p): Check call used hard regs
8344         living through calls.
8346         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
8347         call used regs for call insn.
8349 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
8351         PR target/57736
8352         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
8353         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
8355 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
8357         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
8359 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
8361         * lra-constraints.c (need_for_split_p): Check call used hard regs
8362         living through calls.
8364 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
8366         PR target/57744
8367         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
8368         to tie with any other modes.  Eliminate Altivec vector mode tests,
8369         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
8370         code, to return 0 if testing MODE2 for a condition, if we've
8371         already tested MODE1 for the same condition.
8373 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8375         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
8376         layout.
8378 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8380         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
8381         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
8383 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8385         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
8386         Define.
8387         (aarch64_symbolic_constant_p): Remove.
8388         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
8389         static.  Fix line length and white space.
8390         (aarch64_symbolic_constant_p): Remove.
8391         * config/aarch64/predicates.md (aarch64_valid_symref):
8392         Use aarch64_classify_symbol_expression.
8394 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8396         * config/arm/constraints.md (Ts): New constraint.
8397         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
8398         16-bit encodings.
8399         (compare_scc): Use "Ts" constraint for operand 0.
8400         (ior_scc_scc): Likewise.
8401         (and_scc_scc): Likewise.
8402         (and_scc_scc_nodom): Likewise.
8403         (ior_scc_scc_cmp): Likewise for operand 7.
8404         (and_scc_scc_cmp): Likewise.
8405         * config/arm/thumb2.md (thumb2_movsi_insn):
8406         Add alternatives for 16-bit encodings.
8407         (thumb2_movhi_insn): Likewise.
8408         (thumb2_movsicc_insn): Likewise.
8409         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
8410         (thumb2_negscc): Use "Ts" constraint.
8411         Move mvn instruction outside cond_exec block.
8412         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
8413         for 16-bit encodings.
8415 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8417         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
8418         encoding.
8419         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
8420         (mulsi3subsi): Likewise.
8421         (mulsidi3adddi): Likewise.
8422         (mulsidi3_v6): Likewise.
8423         (umulsidi3_v6): Likewise.
8424         (umulsidi3adddi_v6): Likewise.
8425         (smulsi3_highpart_v6): Likewise.
8426         (umulsi3_highpart_v6): Likewise.
8427         (mulhisi3tb): Likewise.
8428         (mulhisi3bt): Likewise.
8429         (mulhisi3tt): Likewise.
8430         (maddhisi4): Likewise.
8431         (maddhisi4tb): Likewise.
8432         (maddhisi4tt): Likewise.
8433         (maddhidi4): Likewise.
8434         (maddhidi4tb): Likewise.
8435         (maddhidi4tt): Likewise.
8436         (zeroextractsi_compare0_scratch): Likewise.
8437         (insv_zero): Likewise.
8438         (insv_t2): Likewise.
8439         (anddi_notzesidi_di): Likewise.
8440         (anddi_notsesidi_di): Likewise.
8441         (andsi_notsi_si): Likewise.
8442         (iordi_zesidi_di): Likewise.
8443         (xordi_zesidi_di): Likewise.
8444         (andsi_iorsi3_notsi): Likewise.
8445         (smax_0): Likewise.
8446         (smax_m1): Likewise.
8447         (smin_0): Likewise.
8448         (not_shiftsi): Likewise.
8449         (unaligned_loadsi): Likewise.
8450         (unaligned_loadhis): Likewise.
8451         (unaligned_loadhiu): Likewise.
8452         (unaligned_storesi): Likewise.
8453         (unaligned_storehi): Likewise.
8454         (extv_reg): Likewise.
8455         (extzv_t2): Likewise.
8456         (divsi3): Likewise.
8457         (udivsi3): Likewise.
8458         (arm_zero_extendhisi2addsi): Likewise.
8459         (arm_zero_extendqisi2addsi): Likewise.
8460         (compareqi_eq0): Likewise.
8461         (arm_extendhisi2_v6): Likewise.
8462         (arm_extendqisi2addsi): Likewise.
8463         (arm_movt): Likewise.
8464         (thumb2_ldrd): Likewise.
8465         (thumb2_ldrd_base): Likewise.
8466         (thumb2_ldrd_base_neg): Likewise.
8467         (thumb2_strd): Likewise.
8468         (thumb2_strd_base): Likewise.
8469         (thumb2_strd_base_neg): Likewise.
8470         (arm_negsi2): Add alternative for 16-bit encoding.
8471         (arm_one_cmplsi2): Likewise.
8473 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8475         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
8476         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
8477         (movdfcc): Likewise.
8478         * config/arm/vfp.md (*thumb2_movsf_vfp):
8479         Disable predication for arm_restrict_it.
8480         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
8481         (*thumb2_movdfcc_vfp): Likewise.
8482         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
8483         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
8484         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
8485         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
8486         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
8487         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
8488         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
8489         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
8490         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
8491         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
8492         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
8493         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
8494         Disable predication for arm_restrict_it.
8496 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
8498         * config/i386/bmiintrin.h (_bextr_u32): New.
8499         (_bextr_u64): Ditto.
8501 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
8503         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
8504         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
8505         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
8506         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
8507         * config/mips/n32-elf.h: ...this new file.
8509 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
8511         PR target/57224
8512         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
8513         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
8515 2013-06-27  Catherine Moore  <clm@codesourcery.com>
8517         * config/mips/mips-tables.opt: Regenerate.
8518         * config/mips/mips-cpus.def: Add m14ke and m14kec.
8519         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
8520         * doc/invoke.texi: Add -m14kc.
8522 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
8524         PR target/57623
8525         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
8526         constraints of operand 1 and 2.
8528         PR target/57623
8529         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
8530         to match RTL canonicalization.  Swap predicates and
8531         constraints of operand 1 and 2.
8533 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
8535         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
8536         Process OP_INOUT regs for splitting too.
8538 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
8540         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
8541         decl before the loop, initialize to NULL.
8542         (vectorizable_load): Initialize ptr_incr to NULL.
8544 2013-06-27  Martin Jambor  <mjambor@suse.cz>
8546         PR lto/57208
8547         * ipa-ref.h (ipa_maybe_record_reference): Declare.
8548         * ipa-ref.c (ipa_maybe_record_reference): New function.
8549         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
8550         * ipa-cp.c (create_specialized_node): Record potential references from
8551         aggvals.
8552         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
8554 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8556         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
8557         parameter 'mode' of type 'enum machine_mode mode'; change to pass
8558         'mode' to force_reg.
8559         (aarch64_add_offset): Update calls to aarch64_force_temporary.
8560         (aarch64_expand_mov_immediate): Likewise.
8562 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8564         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
8565         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
8567 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8569         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
8570         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
8571         (struct machine_function): Add tbegin_p.
8572         (s390_canonicalize_comparison): Fold CC mode compares to
8573         conditional jump if possible.
8574         (s390_emit_jump): Return the emitted jump.
8575         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
8576         Handle CCRAWmode compares.
8577         (s390_option_override): Default to -mhtm if available.
8578         (s390_reg_clobbered_rtx): Handle floating point regs as well.
8579         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
8580         FPRs instead of df_regs_ever_live_p.
8581         (s390_optimize_nonescaping_tx): New function.
8582         (s390_init_frame_layout): Extend clobbered_regs array to cover
8583         FPRs as well.
8584         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
8585         (s390_expand_tbegin): New function.
8586         (enum s390_builtin): New enum definition.
8587         (code_for_builtin): New array definition.
8588         (s390_init_builtins): New function.
8589         (s390_expand_builtin): New function.
8590         (TARGET_INIT_BUILTINS): Define.
8591         (TARGET_EXPAND_BUILTIN): Define.
8592         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
8593         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
8594         (s390_alc_comparison): Likewise.
8595         * config/s390/s390-modes.def: Add CCRAWmode.
8596         * config/s390/s390.h (processor_flags): Add PF_TX.
8597         (TARGET_CPU_HTM): Define macro.
8598         (TARGET_HTM): Define macro.
8599         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
8600         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
8601         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
8602         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
8603         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
8604         (TBEGIN_MASK, TBEGINC_MASK): New constants.
8605         ("*cc_to_int"): Move up.
8606         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
8607         constants other than 0.
8608         ("*ccraw_to_int"): New insn and splitter definition.
8609         ("tbegin", "tbegin_nofloat", "tbegin_retry")
8610         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
8611         ("tx_assist"): New expander.
8612         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
8613         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
8614         * config/s390/s390.opt: Add -mhtm option.
8615         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
8616         * config/s390/htmxlintrin.h: New file.
8617         * config/s390/htmintrin.h: New file.
8618         * config/s390/s390intrin.h: New file.
8619         * doc/extend.texi: Document htm builtins.
8620         * config.gcc: Add the new header files to extra_headers.
8622 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
8624         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
8625         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
8627 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
8628             Pat Haugen  <pthaugen@us.ibm.com>
8629             Peter Bergner  <bergner@vnet.ibm.com>
8631         * config/rs6000/power8.md: New.
8632         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
8633         setting for power8 entry.
8634         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
8635         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
8636         test for Power4/Power5 only.
8637         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
8638         support.
8639         (force_new_group): Adjust comment.
8640         * config/rs6000/rs6000.md: Include power8.md.
8642 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
8644         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
8645         * config/arm/arm-protos.h (arm_max_conditional_execute): New
8646         declaration.
8647         (tune_params): Update comment.
8648         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
8649         (arm_max_conditional_execute): New function.
8650         (thumb2_final_prescan_insn): Use max_insn_skipped and
8651         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
8653 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
8655         PR tree-optimization/57705
8656         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
8657         SSA_NAME step, provided that it is not defined inside the loop.
8658         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
8659         (get_initial_def_for_induction): Handle SSA_NAME IV step.
8661 2013-06-25  Martin Jambor  <mjambor@suse.cz>
8663         PR middle-end/57670
8664         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
8665         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
8666         calls in the dump.
8667         (ipa_note_param_call): Initialize member_ptr flag.
8668         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
8669         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
8670         (ipa_write_indirect_edge_info): Stream member_ptr flag.
8671         (ipa_read_indirect_edge_info): Likewise.
8673 2013-06-25  Richard Biener  <rguenther@suse.de>
8675         PR middle-end/56977
8676         * passes.c (init_optimization_passes): Move pass_fold_builtins
8677         and pass_dce earlier with -Og.
8679 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
8681         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
8682         <BIT_FIELD_REF>: Remove trailing TAB.
8683         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
8684         remove blank line.
8686 2013-06-24  Martin Jambor  <mjambor@suse.cz>
8688         PR tree-optimization/57358
8689         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
8690         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
8691         (ipa_analyze_params_uses): Generate pessimistic info when true.
8693 2013-06-24  Martin Jambor  <mjambor@suse.cz>
8695         PR tree-optimization/57539
8696         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
8697         global.inlined_to of the new node to it.  All callers changed.
8698         * ipa-inline-transform.c (clone_inlined_nodes): New variable
8699         inlining_into, pass it to cgraph_clone_node.
8700         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
8701         ipa_free_edge_args_substructures.
8702         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
8703         rdesc linked list.  Do not assert rdesc edges have inlined caller.
8704         Assert we have found an rdesc in the rdesc list.
8706 2013-06-24  Richard Biener  <rguenther@suse.de>
8708         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
8709         (pointer_set_lookup): Declare.
8710         (class pointer_map): New template class implementing a
8711         generic pointer to T map.
8712         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
8713         pointer_map<T>::contains, pointer_map<T>::insert,
8714         pointer_map<T>::traverse): New functions.
8715         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
8716         (pointer_set_lookup): New function.
8717         (pointer_set_contains): Use pointer_set_lookup.
8718         (pointer_set_insert): Likewise.
8719         (insert_aux): Remove.
8720         (struct pointer_map_t): Embed a pointer_set_t.
8721         (pointer_map_create): Adjust.
8722         (pointer_map_destroy): Likewise.
8723         (pointer_map_contains): Likewise.
8724         (pointer_map_insert): Likewise.
8725         (pointer_map_traverse): Likewise.
8726         * tree-streamer.h (struct streamer_tree_cache_d): Use a
8727         pointer_map<unsigned> instead of a pointer_map_t.
8728         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
8729         (streamer_tree_cache_lookup): Likewise.
8730         (streamer_tree_cache_create): Likewise.
8731         (streamer_tree_cache_delete): Likewise.
8732         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
8733         pointer_map<unsigned> instead of a pointer_map_t.
8734         (lto_init_tree_ref_encoder): Adjust.
8735         (lto_destroy_tree_ref_encoder): Likewise.
8736         * lto-section-out.c (lto_output_decl_index): Likewise.
8737         (lto_record_function_out_decl_state): Likewise.
8738         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
8740 2013-06-24  Richard Biener  <rguenther@suse.de>
8742         PR tree-optimization/57488
8743         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
8745 2013-06-24  Alan Modra  <amodra@gmail.com>
8747         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
8748         (gen_easy_altivec_constant): Likewise.
8749         * config/rs6000/predicates.md (easy_vector_constant_add_self,
8750         easy_vector_constant_msb): Likewise.
8752 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
8754         PR target/57688
8755         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
8756         add missing return true.
8758 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
8760         PR target/52483
8761         * config/sh/predicates.md (general_extend_operand): Invoke
8762         general_movsrc_operand for memory operands.
8763         (general_movsrc_operand): Allow reg+reg addressing, do not use
8764         general_operand for memory operands.
8766 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
8768         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
8769         when current target options does not apply.
8770         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
8771         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
8772         * config/i386/bmiintrin.h: Pass appropriate target
8773         attributes to header.
8774         * config/i386/mmintrin.h: Ditto.
8775         * config/i386/nmmintrin.h: Ditto.
8776         * config/i386/avx2intrin.h: Ditto.
8777         * config/i386/fxsrintrin.h: Ditto.
8778         * config/i386/tbmintrin.h: Ditto.
8779         * config/i386/xsaveintrin.h: Ditto.
8780         * config/i386/f16cintrin.h: Ditto.
8781         * config/i386/xtestintrin.h: Ditto.
8782         * config/i386/xsaveoptintrin.h: Ditto.
8783         * config/i386/bmi2intrin.h: Ditto.
8784         * config/i386/lzcntintrin.h: Ditto.
8785         * config/i386/smmintrin.h: Ditto.
8786         * config/i386/wmmintrin.h: Ditto.
8787         * config/i386/x86intrin.h: Remove all header include guards.
8788         * config/i386/prfchwintrin.h: Ditto.
8789         * config/i386/pmmintrin.h: Ditto.
8790         * config/i386/tmmintrin.h: Ditto.
8791         * config/i386/xmmintrin.h: Ditto.
8792         * config/i386/popcntintrin.h: Ditto.
8793         * config/i386/rdseedintrin.h: Ditto.
8794         * config/i386/ammintrin.h: Ditto.
8795         * config/i386/emmintrin.h: Ditto.
8796         * config/i386/immintrin.h: Remove all header include guards.
8797         * config/i386/fma4intrin.h: Ditto.
8798         * config/i386/lwpintrin.h: Ditto.
8799         * config/i386/xopintrin.h: Ditto.
8800         * config/i386/ia32intrin.h: Ditto.
8801         * config/i386/avxintrin.h: Ditto.
8802         * config/i386/rtmintrin.h: Ditto.
8803         * config/i386/fmaintrin.h: Ditto.
8804         * config/i386/mm3dnow.h: Ditto.
8806 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
8808         * common/config/i386/i386-common.c: Handle LZCNT.
8810 2013-06-22  Andi Kleen  <ak@linux.intel.com>
8812         * doc/extend.texi: Use __atomic_store_n instead of
8813         __atomic_store in HLE example.
8815 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
8817         * config/sh/sh.c: Remove <cstdlib> workaround.
8819 2013-06-21  Andi Kleen  <ak@linux.intel.com>
8821         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
8823 2013-06-21  Andi Kleen  <ak@linux.intel.com>
8825         * doc/extend.texi: Document that __atomic_clear and
8826           __atomic_test_and_set should only be used with bool.
8828 2013-06-20  Jan Hubicka  <jh@suse.cz>
8830         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
8831         types_same_for_odr.
8832         * tree.c (decls_same_for_odr): New function.
8833         (same_for_edr): New function.
8834         (types_same_for_odr): New function.
8835         (get_binfo_at_offset): Use it.
8836         * tree.h (types_same_for_odr): Declare.
8838 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
8839             Jason Merrill  <jason@redhat.com>
8841         * system.h: Include <cstdlib> as well as <stdlib.h>.
8843 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
8845         PR target/57655
8846         * config/i386/i386.c (construct_container): Report error if
8847         long double is used with disabled x87 float returns.
8849 2013-06-20  Jan Hubicka  <jh@suse.cz>
8851         * lto-cgraph.c (input_symtab): Do not set cgraph state.
8853 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
8855         PR rtl-optimization/57425
8856         PR rtl-optimization/57569
8857         * alias.c (write_dependence_p): Remove parameters mem_mode and
8858         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
8859         Changed all callers.
8860         (canon_anti_dependence): Get comments and semantics in sync.
8861         Add parameter mem_canonicalized.  Changed all callers.
8862         * rtl.h (canon_anti_dependence): Update prototype.
8864 2013-06-20  Richard Biener  <rguenther@suse.de>
8866         * data-streamer-in.c (streamer_read_uhwi): Optimize single
8867         byte case, inline streamer_read_uchar and defer section
8868         overrun check.
8870 2013-06-20  Richard Biener  <rguenther@suse.de>
8872         PR tree-optimization/57584
8873         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
8874         SSA names into the expanded expression that take part in
8875         abnormal coalescing.
8877 2013-06-19  Sharad Singhai  <singhai@google.com>
8879         * gcov.c (print_usage): Handle new option.
8880         (process_args): Ditto.
8881         (get_gcov_intermediate_filename): New function.
8882         (output_intermediate_file): New function.
8883         (output_gcov_file): New function
8884         (generate_results): Handle new option.
8885         (release_function): Relase demangled name.
8886         (read_graph_file): Handle demangled name.
8887         (output_lines): Ditto.
8888         * doc/gcov.texi: Document gcov intermediate format.
8890 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
8892         PR bootstrap/57604
8893         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
8894         (lra_emit_add): Use the functions.  Add comment about Y as an
8895         address segment.
8897 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
8899         PR driver/57652
8900         * collect2.c (collect_atexit): New.
8901         (collect_exit): Delete.
8902         (main): Register collect_atexit with atexit.
8903         (collect_wait): Change collect_exit to exit.
8904         (do_wait): Same.
8905         * collect2.h (collect_exit): Delete.
8906         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
8908 2013-06-19  Wei Mi  <wmi@google.com>
8910         PR rtl-optimization/57518
8911         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
8912         if regno is used in paradoxical subreg.
8913         (update_equiv_regs): Check pdx_subregs[regno] before
8914         set a reg to be equivalent with a mem.
8916 2013-06-19  Matthias Klose  <doko@ubuntu.com>
8918         PR driver/57651
8919         * file-find.h (find_a_file): Add a mode parameter.
8920         * file-find.c (find_a_file): Likewise.
8921         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
8922         with X_OK for the executables.
8923         * collect2.c (main): Call find_a_file with X_OK.
8925 2013-06-19  Steve Ellcey  <sellcey@mips.com>
8927         PR target/56942
8928         * config/mips/mips.md (casesi_internal_mips16_<mode>):
8929         Use NEXT_INSN instead of next_real_insn.
8931 2013-06-19  Jan Hubicka  <jh@suse.cz>
8933         * cgraph.h (const_value_known_p): Replace by ...
8934         (ctor_for_folding): .. this one.
8935         * cgraphunit.c (process_function_and_variable_attributes): Use it.
8936         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
8937         * expr.c (expand_expr_real_1): Likewise.
8938         (string_constant): Likewise.
8939         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
8940         * ipa.c (process_references): Likewise.
8941         (symtab_remove_unreachable_nodes): Likewise.
8942         * ipa-inline-analysis.c (param_change_prob): Likewise.
8943         * gimple-fold.c (canonicalize_constructor_val): Likewise.
8944         (get_base_constructor): Likwise.
8945         * varpool.c (varpool_remove_node): Likewise.
8946         (varpool_remove_initializer): LIkewise.
8947         (dump_varpool_node): LIkwise.
8948         (const_value_known_p): Rewrite to ...
8949         (ctor_for_folding): ... this one.
8951 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
8953         PR driver/57651
8954         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
8955         PERSONALITY in $PATH derived prefixes.
8957 2013-06-19  Jeff Law  <law@redhat.com>
8959         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
8960         in comment.
8962         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
8963         (simplify_bitwise_binary): Use it to simpify certain binary ops on
8964         booleans.
8966 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
8968         * config/arm/vfp.md: Move VFP instruction classification documentation
8969         to ...
8970         * config/arm/arm.md: ... here.  Update instruction classification
8971         documentation.
8973 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
8975         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
8976         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
8977         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
8979 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
8981         PR target/57609
8982         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
8983         with NEXT_INSN.  Use tablejump_p to check for jump table data
8984         insns.
8986 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
8988         PR c++/56544
8989         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
8990         that now in C++ the value is correct per the C++ standards.
8992 2013-06-19  Richard Biener  <rguenther@suse.de>
8994         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
8995         for global context.
8997 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8999         Revert:
9000         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9002         PR target/57609
9003         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
9004         with next_active_insn.
9006 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
9008         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
9009         functions are inlined during failures to flag an error.
9010         * tree-inline.c (expand_call_inline): Allow the error to be flagged
9011         in early inline pass.
9013 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
9015         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
9016         in comments.
9018 2013-06-18  Julian Brown  <julian@codesourcery.com>
9020         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
9021         Permit virtual register pre-reload if !strict.
9022         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
9023         change.
9024         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
9025         prototype.
9026         * config/arm/neon.md (movmisalign<mode>): Use
9027         neon_perm_struct_or_reg_operand instead of
9028         neon_struct_or_register_operand.
9029         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
9030         neon_permissive_struct_operand instead of neon_struct_operand.
9031         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
9032         neon_vector_mem_operand.
9033         * config/arm/predicates.md (neon_struct_operand): Adjust call to
9034         neon_vector_mem_operand.
9035         (neon_permissive_struct_operand): New.
9036         (neon_struct_or_register_operand): Rename to...
9037         (neon_perm_struct_or_reg_operand): This. Adjust call to
9038         neon_vector_mem_operand.
9040 2013-06-18  Richard Biener  <rguenther@suse.de>
9042         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
9043         * lto-streamer.h: Include pointer-set.h.
9044         (struct lto_decl_slot): Remove.
9045         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
9046         Remove next_index entry.
9047         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
9048         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
9049         (lto_init_tree_ref_encoder): Adjust.
9050         (lto_destroy_tree_ref_encoder): Likewise.
9051         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
9052         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
9053         (lto_output_decl_index): Adjust.
9054         (lto_new_out_decl_state): Likewise.
9055         (lto_record_function_out_decl_state): Likewise.
9056         * lto-streamer-out.c (copy_function): Likewise.
9058 2013-06-18  Richard Biener  <rguenther@suse.de>
9060         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
9061         * cgraphunit.c: Include cfgloop.h.
9062         (init_lowered_empty_function): Initialize the loop tree.
9063         (assemble_thunk): Insert new BBs into loops.
9065 2013-06-18  Richard Biener  <rguenther@suse.de>
9067         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
9068         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
9069         the map from cache entry to cache index optional.
9070         (streamer_tree_cache_replace_tree): Adjust accordingly.
9071         (streamer_tree_cache_append): Likewise.
9072         (streamer_tree_cache_delete): Likewise.
9073         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
9074         streamer cache map from cache entry to cache index.
9075         * lto-streamer-out.c (create_output_block): Adjust.
9077 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
9079         * config/arm/arm.md (attribute "insn"): Move multiplication and
9080         division attributes to...
9081         (attribute "type"): ... here.  Remove mult.
9082         (attribute "mul32"): New attribute.
9083         (attribute "mul64"): Add umaal.
9084         (*arm_mulsi3): Update attributes.
9085         (*arm_mulsi3_v6): Likewise.
9086         (*thumb_mulsi3): Likewise.
9087         (*thumb_mulsi3_v6): Likewise.
9088         (*mulsi3_compare0): Likewise.
9089         (*mulsi3_compare0_v6): Likewise.
9090         (*mulsi_compare0_scratch): Likewise.
9091         (*mulsi_compare0_scratch_v6): Likewise.
9092         (*mulsi3addsi): Likewise.
9093         (*mulsi3addsi_v6): Likewise.
9094         (*mulsi3addsi_compare0): Likewise.
9095         (*mulsi3addsi_compare0_v6): Likewise.
9096         (*mulsi3addsi_compare0_scratch): Likewise.
9097         (*mulsi3addsi_compare0_scratch_v6): Likewise.
9098         (*mulsi3subsi): Likewise.
9099         (*mulsidi3adddi): Likewise.
9100         (*mulsi3addsi_v6): Likewise.
9101         (*mulsidi3adddi_v6): Likewise.
9102         (*mulsidi3_nov6): Likewise.
9103         (*mulsidi3_v6): Likewise.
9104         (*umulsidi3_nov6): Likewise.
9105         (*umulsidi3_v6): Likewise.
9106         (*umulsidi3adddi): Likewise.
9107         (*umulsidi3adddi_v6): Likewise.
9108         (*smulsi3_highpart_nov6): Likewise.
9109         (*smulsi3_highpart_v6): Likewise.
9110         (*umulsi3_highpart_nov6): Likewise.
9111         (*umulsi3_highpart_v6): Likewise.
9112         (mulhisi3): Likewise.
9113         (*mulhisi3tb): Likewise.
9114         (*mulhisi3bt): Likewise.
9115         (*mulhisi3tt): Likewise.
9116         (maddhisi4): Likewise.
9117         (*maddhisi4tb): Likewise.
9118         (*maddhisi4tt): Likewise.
9119         (maddhidi4): Likewise.
9120         (*maddhidi4tb): Likewise.
9121         (*maddhidi4tt): Likewise.
9122         (divsi3): Likewise.
9123         (udivsi3): Likewise.
9124         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
9125         (thumb2_mulsi_short_compare0): Likewise.
9126         (thumb2_mulsi_short_compare0_scratch): Likewise.
9127         * config/arm/arm1020e.md (1020mult1): Update attribute change.
9128         (1020mult2): Likewise.
9129         (1020mult3): Likewise.
9130         (1020mult4): Likewise.
9131         (1020mult5): Likewise.
9132         (1020mult6): Likewise.
9133         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
9134         change.
9135         (cortex_a15_mult64): Likewise.
9136         (cortex_a15_sdiv): Likewise.
9137         (cortex_a15_udiv): Likewise.
9138         * config/arm/arm1026ejs.md (mult1): Update attribute change.
9139         (mult2): Likewise.
9140         (mult3): Likewise.
9141         (mult4): Likewise.
9142         (mult5): Likewise.
9143         (mult6): Likewise.
9144         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
9145         (pj4_ir_div): Likewise.
9146         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
9147         (11_mult2): Likewise.
9148         (11_mult3): Likewise.
9149         (11_mult4): Likewise.
9150         (11_mult5): Likewise.
9151         (11_mult6): Likewise.
9152         (11_mult7): Likewise.
9153         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
9154         (cortex_a8_mla): Likewise.
9155         (cortex_a8_mull): Likewise.
9156         (cortex_a8_smulwy): Likewise.
9157         (cortex_a8_smlald): Likewise.
9158         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
9159         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
9160         (cortex_r4_mul_3): Likewise.
9161         (cortex_r4_mla_4): Likewise.
9162         (cortex_r4_mla_3): Likewise.
9163         (cortex_r4_smlald): Likewise.
9164         (cortex_r4_mull): Likewise.
9165         (cortex_r4_sdiv): Likewise.
9166         (cortex_r4_udiv): Likewise.
9167         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
9168         (cortex_a7_idiv): Likewise.
9169         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
9170         (9_mult2): Likewise.
9171         (9_mult3): Likewise.
9172         (9_mult4): Likewise.
9173         (9_mult5): Likewise.
9174         (9_mult6): Likewise.
9175         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
9176         (cortex_a53_sdiv): Likewise.
9177         (cortex_a53_udiv): Likewise.
9178         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
9179         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
9180         (mp626_mult2): Likewise.
9181         (mp626_mult3): Likewise.
9182         (mp626_mult4): Likewise.
9183         * config/arm/fa526.md (526_mult1): Update attribute change.
9184         (526_mult2): Likewise.
9185         * config/arm/arm-generic.md (mult): Update attribute change.
9186         (mult_ldsched_strongarm): Likewise.
9187         (mult_ldsched): Likewise.
9188         (multi_cycle): Likewise.
9189         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
9190         * config/arm/fa606te.md (606te_mult1): Update attribute change.
9191         (606te_mult2): Likewise.
9192         (606te_mult3): Likewise.
9193         (606te_mult4): Likewise.
9194         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
9195         (cortex_a9_mac16): Likewise.
9196         (cortex_a9_multiply): Likewise.
9197         (cortex_a9_mac): Likewise.
9198         (cortex_a9_multiply_long): Likewise.
9199         * config/arm/fa626te.md (626te_mult1): Update attribute change.
9200         (626te_mult2): Likewise.
9201         (626te_mult3): Likewise.
9202         (626te_mult4): Likewise.
9204 2013-06-18  Richard Biener  <rguenther@suse.de>
9206         PR lto/57334
9207         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
9209 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9211         PR target/57609
9212         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
9213         with next_active_insn.
9215 2013-06-18  Alan Modra  <amodra@gmail.com>
9217         * config/rs6000/rs6000.h (enum data_align): New.
9218         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
9219         (DATA_ABI_ALIGNMENT): Define.
9220         (CONSTANT_ALIGNMENT): Correct comment.
9221         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
9222         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
9224 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
9226         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
9227         ATTRIBUTE_UNUSED marking.
9229 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
9231         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
9232         alternative and update.
9233         (aarch64_dup_lanedi): Delete.
9234         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
9235         * config/aarch64/aarch64-simd-builtins.def: Update.
9237 2013-06-17  Richard Biener  <rguenther@suse.de>
9239         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
9240         (lto_input_scc): Declare.
9241         (lto_input_tree_1): Likewise.
9242         (struct lto_stats_d): Add num_tree_bodies_output and
9243         num_pickle_refs_output.
9244         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
9245         (lto_read_tree_1): Split out from ...
9246         (lto_read_tree): ... this.
9247         (lto_input_scc): New function.
9248         (lto_input_tree_1): Split out from ...
9249         (lto_input_tree): ... this.  Handle LTO_tree_scc.
9250         (lto_data_in_create): Create the streamer cache without hashes.
9251         * lto-streamer-out.c (create_output_block): Create the streamer
9252         cache with hashes when not doing WPA.
9253         (lto_write_tree_1): Split out from ...
9254         (lto_write_tree): ... this.
9255         (get_symbol_initial_value): New function.
9256         (lto_output_tree_1): Split out from ...
9257         (lto_output_tree): ... this.  Write trees as series of SCCs
9258         using a DFS walk via DFS_write_tree.
9259         (struct sccs, struct scc_entry): New types.
9260         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
9261         (DFS_write_tree_body): New function.
9262         (DFS_write_tree): Likewise.
9263         (hash_tree): Likewise.
9264         (scc_entry_compare): Likewise.
9265         (hash_scc): Likewise.
9266         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
9267         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
9268         TREE_CHAIN as regular reference.
9269         (streamer_read_integer_cst): Remove.
9270         (streamer_get_pickled_tree): Adjust.
9271         * tree-streamer-out.c (streamer_write_chain): Disable streaming
9272         of DECL_EXTERNALs in BLOCK_VARS for now.
9273         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
9274         reference.
9275         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
9276         Add hash value argument and record that if hashes are recorded
9277         in the cache.
9278         (streamer_tree_cache_insert_1): Adjust.
9279         (streamer_tree_cache_insert): Likewise.
9280         (streamer_tree_cache_insert_at): Rename to ...
9281         (streamer_tree_cache_replace_tree): ... this and adjust.
9282         (streamer_tree_cache_append): Adjust.
9283         (record_common_node): Likewise.
9284         (streamer_tree_cache_create): Add argument whether to
9285         record hash values together with trees.
9286         (streamer_tree_cache_delete): Adjust.
9287         * tree-streamer.h (struct streamer_tree_cache_d): Add
9288         vector of hashes.
9289         (streamer_read_integer_cst): Remove.
9290         (streamer_tree_cache_insert): Adjust.
9291         (streamer_tree_cache_append): Likewise.
9292         (streamer_tree_cache_insert_at): Rename to ...
9293         (streamer_tree_cache_replace_tree): ... this and adjust.
9294         (streamer_tree_cache_create): Add argument whether to record hashes.
9295         (streamer_tree_cache_get): Rename to ...
9296         (streamer_tree_cache_get_tree): ... this.
9297         (streamer_tree_cache_get_hash): New function.
9298         * tree.c (cache_integer_cst): New function.
9299         * tree.h (cache_integer_cst): Declare.
9300         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
9301         * lto-symtab.c (lto_varpool_replace_node): Only release
9302         DECL_INITIAL of non-prevailing decls.
9303         * varpool.c (varpool_remove_initializer): Do not release
9304         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
9306 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
9308         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
9309         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
9310         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
9311         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
9312         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
9313         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
9314         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
9315         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
9316         instead of TARGET_64BIT.
9317         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
9318         Require ISA_HAS_<D>DIV.
9320 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
9322         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
9323         (mips*-*-linux*): Move default with_llsc setting to where other
9324         defaults are set.
9325         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
9326         with_arch block.
9327         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
9328         Likewise.  Remove default with_tune setting.  Move default float
9329         setting to its own block.  Handle with_llsc in the same block as above.
9331 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
9333         PR rtl-optimization/57425
9334         PR rtl-optimization/57569
9335         * alias.c (write_dependence_p): Add new parameters mem_mode,
9336         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
9337         Changed all callers.
9338         (canon_anti_dependence): New function.
9339         * cse.c (check_dependence): Use canon_anti_dependence.
9340         * cselib.c (cselib_invalidate_mem): Likewise.
9341         * rtl.h (canon_anti_dependence): Declare.
9343 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
9345         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
9346         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
9347         ".set mips3" for 64-bit targets.
9349 2013-06-15  Dehao Chen  <dehao@google.com>
9351         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
9352         * gimple-low.c (gimple_check_call_matching_types): Likewise.
9353         (gimple_check_call_args): Likewise.
9354         * value-prof.c (check_ic_target): Likewise.
9355         * ipa-inline.c (early_inliner): Likewise.
9356         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
9357         * cgraph.c (cgraph_create_edge_1): Likewise.
9358         (cgraph_make_edge_direct): Likewise.
9360 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
9362         PR target/57615
9363         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
9364         rs6000_output_move_128bit to handle emitting quad memory
9365         operations.  Set attribute length to 8 bytes.
9367 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
9369         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
9370         New pattern.
9371         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
9372         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
9373         (aarch64_<su>mlsl<mode>): Likewise.
9375 2013-06-14  Mike Stump  <mikestump@comcast.net>
9377         * Makefile.in (TARGET_H): Add insn-codes.h.
9379 2013-06-14  Alan Modra  <amodra@gmail.com>
9381         PR middle-end/57134
9382         PR middle-end/57586
9383         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
9384         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
9385         bitfield expansion when EXPAND_MEMORY.
9386         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
9388 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
9390         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
9391         test for clearing quad memory on 32-bit later.
9393 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
9395         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
9396         (fold_negate_expr): Likewise.
9397         (fold_real_zero_addition_p): Handle vectors.
9398         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
9400 2013-06-14  Alan Modra  <amodra@gmail.com>
9402         * varasm.c (force_const_mem): Revert 2013-06-07 change.
9404 2013-06-13  Jan Hubicka  <jh@suse.cz>
9406         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
9407         Local comdats are not externally visible.
9408         * symtab.c (dump_symtab_base): Dump externally visible.
9409         (verify_symtab_base): Verify back links in the symtab hash.
9411 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
9413         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
9414         CONVERT_EXPR as equal nodes.
9416 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
9418         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
9420 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
9422         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
9423         Generalize to complex and vector.
9424         * tree.c (build_all_ones_cst): New function.
9425         * tree.h (build_all_ones_cst): Declare it.
9427 2013-06-13  Alan Modra  <amodra@gmail.com>
9429         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
9430         * config/rs6000/rs6000.md (signbittf2): New insn.
9431         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
9432         (abstf2_internal, cmptf_internal2): Likewise.
9433         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
9435 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
9436             Pat Haugen  <pthaugen@us.ibm.com>
9437             Peter Bergner  <bergner@vnet.ibm.com>
9439         * config/rs6000/rs6000.c (emit_load_locked): Add support for
9440         power8 byte, half-word, and quad-word atomic instructions.
9441         (emit_store_conditional): Likewise.
9442         (rs6000_expand_atomic_compare_and_swap): Likewise.
9443         (rs6000_expand_atomic_op): Likewise.
9445         * config/rs6000/sync.md (larx): Add new modes for power8.
9446         (stcx): Likewise.
9447         (AINT): New mode iterator to include TImode as well as normal
9448         integer modes on power8.
9449         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
9450         that VSX registers are not considered.  Use AINT mode iterator
9451         instead of INT1 to allow inclusion of quad word atomic operations
9452         on power8.
9453         (load_locked<mode>): Likewise.
9454         (store_conditional<mode>): Likewise.
9455         (atomic_compare_and_swap<mode>): Likewise.
9456         (atomic_exchange<mode>): Likewise.
9457         (atomic_nand<mode>): Likewise.
9458         (atomic_fetch_<fetchop_name><mode>): Likewise.
9459         (atomic_nand_fetch<mode>): Likewise.
9460         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
9461         each type.
9462         (ATOMIC): On power8, add QImode, HImode modes.
9463         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
9464         modes that promote to SImode.
9465         (load_lockedti): Convert TImode arguments to PTImode, so that we
9466         get a guaranteed even/odd register pair.
9467         (load_lockedpti): Likewise.
9468         (store_conditionalti): Likewise.
9469         (store_conditionalpti): Likewise.
9471         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
9472         atomic load/store instructions.
9473         (HSI): Likewise.
9475 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
9477         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
9478         loads.
9479         (insn_count): New attribute, with most cases extracted from...
9480         (length): ...here.  Redefine most cases in terms of insn_count.
9481         (single_insn): Delete.
9482         (can_delay): Use insn_count to check for single instructions.
9483         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
9484         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
9485         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
9486         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
9487         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
9488         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
9489         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
9490         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
9491         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
9492         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
9493         rather than "length".
9494         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
9495         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
9496         Use "insn_count" rather than "length".
9497         * config/mips/mips-dsp.md
9498         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
9499         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
9500         length attributes.
9502 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
9504         PR tree-optimization/57361
9505         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
9507 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
9509         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
9510         to split.
9511         (aarch64_simd_combine<mode>): New instruction expansion.
9512         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
9513         function prototype.
9514         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
9515         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
9517 2013-06-12  Jan Hubicka  <jh@suse.cz>
9519         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
9520         decl has when in streaming stage.
9521         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
9522         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
9524 2013-06-12  Roland Stigge  <stigge@antcom.de>
9526         PR target/57578
9527         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
9529 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
9531         PR tree-optimization/57537
9532         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
9533         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
9535 2013-06-12  Richard Biener  <rguenther@suse.de>
9537         * data-streamer.h (streamer_write_char_stream): CSE
9538         obs->current_pointer.
9539         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
9540         streamer_write_char_stream manually and optimize the resulting loop.
9541         (streamer_write_hwi_stream): Likewise.
9543 2013-06-12  Jan Hubicka  <jh@suse.cz>
9545         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
9546         * cgraph.h (varpool_create_empty_node): Declare.
9547         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
9548         duplicated nodes.
9549         * symtab.c (symtab_unregister_node): Be lax about missin entries
9550         in node hash.
9551         (symtab_get_node): Update comment.
9552         * varpool.c (varpool_create_empty_node): Break out from ...
9553         (varpool_node_for_decl): ... here.
9554         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
9556 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
9558         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
9559         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
9560         part.  Use straight-line flow at the end.
9561         <COMPONENT_REF>: Remove superfluous else.
9562         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
9564 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
9566         PR target/56564
9567         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
9568         target hook even for !TREE_PUBLIC decls.  If no resolution info
9569         is available, return false for common and external decls.
9571 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
9573         * config/rl78/constraints.md (U): New constraint.
9574         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
9575         valloc attribute.
9577 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
9579         PR target/57589
9580         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
9581         to allow returning address to AT_PLATFORM name.
9583 2013-06-11  Jan Hubicka  <jh@suse.cz>
9585         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
9586         * cgraph.h (symtab_node_base): Add weakref flag.
9587         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
9588         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
9589         (output_weakrefs): Use weakref flag.
9590         * fold-const.c (simple_operand_p): Handle WEAK.
9591         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
9592         * ipa.c (varpool_externally_visible_p): Drop weakref.
9593         (function_and_variable_visibility): Update comment; fix weakref
9594         sanity checks; do not clear DECL_WEAK on them.
9595         * lto-cgraph.c (lto_output_node): update.
9596         (lto_output_varpool_node): Update.
9597         (input_overwrite_node): Update.
9598         (input_node): Update.
9599         (input_varpool_node): Update.
9600         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
9601         (lto_symtab_merge_symbols): Add sanity check.
9602         (lto_symtab_prevailing_decl): Do not special case weakrefs.
9603         * passes.c (rest_of_decl_compilation): Set static flag, too.
9604         * symtab.c (dump_symtab_base): Dump weakref.
9605         (verify_symtab_base): Sanity check weakrefs.
9606         (symtab_make_decl_local): Remove duplicated code.
9607         (symtab_alias_ultimate_target): Simplify.
9608         * varpool.c (varpool_create_variable_alias): Set weakref flag.
9610 2013-06-11  Tom de Vries  <tom@codesourcery.com>
9612         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
9613         sequence_vect == NULL.
9615 2013-06-11  DJ Delorie  <dj@redhat.com>
9617         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
9618         (rl78_unwind_word_mode): New.
9620 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
9622         * final.c (debug_prefix_maps): Make static.
9624 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
9626         * function.c (initial_trampoline): Remove stray copy.
9628 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
9630         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
9632 2013-06-11  Martin Jambor  <mjambor@suse.cz>
9634         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
9635         within bounds at the beginning of the function.
9637 2013-06-11  Alan Modra  <amodra@gmail.com>
9639         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
9640         reporting.
9641         (get_named_section): Don't NULL !DECL_P decl.
9643 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
9645         * doc/invoke.texi (core-avx2): Document.
9646         (slm): Likewise.
9647         (atom): Updated with MOVBE.
9649 2013-06-11  Richard Biener  <rguenther@suse.de>
9651         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
9653 2013-06-11  Anton Blanchard  <anton@samba.org>
9655         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
9656         correct shift value in little-endian mode.
9658 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
9660         PR target/56564
9661         * varasm.c (get_variable_align): Move #endif to the right place.
9663 2013-06-10  Cary Coutant  <ccoutant@google.com>
9665         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
9666         for hash so that hash table traversal order is deterministic.
9668 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
9669             Pat Haugen  <pthaugen@us.ibm.com>
9670             Peter Bergner  <bergner@vnet.ibm.com>
9672         * config/rs6000/vector.md (GPR move splitter): Do not split moves
9673         of vectors in GPRS if they are direct moves or quad word load or
9674         store moves.
9676         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
9677         declaration.
9678         (direct_move_p): Likewise.
9679         (quad_load_store_p): Likewise.
9681         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
9682         classes into bins based on the physical register type.
9683         (reg_class_to_reg_type): Likewise.
9684         (IS_STD_REG_TYPE): Likewise.
9685         (IS_FP_VECT_REG_TYPE): Likewise.
9686         (reload_fpr_gpr): Arrays to determine what insn to use if we can
9687         use direct move instructions.
9688         (reload_gpr_vsx): Likewise.
9689         (reload_vsx_gpr): Likewise.
9690         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
9691         information that is a simplification of register classes.  Also
9692         precalculate direct move reload helpers.
9693         (direct_move_p): New function to return true if the operation can
9694         be done as a direct move instruciton.
9695         (quad_load_store_p): New function to return true if the operation
9696         is a quad memory operation.
9697         (rs6000_legitimize_address): If quad memory, only allow register
9698         indirect for TImode addresses.
9699         (rs6000_legitimate_address_p): Likewise.
9700         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
9701         (rs6000_reload_register_type): Likewise.
9702         (register_to_reg_type): Return register type.
9703         (rs6000_secondary_reload_simple_move): New helper function for
9704         secondary reload and secondary memory needed to identify anything
9705         that is a simple move, and does not need reloading.
9706         (rs6000_secondary_reload_direct_move): New helper function for
9707         secondary reload to identify cases that can be done with several
9708         instructions via the direct move instructions.
9709         (rs6000_secondary_reload_move): New helper function for secondary
9710         reload to identify moves between register types that can be done.
9711         (rs6000_secondary_reload): Add support for quad memory operations
9712         and for direct move.
9713         (rs6000_secondary_memory_needed): Likewise.
9714         (rs6000_debug_secondary_memory_needed): Change argument names.
9715         (rs6000_output_move_128bit): New function to return the move to
9716         use for 128-bit moves, including knowing about the various
9717         limitations of quad memory operations.
9719         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
9720         memory operations.  call rs6000_output_move_128bit for the actual
9721         instruciton(s) to generate.
9722         (vsx_movti_64bit): Likewise.
9724         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
9725         (UNSPEC_P8V_MTVSRWZ): Likewise.
9726         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
9727         (UNSPEC_P8V_MTVSRD): Likewise.
9728         (UNSPEC_P8V_XXPERMDI): Likewise.
9729         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
9730         (UNSPEC_FUSION_GPR): Likewise.
9731         (FMOVE128_GPR): New iterator for direct move.
9732         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
9733         (f32_sv): Likewise.
9734         (f32_dm): Likewise.
9735         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
9736         loads and direct move instructions.
9737         (zero_extendsidi2_lfiwzx): Likewise.
9738         (extendsidi2_lfiwax): Likewise.
9739         (extendsidi2_nocell): Likewise.
9740         (floatsi<mode>2_lfiwax): Likewise.
9741         (lfiwax): Likewise.
9742         (floatunssi<mode>2_lfiwzx): Likewise.
9743         (lfiwzx): Likewise.
9744         (fix_trunc<mode>_stfiwx): Likewise.
9745         (fixuns_trunc<mode>_stfiwx): Likewise.
9746         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
9747         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
9748         (parity<mode>2_cmpb): Set length/type attr.
9749         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
9750         for 'mr.' to fast_compare.
9751         (bpermd_<mode>): Change type attr to popcnt.
9752         (p8_fmrgow_<mode>): New insns for power8 direct move support.
9753         (p8_mtvsrwz_1): Likewise.
9754         (p8_mtvsrwz_2): Likewise.
9755         (reload_fpr_from_gpr<mode>): Likewise.
9756         (p8_mtvsrd_1): Likewise.
9757         (p8_mtvsrd_2): Likewise.
9758         (p8_xxpermdi_<mode>): Likewise.
9759         (reload_vsx_from_gpr<mode>): Likewise.
9760         (reload_vsx_from_gprsf): Likewise.
9761         (p8_mfvsrd_3_<mode>): LIkewise.
9762         (reload_gpr_from_vsx<mode>): Likewise.
9763         (reload_gpr_from_vsxsf): Likewise.
9764         (p8_mfvsrd_4_disf): Likewise.
9765         (multi-word GPR splits): Do not split direct moves or quad memory
9766         operations.
9768 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
9770         * tree-into-ssa.c (interesting_blocks): Make static.
9772 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
9774         PR target/56564
9775         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
9776         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
9777         Use DATA_ABI_ALIGNMENT for that case instead if defined.
9778         (get_variable_align): New function.
9779         (get_variable_section, emit_bss, emit_common,
9780         assemble_variable_contents, place_block_symbol): Use
9781         get_variable_align instead of DECL_ALIGN.
9782         (assemble_noswitch_variable): Add align argument, use it
9783         instead of DECL_ALIGN.
9784         (assemble_variable): Adjust caller.  Use get_variable_align
9785         instead of DECL_ALIGN.
9786         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
9787         caller.
9788         (DATA_ABI_ALIGNMENT): Define.
9789         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
9790         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
9791         opt is false, only return the psABI mandated alignment increase.
9792         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
9793         (DATA_ABI_ALIGNMENT): ... this.
9794         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
9795         (DATA_ABI_ALIGNMENT): ... this.
9796         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
9797         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
9798         (DATA_ABI_ALIGNMENT): ... this.
9799         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
9800         * doc/tm.texi: Regenerated.
9802 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
9804         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
9805         cmp_code to construct REG_EQUAL note.
9807 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
9809         PR target/57568
9810         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
9811         that operands[2] doesn't overlap with operands[0].
9813 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
9814             Jan Hubicka  <jh@suse.cz>
9816         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
9817         hack to mark symbols as used.
9819 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
9821         PR rtl-optimization/57559
9822         * lra-constraints.c (process_alt_operands): Don't discourage
9823         memory with known offset for offsetable memory constraint.
9824         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
9826 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
9828         * varasm.c (struct oc_local_state): Reorder fields.
9829         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
9830         and adjust accordingly.
9831         (output_constructor): Reorder initialization code and adjust call to
9832         output_constructor_bitfield.
9834 2013-06-07  Jan Hubicka  <jh@suse.cz>
9836         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
9838 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
9840         * tree-object-size.c (unknown): Make const.
9842 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9844         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
9845         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
9846         for last alternative in the cpu_facility attribute.
9848 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9850         PR target/56315
9851         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
9852         (xordi3): Change operand 2 constraint to arm_xordi_operand.
9853         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
9854         * config/arm/constraints.md (Dg): New constraint.
9855         * config/arm/neon.md (xordi3_neon): Remove.
9856         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
9857         * config/arm/predicates.md (arm_xordi_operand): New predicate.
9859 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9861         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
9862         Clean up alternatives.
9864 2013-06-07  Alan Modra  <amodra@gmail.com>
9866         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
9867         va_list_gpr_size.
9869 2013-06-07  Alan Modra  <amodra@gmail.com>
9871         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
9873 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9875         * config/arm/constraints.md (Df): New constraint.
9876         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
9877         Correct length attribute for last two alternatives.
9879 2013-06-07  Alan Modra  <amodra@gmail.com>
9881         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
9882         override user -mfp-in-toc.
9883         (offsettable_ok_by_alignment): Consider just the current access
9884         rather than the whole object, unless BLKmode.  Handle
9885         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
9886         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
9887         for -mcmodel=medium.
9888         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
9889         override user -mfp-in-toc or -msum-in-toc.  Default to
9890         -mno-fp-in-toc for -mcmodel=medium.
9892 2013-06-06  DJ Delorie  <dj@redhat.com>
9894         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
9895         TARGET_VALID_POINTER_MODE.
9897 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
9898             Pat Haugen  <pthaugen@us.ibm.com>
9899             Peter Bergner  <bergner@vnet.ibm.com>
9901         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
9902         Document new power8 builtins.
9904         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
9905         condition code register, to allow 128-bit logical operations to be
9906         done in the VSX or GPR registers.
9907         (nor<mode>3): Use the canonical form for nor.
9908         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
9909         vclz*, and vpopcnt* vector instructions.
9910         (nand<mode>3): Likewise.
9911         (orc<mode>3): Likewise.
9912         (clz<mode>2): LIkewise.
9913         (popcount<mode>2): Likewise.
9915         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
9916         that only the GPRs are recognized.
9918         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9919         support for new power8 builtins.
9921         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
9922         builtin functions.
9923         (xscvdpspn): Likewise.
9924         (vclz): Likewise.
9925         (vclzb): Likewise.
9926         (vclzh): Likewise.
9927         (vclzw): Likewise.
9928         (vclzd): Likewise.
9929         (vpopcnt): Likewise.
9930         (vpopcntb): Likewise.
9931         (vpopcnth): Likewise.
9932         (vpopcntw): Likewise.
9933         (vpopcntd): Likewise.
9934         (vgbbd): Likewise.
9935         (vmrgew): Likewise.
9936         (vmrgow): Likewise.
9937         (eqv): Likewise.
9938         (eqv_v16qi3): Likewise.
9939         (eqv_v8hi3): Likewise.
9940         (eqv_v4si3): Likewise.
9941         (eqv_v2di3): Likewise.
9942         (eqv_v4sf3): Likewise.
9943         (eqv_v2df3): Likewise.
9944         (nand): Likewise.
9945         (nand_v16qi3): Likewise.
9946         (nand_v8hi3): Likewise.
9947         (nand_v4si3): Likewise.
9948         (nand_v2di3): Likewise.
9949         (nand_v4sf3): Likewise.
9950         (nand_v2df3): Likewise.
9951         (orc): Likewise.
9952         (orc_v16qi3): Likewise.
9953         (orc_v8hi3): Likewise.
9954         (orc_v4si3): Likewise.
9955         (orc_v2di3): Likewise.
9956         (orc_v4sf3): Likewise.
9957         (orc_v2df3): Likewise.
9959         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
9960         allow power8 quad mode in 64-bit.
9961         (rs6000_builtin_vectorized_function): Add support to vectorize
9962         ISA 2.07 count leading zeros, population count builtins.
9963         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
9964         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
9965         (builtin_function_type): Add vgbbd builtin function which takes an
9966         unsigned argument.
9967         (altivec_expand_vec_perm_const): Add support for new power8 merge
9968         instructions.
9970         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
9971         that does not include TImdoe for use with 32-bit.
9972         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
9973         instructions.
9974         (UNSPEC_VSX_CVDPSPN): Likewise.
9975         (vsx_xscvdpspn): Likewise.
9976         (vsx_xscvspdpn): Likewise.
9977         (vsx_xscvdpspn_scalar): Likewise.
9978         (vsx_xscvspdpn_directmove): Likewise.
9979         (vsx_and<mode>3): Split logical operations into 32-bit and
9980         64-bit. Add support to do logical operations on TImode as well as
9981         VSX vector types.  Allow logical operations to be done in either
9982         VSX registers or in general purpose registers in 64-bit mode.  Add
9983         splitters if GPRs were used. For AND, add clobber of CCmode to
9984         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
9985         encoding.
9986         (vsx_and<mode>3_32bit): Likewise.
9987         (vsx_and<mode>3_64bit): Likewise.
9988         (vsx_ior<mode>3): Likewise.
9989         (vsx_ior<mode>3_32bit): Likewise.
9990         (vsx_ior<mode>3_64bit): Likewise.
9991         (vsx_xor<mode>3): Likewise.
9992         (vsx_xor<mode>3_32bit): Likewise.
9993         (vsx_xor<mode>3_64bit): Likewise.
9994         (vsx_one_cmpl<mode>2): Likewise.
9995         (vsx_one_cmpl<mode>2_32bit): Likewise.
9996         (vsx_one_cmpl<mode>2_64bit): Likewise.
9997         (vsx_nor<mode>3): Likewise.
9998         (vsx_nor<mode>3_32bit): Likewise.
9999         (vsx_nor<mode>3_64bit): Likewise.
10000         (vsx_andc<mode>3): Likewise.
10001         (vsx_andc<mode>3_32bit): Likewise.
10002         (vsx_andc<mode>3_64bit): Likewise.
10003         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
10004         and xxlorc instructions.
10005         (vsx_eqv<mode>3_64bit): Likewise.
10006         (vsx_nand<mode>3_32bit): Likewise.
10007         (vsx_nand<mode>3_64bit): Likewise.
10008         (vsx_orc<mode>3_32bit): Likewise.
10009         (vsx_orc<mode>3_64bit): Likewise.
10011         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
10013         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
10014         instruction.
10015         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
10016         (p8_vmrgow): Likewise.
10017         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
10018         GPRs to be split under VSX.
10019         (p8v_clz<mode>2): Add power8 count leading zero support.
10020         (p8v_popcount<mode>2): Add power8 population count support.
10021         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
10022         support.
10024         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
10025         instruction.
10027         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
10028         builtin functions.
10029         (vec_nand): Likewise.
10030         (vec_vclz): Likewise.
10031         (vec_vclzb): Likewise.
10032         (vec_vclzd): Likewise.
10033         (vec_vclzh): Likewise.
10034         (vec_vclzw): Likewise.
10035         (vec_vgbbd): Likewise.
10036         (vec_vmrgew): Likewise.
10037         (vec_vmrgow): Likewise.
10038         (vec_vpopcnt): Likewise.
10039         (vec_vpopcntb): Likewise.
10040         (vec_vpopcntd): Likewise.
10041         (vec_vpopcnth): Likewise.
10042         (vec_vpopcntw): Likewise.
10044 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
10046         PR rtl-optimization/57468
10047         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
10048         spilled pseudos.
10050 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
10052         PR rtl-optimization/57459
10053         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
10054         type when setting live regs.
10056 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
10058         * config/s390/s390.opt (mlra): New option.
10059         * config/s390/s390.c (s390_decompose_address): Check displacement
10060         for all registers for LRA.
10061         (s390_secondary_reload): Don't used secondary reloads for LRA.
10062         (s390_lra_p): New function.
10063         (TARGET_LRA_P): Define.
10064         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
10065         of attribute cpu_facility to zarch for the last alternative.
10066         (*cmpmem_short): Ditto.
10068 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
10070         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
10071         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
10072         (arm_expand_prologue): Likewise.
10074 2013-06-06  Teresa Johnson  <tejohnson@google.com>
10076         PR c++/53743
10077         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
10078         as this is now done by redirect_edge_and_branch_force.
10079         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
10080         barriers, and fix interaction with splitting.
10081         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
10082         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
10083         reflect changes made in the routine.
10084         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
10085         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
10086         since this is called in cfglayout mode, and replace partition fixup
10087         with assert as that is now done by force_nonfallthru_and_redirect.
10088         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
10089         already be marked with region crossing note.
10090         (insert_section_boundary_note): Make non-static, gate on flag
10091         has_bb_partition, rewrite to also check for multiple partitions.
10092         (rest_of_handle_reorder_blocks): Remove call to
10093         insert_section_boundary_note, now done later during free_cfg.
10094         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
10095         * bb-reorder.h (insert_section_boundary_note): Declare.
10096         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
10097         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
10098         invoke insert_section_boundary_note.
10099         (try_redirect_by_replacing_jump): Remove unnecessary
10100         check for region crossing note.
10101         (fixup_partition_crossing): New function.
10102         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
10103         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
10104         in non-cfglayout mode.
10105         (force_nonfallthru_and_redirect): Fixup partition boundaries,
10106         remove old code that tried to do this. Emit barrier correctly
10107         when we are in cfglayout mode.
10108         (last_bb_in_partition): New function.
10109         (rtl_split_edge): Correctly fixup partition boundaries.
10110         (commit_one_edge_insertion): Remove old code that tried to
10111         fixup region crossing edge since this is now handled in
10112         split_block, and set up insertion point correctly since
10113         block may now end in a jump.
10114         (verify_hot_cold_block_grouping): Guard against checking when not in
10115         linearized RTL mode.
10116         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
10117         notes.
10118         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
10119         rtl_verify_flow_info, so not called in cfglayout mode.
10120         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
10121         (fixup_reorder_chain): Remove old code that attempted to fixup region
10122         crossing note as this is now handled in force_nonfallthru_and_redirect.
10123         (duplicate_insn_chain): Don't duplicate switch section notes.
10124         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
10125         note.
10126         * basic-block.h (emit_barrier_after_bb): Declare.
10128 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10130         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
10131         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
10132         arm_usatsihi): Adjust alternatives for arm_restrict_it.
10134 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10136         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
10137         where appropriate.
10138         * config/arm/ldmstm.md: Regenerate.
10140 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10142         * config/arm/sync.md (atomic_loaddi_1):
10143         Disable predication for arm_restrict_it.
10144         (arm_load_exclusive<mode>): Likewise.
10145         (arm_load_exclusivesi): Likewise.
10146         (arm_load_exclusivedi): Likewise.
10147         (arm_load_acquire_exclusive<mode>): Likewise.
10148         (arm_load_acquire_exclusivesi): Likewise.
10149         (arm_load_acquire_exclusivedi): Likewise.
10150         (arm_store_exclusive<mode>): Likewise.
10151         (arm_store_exclusive<mode>): Likewise.
10152         (arm_store_release_exclusivedi): Likewise.
10153         (arm_store_release_exclusive<mode>): Likewise.
10155 2013-06-06  Richard Biener  <rguenther@suse.de>
10157         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
10158         after LTO_null.
10159         (lto_tag_is_tree_code_p): Adjust.
10160         (lto_tag_is_gimple_code_p): Likewise.
10161         (lto_gimple_code_to_tag): Likewise.
10162         (lto_tag_to_gimple_code): Likewise.
10163         (lto_tree_code_to_tag): Likewise.
10164         (lto_tag_to_tree_code): Likewise.
10165         * data-streamer.h (streamer_write_hwi_in_range): Use
10166         uhwi streaming to stream the normalized range.
10167         (streamer_read_hwi_in_range): Likewise.
10169 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10171         * config/arm/arm.md (enabled_for_depr_it): New attribute.
10172         (predicable_short_it): Likewise.
10173         (predicated): Likewise.
10174         (enabled): Handle above.
10175         (define_cond_exec): Set predicated attribute to yes.
10177 2013-06-05  Mike Stump  <mikestump@comcast.net>
10179         * gdbinit.in (__FUNCTION__): Add.
10181 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
10183         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
10184         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
10186 2013-06-05  Jan Hubicka  <jh@suse.cz>
10188         * varasm.c (mark_decl_referenced): Revert the removal until targets
10189         are fixed.
10191 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
10193         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
10194         instead of mark_decl_referenced.
10196 2013-06-05  Jan Hubicka  <jh@suse.cz>
10198         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
10199         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
10200         and symtab_used_from_object_file_p.
10201         (cgraph_make_node_local_1): Clear forced_by_abi.
10202         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
10203         * cgraph.h (symtab_node_base): Add forced_by_abi.
10204         (decide_is_variable_needed): Remove.
10205         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
10206         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
10207         (decide_is_symbol_needed): ... this one; handle symbols in general;
10208         always analyze virtuals; honnor forced_by_abi.
10209         (cgraph_finalize_function): Update.
10210         (varpool_finalize_decl): Update.
10211         (symbol_defined_and_needed): Remove.
10212         (analyze_functions): Update.
10213         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
10214         output_refs, input_overwrite_node): Handle forced_by_abi.
10215         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
10216         (address_taken_from_non_vtable_p): ... this one.
10217         (comdat_can_be_unshared_p_1): New function.
10218         (cgraph_comdat_can_be_unshared_p): Rename to ...
10219         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
10220         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
10221         (function_and_variable_visibility): Clear forced_by_abi as needed.
10222         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
10223         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
10224         * symtab.c (dump_symtab_base): Dump forced_by_abi.
10225         * varpool.c (decide_is_variable_needed): Remove.
10227 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10229         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
10230         (arm_option_override): Override arm_restrict_it where appropriate.
10231         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
10232         * config/arm/arm.opt (mrestrict-it): New command-line option.
10233         * doc/invoke.texi: Document -mrestrict-it.
10235 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
10237         * tsan.c (tsan_atomic_table): Make const.
10239 2013-06-05  Richard Biener  <rguenther@suse.de>
10241         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
10242         index associated with the tree we are supposed to replace.
10243         * tree-streamer-out.c (pack_ts_base_value_fields): Output
10244         TREE_ASM_WRITTEN as zero for everything but SSA names.
10246 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
10248         * tree-ssa-structalias.c (call_stmt_vars): Make static.
10250 2013-06-04  Jan Hubicka  <jh@suse.cz>
10252         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
10253         (input_node, input_varpool_node): Handle correctly external same
10254         body aliases.
10255         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
10256         nodes at ltrans stage.
10258 2013-06-04  Jan Hubicka  <jh@suse.cz>
10260         * ipa-inline.c (update_caller_keys): Fix availability test.
10261         (update_callee_keys): Likewise.
10262         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
10263         to follow ELF standard.
10265 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
10267         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
10268         (mips64r5900el-*-elf*): New configurations.
10269         * config/mips/mips-cpus.def (r5900): New processor.
10270         * config/mips/mips-tables.opt: Regenerate.
10271         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
10272         (mips_issue_rate): Handle PROCESSOR_R5900.
10273         (mips_reorg_process_insns): Force reorder mode for the R5900.
10274         * config/mips/mips.h (TARGET_MIPS5900): Define.
10275         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
10276         TARGET_MIPS5900.
10277         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
10278         TARGET_MIPS5900.
10279         * config/mips/mips.md (processor): Add r5900.
10280         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
10282 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
10284         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
10285         into function to generate MOVI instruction.
10286         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
10287         (aarch64_preferred_simd_mode): Turn into wrapper.
10288         (aarch64_output_scalar_simd_mov_immediate): New function.
10289         * config/aarch64/aarch64-protos.h: Add prototype for above.
10291 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
10293         * config/aarch64/aarch64.c (simd_immediate_info): Remove
10294         element_char member.
10295         (sizetochar): Return signed char.
10296         (aarch64_simd_valid_immediate): Remove elchar and other
10297         unnecessary variables.
10298         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
10299         Calculate element_char as required.
10300         * config/aarch64/aarch64-protos.h: Update and move prototype
10301         for aarch64_output_simd_mov_immediate.
10302         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
10303         Update arguments.
10305 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
10307         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
10308         information completed by aarch64_simd_valid_immediate.
10309         (aarch64_legitimate_constant_p): Update arguments.
10310         (aarch64_simd_valid_immediate): Work with struct rather than many
10311         pointers.
10312         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
10313         (aarch64_simd_make_constant): Update arguments.
10314         (aarch64_output_simd_mov_immediate): Work with struct rather than
10315         many pointers.  Output immediate directly rather than as operand.
10316         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
10317         Update prototype.
10318         * config/aarch64/constraints.md (Dn): Update arguments.
10320 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
10322         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
10323         longer static.
10324         (aarch64_simd_immediate_valid_for_move): Remove.
10325         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
10326         (aarch64_simd_make_constant): Update call.
10327         (aarch64_output_simd_mov_immediate): Update call.
10328         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
10329         Add prototype.
10330         * config/aarch64/constraints.md (Dn): Update call.
10332 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
10334         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
10335         return type to bool for prototype.
10336         (aarch64_legitimate_constant_p): Check for true instead of not -1.
10337         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
10338         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
10340 2013-06-04  Catherine Moore  <clm@codesourcery.com>
10342         * config/mips/mips.opt (meva): New.
10343         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
10344         (ASM_SPEC): Handle -meva.
10345         * doc/invoke.texi (meva):  Document.
10347 2013-06-04  Alan Modra  <amodra@gmail.com>
10349         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
10350         constant output.
10352 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10354         * rtl.def: Add extra fourth optional field to define_cond_exec.
10355         * gensupport.c (process_one_cond_exec): Process attributes from
10356         define_cond_exec.
10357         * doc/md.texi: Document fourth field in define_cond_exec.
10359 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
10361         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
10362         out the processing order as in store_bit_field_1.
10364 2013-06-04  Jan Hubicka  <jh@suse.cz>
10366         PR middle-end/57500
10367         * cgraphunit.c (cgraph_process_same_body_aliases): Create
10368         non-VAR_DECL node if it does not exist yet.
10370 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
10372         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
10373         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
10374         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
10375         target_cpu_default setting.
10377 2013-06-03  Teresa Johnson  <tejohnson@google.com>
10379         * dumpfile.c (opt_info_switch_p): Change -fopt-info
10380         default to -fopt-info=optimized instead of all.
10381         * doc/invoke.texi: Ditto.
10382         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
10383         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
10384         (execute_vect_slp): Emit BB vectorization success under
10385         MSG_OPTIMIZED_LOCATIONS.
10386         * tree-vect-slp.c (vect_slp_transform_bb): Change
10387         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
10388         * tree-vect-loop.c (vect_transform_loop): Ditto.
10390 2013-06-03  Jason Merrill  <jason@redhat.com>
10392         PR c++/57415
10393         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10394         Use TARGET_EXPR for C++.
10396 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
10398         PR rtl-optimization/57268
10399         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
10400         if DEBUG_INSN_P (insn).
10402         Reapply
10403         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
10405         PR rtl-optimization/57268
10406         * sched-deps.c (sched_analyze_2): Flush dependence lists if
10407         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
10409 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
10411         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
10412         (ix86_avoid_lea_for_addr): Likewise.
10413         (exact_dependency_1): Likewise.
10414         (ix86_adjust_cost): Likewise.
10415         (swap_top_of_ready_list): Fix formatting and !reload_completed check
10416         removed.
10417         (do_reorder_for_imul): Fix typo, formatting and
10418         !reload_completed check removed.
10419         (ix86_sched_reorder): Fix typo and formatting.
10420         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
10421         list.
10423 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
10425         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
10427 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
10429         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
10430         <STRING_CST>: Likewise.
10431         <VECTOR_CST>: Likewise.
10433 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
10434             Mikael Morin  <mikael@gcc.gnu.org>
10436         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
10437         * config.in: Regenerated.
10438         * configure: Regenerated.
10440 2013-06-01  Jan Hubicka  <jh@suse.cz>
10442         PR middle-end/57366
10443         * cgraphunit.c (compile): When weakref is not supported,
10444         set up transparent aliases before final output pass.
10445         * varasm.c (assemble_alias): Do not try to do it here.
10447 2013-06-01  Jan Hubicka  <jh@suse.cz>
10449         PR middle-end/57467
10450         * passes.c (for_per_function): Skip unanalyzed functions.
10452 2013-06-01  Jan Hubicka  <jh@suse.cz>
10454         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
10455         (lto_symtab_merge_symbols_1): ... this one.
10456         (lto_symtab_merge_cgraph_nodes): Rename to ...
10457         (lto_symtab_merge_symbols): ... this one; simplify.
10458         * cgraph.c (same_body_aliases_done): Rename to ...
10459         (cpp_implicit_aliases_done): ... this one.
10460         (cgraph_create_function_alias): Update.
10461         (cgraph_same_body_alias): Update.
10462         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
10463         (verify_edge_corresponds_to_fndecl): Simplify.
10464         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
10465         (cgraph_node): Remove same_body_alias.
10466         (varpool_node): Remove alias_of and extra_name_alias.
10467         (same_body_aliases_done): Rename to ..
10468         (cpp_implicit_aliases_done): ... this one.
10469         (symtab_alias_ultimate_target): Add default parameter.
10470         (symtab_resolve_alias): New function.
10471         (fixup_same_cpp_alias_visibility): Declare.
10472         (cgraph_function_node): Add default parameter.
10473         (cgraph_node_asm_name): Likewise.
10474         (cgraph_function_or_thunk_node): Add default parameter; do
10475         not ICE when it is NULL.
10476         (varpool_variable_node): Likewise.
10477         * tree-emutls.c (create_emultls_var): Update.
10478         (ipa_lower_emutls): Update.
10479         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
10480         (cgraph_reset_node): Reset alias info.
10481         (cgraph_finalize_function): Update.
10482         (fixup_same_cpp_alias_visibility): Move to symtab.c.
10483         (analyze_function): Simplify.
10484         (cgraph_process_same_body_aliases): Simplify.
10485         (analyze_functions): Fixup same body aliases.
10486         (handle_alias_pairs): Simplify.
10487         (assemble_thunk): Update.
10488         (assemble_thunks_and_aliases): Update.
10489         (output_weakrefs): Rewrite.
10490         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
10491         (lto_output_varpool_node): Likewise.
10492         (compute_ltrans_boundary): Remve assert.
10493         (get_alias_symbol): New functoin.
10494         (input_node): Rewrite alias handling.
10495         (input_varpool_node): Likewise.
10496         * ipa-pure-const.c (propagate_pure_const): Fix formating.
10497         * ipa.c (process_references): Handle weakrefs correctly.
10498         (symtab_remove_unreachable_nodes): Likewise.
10499         * trans-mem.c (get_cg_data): Update.
10500         (ipa_tm_create_version_alias): Update.
10501         (ipa_tm_execute): Update.
10502         * symtab.c (dump_symtab_base): Dump aliases.
10503         (verify_symtab_base): Verify aliases.
10504         (symtab_node_availability): New function.
10505         (symtab_alias_ultimate_target): Simplify.
10506         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
10507         handle all the fixup cases.
10508         (symtab_resolve_alias): New function.
10509         * passes.c (ipa_write_summaries): Handle weakrefs.
10510         * varpool.c (varpool_analyze_node): Simplify.
10511         (assemble_aliases): Update.
10512         (varpool_create_variable_alias): Simplify.
10513         (varpool_extra_name_alias): Simplify.
10514         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
10515         (lto_symtab_merge_symbols): ... this one.
10517 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
10519         Revert
10520         PR rtl-optimization/57268
10521         * sched-deps.c (sched_analyze_2): Flush dependence lists if
10522         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
10524 2013-06-01  Tobias Burnus  <burnus@net-b.de>
10526         Partially reverted:
10527         2013-05-31  Tobias Burnus  <burnus@net-b.de>
10529         PR middle-end/57073
10530         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
10531         further up.
10533 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
10535         PR rtl-optimization/57268
10536         * sched-deps.c (sched_analyze_2): Flush dependence lists if
10537         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
10539 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
10541         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
10542         unordered comparison operators when -fno-trapping-math is in effect
10543         on the e500.
10544         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
10545         and implement unordered comparison operators properly on the e500.
10547 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
10549         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
10550         for constant scalar integers.
10551         (simplify_relational_operation_1): Likewise.
10553 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
10555         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
10556         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
10557         Fix comment.
10559 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
10560             Igor Zamyatin  <igor.zamyatin@intel.com>
10562         Silvermont (SLM) architecture performance tuning.
10563         * config/i386/i386.h (enum ix86_tune_indices): Add
10564         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
10565         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
10567         * config/i386/i386.c (initial_ix86_tune_features)
10568         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
10569         (ix86_lea_outperforms): Handle Silvermont tuning.
10570         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
10571         call.
10572         (ix86_use_lea_for_mov): Likewise.
10573         (ix86_avoid_lea_for_addr): Likewise.
10574         (ix86_lea_for_add_ok): Likewise.
10575         (exact_dependency_1): New function.
10576         (exact_store_load_dependency): Likewise.
10577         (ix86_adjust_cost): Handle Silvermont tuning.
10578         (do_reoder_for_imul): Likewise.
10579         (swap_top_of_ready_list): New function.
10580         (ix86_sched_reorder): Changed to handle Silvermont tuning.
10582         * config/i386/i386.md (peepholes that split memory operand in fp
10583         converts): New.
10585 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
10587         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
10588         Remove un-necessary braces.
10590 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
10592         * config/aarch64/aarch64.c (aarch64_classify_symbol):
10593         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
10595 2013-05-31  Tobias Burnus  <burnus@net-b.de>
10597         PR middle-end/57073
10598         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
10600 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10602         PR target/56315
10603         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
10604         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
10605         * config/arm/neon.md (iordi3_neon): Remove.
10606         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
10607         * config/arm/predicates.md (imm_for_neon_logic_operand):
10608         Move to earlier in the file.
10609         (neon_logic_op2): Likewise.
10610         (arm_iordi_operand_neon): New predicate.
10612 2013-05-31  Richard Biener  <rguenther@suse.de>
10614         PR tree-optimization/57478
10615         PR tree-optimization/57453
10616         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
10617         are life as well.
10619 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
10621         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
10622         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
10624 2013-05-30  Tobias Burnus  <burnus@net-b.de>
10625             Thomas Koenig  <tkoenig@gcc.gnu.org>
10627         PR middle-end/57073
10628         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
10629         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
10631 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
10633         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
10635 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
10637         * target.def (register_usage_leveling_p): New hook.
10638         * targhooks.c (default_register_usage_leveling_p): New.
10639         * targhooks.h (default_register_usage_leveling_p): New prototype.
10640         * lra-assigns.c (register_usage_leveling_p): Use the hook.
10641         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
10642         * doc/tm.texi: Update.
10643         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
10645 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
10647         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
10648         (*insv_reg<mode>): New define_insn.
10650 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
10652         PR rtl-optimization/57439
10653         * postreload.c (move2add_valid_value_p): Check that we have
10654         a zero subreg_regno_offset when accessing the register in
10655         the requested mode.
10657 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
10658             Igor Zamyatin  <igor.zamyatin@intel.com>
10660         Silvermont (SLM) architecture pipeline model, tuning and
10661         insn selection.
10662         * config.gcc: Add slm config options and target.
10664         * config/i386/slm.md: New.
10666         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
10668         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
10669         PROCESSOR_SLM.
10670         (ix86_target_macros_internal): Likewise.
10672         * config/i386/i386.c (slm_cost): New cost.
10673         (m_SLM): New macro flag.
10674         (initial_ix86_tune_features): Set m_SLM.
10675         (x86_accumulate_outgoing_args): Likewise.
10676         (x86_arch_always_fancy_math_387): Likewise.
10677         (processor_target_table): Add slm cost.
10678         (cpu_names): Add slm cpu name.
10679         (x86_option_override_internal): Set SLM ISA.
10680         (ix86_issue_rate): New case PROCESSOR_SLM.
10681         (ia32_multipass_dfa_lookahead): Likewise.
10682         (fold_builtin_cpu): Add slm.
10684         * config/i386/i386.h (TARGET_SLM): New target macro.
10685         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
10686         (processor_type): Add PROCESSOR_SLM.
10688         * config/i386/i386.md (cpu): Add new value "slm".
10689         (slm.md): Include slm.md.
10691 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
10692             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
10694         * config/arm/arm-protos.h: Add and update function protos.
10695         * config/arm/arm.c (use_simple_return_p): New added.
10696         (thumb2_expand_return): Check simple_return flag.
10697         * config/arm/arm.md: Add simple_return and conditional simple_return.
10698         * config/arm/iterators.md: Add iterator for return and simple_return.
10700 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
10702         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
10703         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
10704         (arm_emit_vfp_multi_reg_pop): Likewise.
10705         (thumb2_emit_ldrd_pop): Likewise.
10706         (arm_expand_epilogue): Add misc REG_CFA notes.
10707         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
10709 2013-05-29  Lawrence Crowl  <crowl@google.com>
10711         * config/arm/t-arm: Update for below.
10713         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
10714         Change type to hash_table.  Update dependent calls and types.
10716         * config/i386/t-cygming: Update for below.
10718         * config/i386/t-interix: Update for below.
10720         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
10721         Change type to hash_table.  Update dependent calls and types.
10722         (i386_find_on_wrapper_list::wrappers): Likewise.
10724         * config/ia64/t-ia64: Update for below.
10726         * config/ia64/ia64.c (bundle_state_table):
10727         Change type to hash_table.  Update dependent calls and types.
10729         * config/mips/mips.c (mips_reorg_process_insns::htab):
10730         Change type to hash_table.  Update dependent calls and types.
10732         * config/sol2.c (solaris_comdat_htab):
10733         Change type to hash_table.  Update dependent calls and types.
10735         * config/t-sol2: Update for above.
10737 2013-05-29  Teresa Johnson  <tejohnson@google.com>
10739         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
10740         functions are not yet marked as defined.
10742 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
10743             Pat Haugen  <pthaugen@us.ibm.com>
10744             Peter Bergner  <bergner@vnet.ibm.com>
10746         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
10747         instructions.
10748         (VEC_A): Likewise.
10749         (VEC_C): Likewise.
10750         (vrotl<mode>3): Likewise.
10751         (vashl<mode>3): Likewise.
10752         (vlshr<mode>3): Likewise.
10753         (vashr<mode>3): Likewise.
10755         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10756         support for power8 V2DI builtins.
10758         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
10759         power8 V2DI builtins.
10760         (vupkhsw): Likewise.
10761         (vupklsw): Likewise.
10762         (vaddudm): Likewise.
10763         (vminsd): Likewise.
10764         (vmaxsd): Likewise.
10765         (vminud): Likewise.
10766         (vmaxud): Likewise.
10767         (vpkudum): Likewise.
10768         (vpksdss): Likewise.
10769         (vpkudus): Likewise.
10770         (vpksdus): Likewise.
10771         (vrld): Likewise.
10772         (vsld): Likewise.
10773         (vsrd): Likewise.
10774         (vsrad): Likewise.
10775         (vsubudm): Likewise.
10776         (vcmpequd): Likewise.
10777         (vcmpgtsd): Likewise.
10778         (vcmpgtud): Likewise.
10779         (vcmpequd_p): Likewise.
10780         (vcmpgtsd_p): Likewise.
10781         (vcmpgtud_p): Likewise.
10782         (vupkhsw): Likewise.
10783         (vupklsw): Likewise.
10784         (vaddudm): Likewise.
10785         (vmaxsd): Likewise.
10786         (vmaxud): Likewise.
10787         (vminsd): Likewise.
10788         (vminud): Likewise.
10789         (vpksdss): Likewise.
10790         (vpksdus): Likewise.
10791         (vpkudum): Likewise.
10792         (vpkudus): Likewise.
10793         (vrld): Likewise.
10794         (vsld): Likewise.
10795         (vsrad): Likewise.
10796         (vsrd): Likewise.
10797         (vsubudm): Likewise.
10799         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
10800         support for power8 V2DI instructions.
10802         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
10803         power8 V2DI instructions.  Combine pack and unpack insns to use an
10804         iterator for each mode.  Check whether a particular mode supports
10805         Altivec instructions instead of just checking TARGET_ALTIVEC.
10806         (UNSPEC_VPKUWUM): Likewise.
10807         (UNSPEC_VPKSHSS): Likewise.
10808         (UNSPEC_VPKSWSS): Likewise.
10809         (UNSPEC_VPKUHUS): Likewise.
10810         (UNSPEC_VPKSHUS): Likewise.
10811         (UNSPEC_VPKUWUS): Likewise.
10812         (UNSPEC_VPKSWUS): Likewise.
10813         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
10814         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
10815         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
10816         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
10817         (UNSPEC_VUPKHSB): Likewise.
10818         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
10819         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
10820         (UNSPEC_VUPKHSH): Likewise.
10821         (UNSPEC_VUPKLSB): Likewise.
10822         (UNSPEC_VUPKLSH): Likewise.
10823         (VI2): Likewise.
10824         (VI_char): Likewise.
10825         (VI_scalar): Likewise.
10826         (VI_unit): Likewise.
10827         (VP): Likewise.
10828         (VP_small): Likewise.
10829         (VP_small_lc): Likewise.
10830         (VU_char): Likewise.
10831         (add<mode>3): Likewise.
10832         (altivec_vaddcuw): Likewise.
10833         (altivec_vaddu<VI_char>s): Likewise.
10834         (altivec_vadds<VI_char>s): Likewise.
10835         (sub<mode>3): Likewise.
10836         (altivec_vsubcuw): Likewise.
10837         (altivec_vsubu<VI_char>s): Likewise.
10838         (altivec_vsubs<VI_char>s): Likewise.
10839         (altivec_vavgs<VI_char>): Likewise.
10840         (altivec_vcmpbfp): Likewise.
10841         (altivec_eq<mode>): Likewise.
10842         (altivec_gt<mode>): Likewise.
10843         (altivec_gtu<mode>): Likewise.
10844         (umax<mode>3): Likewise.
10845         (smax<mode>3): Likewise.
10846         (umin<mode>3): Likewise.
10847         (smin<mode>3): Likewise.
10848         (altivec_vpkuhum): Likewise.
10849         (altivec_vpkuwum): Likewise.
10850         (altivec_vpkshss): Likewise.
10851         (altivec_vpkswss): Likewise.
10852         (altivec_vpkuhus): Likewise.
10853         (altivec_vpkshus): Likewise.
10854         (altivec_vpkuwus): Likewise.
10855         (altivec_vpkswus): Likewise.
10856         (altivec_vpks<VI_char>ss): Likewise.
10857         (altivec_vpks<VI_char>us): Likewise.
10858         (altivec_vpku<VI_char>us): Likewise.
10859         (altivec_vpku<VI_char>um): Likewise.
10860         (altivec_vrl<VI_char>): Likewise.
10861         (altivec_vsl<VI_char>): Likewise.
10862         (altivec_vsr<VI_char>): Likewise.
10863         (altivec_vsra<VI_char>): Likewise.
10864         (altivec_vsldoi_<mode>): Likewise.
10865         (altivec_vupkhsb): Likewise.
10866         (altivec_vupkhs<VU_char>): Likewise.
10867         (altivec_vupkls<VU_char>): Likewise.
10868         (altivec_vupkhsh): Likewise.
10869         (altivec_vupklsb): Likewise.
10870         (altivec_vupklsh): Likewise.
10871         (altivec_vcmpequ<VI_char>_p): Likewise.
10872         (altivec_vcmpgts<VI_char>_p): Likewise.
10873         (altivec_vcmpgtu<VI_char>_p): Likewise.
10874         (abs<mode>2): Likewise.
10875         (vec_unpacks_hi_v16qi): Likewise.
10876         (vec_unpacks_hi_v8hi): Likewise.
10877         (vec_unpacks_lo_v16qi): Likewise.
10878         (vec_unpacks_hi_<VP_small_lc>): Likewise.
10879         (vec_unpacks_lo_v8hi): Likewise.
10880         (vec_unpacks_lo_<VP_small_lc>): Likewise.
10881         (vec_pack_trunc_v8h): Likewise.
10882         (vec_pack_trunc_v4si): Likewise.
10883         (vec_pack_trunc_<mode>): Likewise.
10885         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
10886         V2DI builtins.
10887         (vec_vmaxsd): Likewise.
10888         (vec_vmaxud): Likewise.
10889         (vec_vminsd): Likewise.
10890         (vec_vminud): Likewise.
10891         (vec_vpksdss): Likewise.
10892         (vec_vpksdus): Likewise.
10893         (vec_vpkudum): Likewise.
10894         (vec_vpkudus): Likewise.
10895         (vec_vrld): Likewise.
10896         (vec_vsld): Likewise.
10897         (vec_vsrad): Likewise.
10898         (vec_vsrd): Likewise.
10899         (vec_vsubudm): Likewise.
10900         (vec_vupkhsw): Likewise.
10901         (vec_vupklsw): Likewise.
10903 2013-05-29  Jan Hubicka  <jh@suse.cz>
10905         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
10906         flags; reorder rest of fields in more consistent way.
10907         (varpool_node): Remove analyzed, finalized and alias.
10908         (cgraph_ndoe): Likewise.
10909         (symtab_alias_ultimate_target): New function.
10910         (cgraph_function_node): Move offline.
10911         (cgraph_reset_node): Declare.
10912         (cgraph_comdat_can_be_unshared_p): Remove.
10913         (varpool_remove_initializer): Declare.
10914         (varpool_first_defined_variable, varpool_next_defined_variable
10915         cgraph_first_defined_function, cgraph_next_defined_function): Update.
10916         (cgraph_function_with_gimple_body_p): Update.
10917         (varpool_all_refs_explicit_p): Update.
10918         (symtab_alias_target): New function.
10919         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
10920         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
10921         (cgraph_function_or_thunk_node): Simplify using
10922         symtab_alias_ultimate_target.
10923         (varpool_variable_node): Likewise.
10924         * cgraph.c (cgraph_create_function_alias): Update.
10925         (cgraph_add_thunk): Update.
10926         (cgraph_remove_node): Update.
10927         (dump_cgraph_node): Do not dump removed flags.
10928         (cgraph_function_body_availability): Update.
10929         (cgraph_propagate_frequency): Update.
10930         (verify_cgraph_node): Check sanity of local flag.
10931         (cgraph_function_node): Move here from cgraph.h; revamp for
10932         cgraph_function_or_thunk_node.
10933         * lto-symtab.c (lto_varpool_replace_node): Update.
10934         (lto_symtab_resolve_can_prevail_p): Update.
10935         (lto_symtab_merge_cgraph_nodes): Update.
10936         * ipa-cp.c (determine_versionability, initialize_node_lattices,
10937         propagate_constants_accross_call, devirtualization_time_bonus,
10938         ipcp_propagate_stage): Update.
10939         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
10940         * ipa-inline-transform.c (clone_inlined_nodes,
10941         preserve_function_body_p): Update.
10942         * ipa-reference.c (propagate): Update.
10943         (write_node_summary_p): Update.
10944         * toplev.c (wrapup_global_declaration_2): Update.
10945         * cgraphunit.c (cgraph_analyze_function): Rename to ...
10946         (analyze_function) ... this one.
10947         (cgraph_process_new_functions): Update.
10948         (cgraph_reset_node): Export.
10949         (cgraph_finalize_function): Update.
10950         (cgraph_add_new_function): Update.
10951         (process_function_and_variable_attributes): Update.
10952         (varpool_finalize_decl): Update.
10953         (symbol_finalized): Remove.
10954         (symbol_finalized_and_needed): Rename to ...
10955         (symbol_defined_and_needed): ... update.
10956         (cgraph_analyze_functions): Update.
10957         (handle_alias_pairs): Update.
10958         (mark_functions_to_output): Update.
10959         (assemble_thunk): Update.
10960         (output_in_order): Update.
10961         (output_weakrefs): Update.
10962         (finalize_compilation_unit): Update.
10963         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
10964         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
10965         input_node, input_varpool_node): Update.
10966         * dbxout.c (dbxout_expand_expr): Update.
10967         * cgraphclones.c (cgraph_clone_node): Update.
10968         (cgraph_copy_node_for_versioning): Update.
10969         (cgraph_materialize_clone): Update.
10970         (cgraph_materialize_all_clones): Update.
10971         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
10972         propagate_pure_const, propagate_nothrow): Update.
10973         * lto-streamer-out.c (lto_output, write_symbol): Update.
10974         * ipa-utils.c (ipa_reverse_postorder): Update.
10975         * ipa-inline.c (can_inline_edge_p): Update.
10976         (update_caller_keys, ipa_inline): Update.
10977         * dwarf2out.c (reference_to_unused,
10978         premark_types_used_by_global_vars_helper): Update.
10979         * tree-eh.c (tree_could_trap_p): Update.
10980         * ipa-split.c (consider_split, execute_split_functions): Update.
10981         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
10982         has_addr_references_p): Update; move ahead in file for better
10983         readability.
10984         (process_references): Simplify.
10985         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
10986         bodies are removed.
10987         (cgraph_comdat_can_be_unshared_p): Make static.
10988         (cgraph_externally_visible_p): Update.
10989         (varpool_externally_visible_p): Update.
10990         (function_and_variable_visibility): Update.
10991         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
10992         ipa_tm_mark_force_output_node): Update.
10993         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
10994         estimate_edge_devirt_benefit, inline_generate_summary,
10995         inline_write_summary): Update.
10996         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
10997         * ipa-prop.c (ipa_compute_jump_functions): Update.
10998         (ipa_print_node_params, ipa_prop_read_section,
10999         ipa_update_after_lto_read, read_replacements_section): Update.
11000         * varasm.c (mark_decl_referenced): Update.
11001         (assemble_alias, dump_tm_clone_pairs): Update.
11002         * tree-inline.c (copy_bb): Update.
11003         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
11004         Update.
11005         * symtab.c (dump_symtab_base): Print new flags.
11006         (verify_symtab_base): Verify new flags.
11007         (symtab_alias_ultimate_target): New function.
11008         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
11009         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
11010         Update.
11011         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
11012         Update.
11013         * i386.c (ix86_get_function_versions_dispatcher,
11014         ix86_generate_version_dispatcher_body): Update.
11015         (fold_builtin_cpu): Use varpool_add_new_variable.
11016         * varpool.c (varpool_remove_initializer): Break out from ...
11017         (varpool_remove_node): ... this one.
11018         (dump_varpool_node, varpool_node_for_asm,
11019         cgraph_variable_initializer_availability, varpool_analyze_node,
11020         varpool_assemble_decl, varpool_remove_unreferenced_decls,
11021         varpool_finalize_named_section_flags, varpool_create_variable_alias):
11022         Update.
11024 2013-05-29  Jan Hubicka  <jh@suse.cz>
11026         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
11028 2013-05-29  Easwaran Raman  <eraman@google.com>
11030         PR tree-optimization/57442
11031         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
11032         when control exits the main loop.
11034 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
11036         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
11037         and RX600.
11038         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
11039         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
11040         * rx/t-rx: Add rx100 under multi library matches option for nofpu
11041         option.
11043 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11045         PR tree-optimization/57441
11046         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
11047         Don't limit size of incr_vec to number of candidates.
11049 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
11051         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
11052         and mips16 directories.
11053         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
11054         (MULTILIB_DIRNAMES): Ditto.
11055         (MULTILIB_EXCEPTIONS): Add new exceptions.
11056         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
11057         (MULTILIB_DIRNAMES): Ditto.
11058         (MULTILIB_EXCEPTIONS): Add new exceptions.
11060 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
11061             Marcus Shawcroft  <marcus.shawcroft@arm.com>
11063         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
11064         SYMBOL_TINY_ABSOLUTE.
11065         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
11066         SYMBOL_TINY_ABSOLUTE.
11067         (aarch64_expand_mov_immediate): Likewise.
11068         (aarch64_classify_symbol): Likewise.
11069         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
11070         Permit SYMBOL_TINY_ABSOLUTE.
11071         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
11073 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
11074             Marcus Shawcroft  <marcus.shawcroft@arm.com>
11076         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
11077         Refactor if/switch.  Replace gcc_assert with if.
11079 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
11081         * config/i386/i386.c (initial_ix86_tune_features): Enable
11082         FP Reassociation for AMD bdver1 and bdver2.
11084 2013-05-29  Martin Jambor  <mjambor@suse.cz>
11086         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
11087         and IMAGPART_EXPR do not occur within other handled_components.
11089 2013-05-29  Richard Biener  <rguenther@suse.de>
11091         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
11092         access on whether the use is in the BB we currently try to
11093         vectorize.
11094         (vect_bb_vectorization_profitable_p): Pass the BB we currently
11095         vectorize to vect_bb_slp_scalar_cost.
11097 2013-05-29  Richard Biener  <rguenther@suse.de>
11099         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
11100         computing scalar cost offsetted by stmts that are kept live
11101         by scalar uses.
11102         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
11103         for computation of scalar cost.
11105 2013-05-28  Steve Ellcey  <sellcey@mips.com>
11107         * config/mips/mips-cpus.def (mips32r2): Change processor type.
11109 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
11111         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
11112         array notation built-in reduction functions.
11113         * doc/passes.texi (Passes): Added documentation about changes done
11114         for Cilk Plus.
11115         * doc/invoke.texi (C Dialect Options): Added documentation about
11116         the -fcilkplus flag.
11117         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
11118         (BUILTINS_DEF): Depend on cilkplus.def.
11119         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
11120         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
11121         * cilkplus.def: New file.
11123 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
11125         PR rtl-optimization/57439
11126         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
11128 2013-05-28  Easwaran Raman  <eraman@google.com>
11130         PR tree-optimization/57337
11131         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
11132         (find_insert_point): Correctly identify the insertion point
11133         when two statements with the same UID is compared.
11135 2013-05-28  Richard Biener  <rguenther@suse.de>
11137         PR tree-optimization/56787
11138         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
11139         from the list of data references.
11140         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
11141         clobbers.
11142         (vect_analyze_loop_operations): Likewise.
11143         (vect_transform_loop): Remove clobbers.
11145 2013-05-28  Martin Jambor  <mjambor@suse.cz>
11147         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
11148         and REALPART_EXPRs have scalar type.
11150 2013-05-28  Richard Biener  <rguenther@suse.de>
11152         PR tree-optimization/57411
11153         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
11154         virtual operands.
11155         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
11156         virtual operand propagation.
11158 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
11160         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
11161         destination register for bmasksi_vis.
11162         (vector_init_bshuffle): Likewise.
11163         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
11165 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
11167         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
11168         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
11169         mode if the instruction isn't available in the original mode.
11170         * config/sparc/sparc.opt (mfix-ut699): New option.
11171         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
11172         (divdf3): Turn into expander.
11173         (divdf3_nofix): New insn.
11174         (divdf3_fix): Likewise.
11175         (divsf3): Disable if -mfix-ut699.
11176         (sqrtdf2): Turn into expander.
11177         (sqrtdf2_nofix): New insn.
11178         (sqrtdf2_fix): Likewise.
11179         (sqrtsf2): Disable if -mfix-ut699.
11181 2013-05-27  Richard Biener  <rguenther@suse.de>
11183         PR middle-end/57412
11184         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
11185         block for the new loop.
11187 2013-05-27  Richard Biener  <rguenther@suse.de>
11189         PR tree-optimization/57343
11190         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
11191         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
11192         (number_of_iterations_cond): Do not build the folded tree.
11194 2013-05-27  Richard Biener  <rguenther@suse.de>
11196         Revert
11197         PR middle-end/57381
11198         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
11199         OEP_CONSTANT_ADDRESS_OF retained.
11201         PR tree-optimization/57417
11202         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
11203         for unchanged base.
11204         (set_ssa_val_to): Compare addresses using
11205         get_addr_base_and_unit_offset.
11207 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
11209         PR rtl-optimization/56833
11210         * postreload.c (move2add_record_mode): New function.
11211         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
11212         (move2add_use_add2_insn): Use move2add_record_sym_value.
11213         (move2add_use_add3_insn): Likewise.
11214         (reload_cse_move2add): Use move2add_valid_value_p and
11215         move2add_record_mode.  Invalidate call-clobbered and REG_INC
11216         affected regs by setting reg_mode to VOIDmode.
11217         (move2add_note_store): Don't pretend the inside of a SUBREG is
11218         the actual destination.  Invalidate single/leading registers by
11219         setting reg_mode to VOIDmode.
11220         Use move2add_record_sym_value, move2add_valid_value_p and
11221         move2add_record_mode.
11223 2013-05-27  Richard Biener  <rguenther@suse.de>
11225         PR tree-optimization/57396
11226         * tree-affine.c (double_int_constant_multiple_p): Properly
11227         return false for val == 0 and div != 0.
11229 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
11231         * config/mips/mips.h: Use #elif in preprocessor conditions.
11233 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
11235         PR target/53916
11236         * config/mips/constraints.md (kl): New constraint.
11237         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
11238         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
11239         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
11240         from LO for MIPS16.
11241         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
11243 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
11245         PR target/55777
11246         * config/mips/mips.c (mips_can_inline_p): New function.
11247         (TARGET_CAN_INLINE_P): Define.
11249 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
11251         * sched-int.h (ds_t, dw_t): Make unsigned int.
11252         Fix documentation that describes how all the ds_t bits are used.
11253         Reserve the last bit for delayed-branch scheduling.
11254         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
11255         (BITS_PER_DEP_WEAK): Fix definition and documentation.
11256         (gen_dep_weak_1): Remove prototype.
11257         * sched-deps.c (get_dep_weak_1): Make static.
11258         * target.def (speculate_insn, needs_block_p, gen_spec_check,
11259         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
11260         * doc/tm.texi: Regenerate.
11261         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
11263 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
11265         PR debug/56950
11266         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
11268 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
11269             Sandra Loosemore  <sandra@codesourcery.com>
11271         * config.gcc (powerpc-*): Allow native for with-cpu.
11273 2013-05-24  Jeff Law  <law@redhat.com>
11275         PR tree-optimization/57124
11276         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
11277         conversion feeding a condition if the range has an overflow
11278         if -fstrict-overflow.  Add warnings for when we do make the
11279         transformation.
11281 2013-05-24  Dehao Chen  <dehao@google.com>
11283         * tree-cfg.c (locus_discrim_map): Fix the typo.
11284         (locus_discrim_hasher): Likewise.
11285         (locus_discrim_hasher::hash): Likewise.
11286         (locus_discrim_hasher::equal): Likewise.
11288 2013-05-24  Martin Jambor  <mjambor@suse.cz>
11290         PR tree-optimization/57294
11291         * cgraph.h (ipa_record_stmt_references): Declare.
11292         * cgraphbuild.c (ipa_record_stmt_references): New function.
11293         (build_cgraph_edges): Use ipa_record_stmt_references.
11294         (rebuild_cgraph_edges): Likewise.
11295         (cgraph_rebuild_references): Likewise.
11296         * ipa-prop.c (ipa_modify_call_arguments): Discard references
11297         associated with the old statement and build references from the
11298         newly built statements.
11299         * ipa-ref.c (ipa_remove_stmt_references): New function.
11300         * ipa-ref.h (ipa_remove_stmt_references): Declare.
11302 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
11304         * lra-constraints.c (emit_spill_move): Use smaller mode for
11305         mem-mem moves.
11306         (check_and_process_move): Consider mem-reg moves for secondary
11307         too.
11308         (curr_insn_transform): Don't lose insns emitted before for
11309         secondary memory moves.
11310         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
11311         reg set up in the current insn.
11313 2013-05-24  Dehao Chen  <dehao@google.com>
11315         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
11316         hash function.
11317         (locus_descrim_hasher::equal): Likewise.
11318         (build_gimple_cfg): New discriminator assignment algorithm.
11319         (make_edges): Likewise.
11320         (next_discriminator_for_locus): Likewise.
11321         (same_line_p): Likewise.
11322         (assign_discriminators): Likewise.
11323         (make_cond_expr_edges): Likewise.
11324         (make_gimple_switch_edges): Likewise.
11325         (make_goto_expr_edges): Likewise.
11326         (make_gimple_asm_edges): Likewise.
11328 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
11330         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
11331         X format specifier to only display bottom 16 bits.
11332         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
11333         immediate to match for operand 2, since it will be masked.
11335 2013-05-24  Richard Biener  <rguenther@suse.de>
11337         PR tree-optimization/57287
11338         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
11339         all SSA names that occur in abnormal PHIs.
11341 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
11343         PR tree-ssa/57385
11344         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
11345         that index is not negative.
11347 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
11349         PR rtl-optimization/55177
11350         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
11351         (simplify_byte_swapping_operation): New.
11352         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
11353         (simplify_relational_operation_1): Deal with BSWAP.
11355 2013-05-23  Richard Henderson  <rth@redhat.com>
11357         PR target/56742
11358         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
11359         (ix86_reorg): Call it.
11361 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
11363         PR target/57379
11364         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
11365         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
11366         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
11368 2013-05-23  Christian Bruel  <christian.bruel@st.com>
11370         PR debug/57351
11371         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
11373 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
11374             Marcus Shawcroft  <marcus.shawcroft@arm.com>
11376         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
11377         * config/aarch64/constraints.md (Usa): Remove.
11378         * doc/md.texi (AArch64 Usa): Remove.
11380 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
11381             Marcus Shawcroft  <marcus.shawcroft@arm.com>
11383         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
11384         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
11385         * config/aarch64/predicates.md (aarch64_const_address): Remove.
11386         (aarch64_mov_operand): Use aarch64_mov_operand_p.
11388 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
11390         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
11391         instruction (AdvSIMD).
11392         * config/aarch64/aarch64-builtins.c
11393         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
11394         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
11396 2013-05-23  Martin Jambor  <mjambor@suse.cz>
11398         PR middle-end/57347
11399         * tree.h (contains_bitfld_component_ref_p): Declare.
11400         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
11401         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
11402         caller.
11403         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
11404         not access a bit-field.  Assert all final offsets are byte-aligned.
11406 2013-05-23  Richard Biener  <rguenther@suse.de>
11408         PR tree-optimization/57380
11409         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
11410         least one invariant or re-used load.
11411         * passes.c (init_optimization_passes): Move pass_phiprop before
11412         pass_forwprop.
11414 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
11416         * config/aarch64/aarch64-simd.md
11417         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
11419 2013-05-23  Richard Biener  <rguenther@suse.de>
11421         PR middle-end/57381
11422         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
11423         OEP_CONSTANT_ADDRESS_OF retained.
11425 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
11427         PR middle-end/57344
11428         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
11429         don't lower unit.  Handle unit not being always BITS_PER_WORD.
11431 2013-05-23  Richard Biener  <rguenther@suse.de>
11433         PR rtl-optimization/57341
11434         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
11435         instead of true_dependence.
11437 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
11439         * bb-reorder.c (branch_threshold): Make const.
11440         (exec_threshold): Ditto.
11442 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
11443             Pat Haugen  <pthaugen@us.ibm.com>
11444             Peter Bergner  <bergner@vnet.ibm.com>
11446         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
11447         documentation for the power8 crypto builtins.
11449         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
11451         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
11452         macros for defining power8 builtin functions.
11453         (BU_P8V_AV_2): Likewise.
11454         (BU_P8V_AV_P): Likewise.
11455         (BU_P8V_VSX_1): Likewise.
11456         (BU_P8V_OVERLOAD_1): Likewise.
11457         (BU_P8V_OVERLOAD_2): Likewise.
11458         (BU_CRYPTO_1): Likewise.
11459         (BU_CRYPTO_2): Likewise.
11460         (BU_CRYPTO_3): Likewise.
11461         (BU_CRYPTO_OVERLOAD_1): Likewise.
11462         (BU_CRYPTO_OVERLOAD_2): Likewise.
11463         (XSCVSPDP): Fix typo, point to the correct instruction.
11464         (VCIPHER): Add power8 crypto builtins.
11465         (VCIPHERLAST): Likewise.
11466         (VNCIPHER): Likewise.
11467         (VNCIPHERLAST): Likewise.
11468         (VPMSUMB): Likewise.
11469         (VPMSUMH): Likewise.
11470         (VPMSUMW): Likewise.
11471         (VPERMXOR_V2DI): Likewise.
11472         (VPERMXOR_V4SI: Likewise.
11473         (VPERMXOR_V8HI: Likewise.
11474         (VPERMXOR_V16QI: Likewise.
11475         (VSHASIGMAW): Likewise.
11476         (VSHASIGMAD): Likewise.
11477         (VPMSUM): Likewise.
11478         (VPERMXOR): Likewise.
11479         (VSHASIGMA): Likewise.
11481         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
11482         __CRYPTO__ if the crypto instructions are available.
11483         (altivec_overloaded_builtins): Add support for overloaded power8
11484         builtins.
11486         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
11487         support for power8 crypto builtins.
11488         (builtin_function_type): Likewise.
11489         (altivec_init_builtins): Add support for builtins that take vector
11490         long long (V2DI) arguments.
11492         * config/rs6000/crypto.md: New file, define power8 crypto
11493         instructions.
11495 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
11496             Pat Haugen  <pthaugen@us.ibm.com>
11497             Peter Bergner  <bergner@vnet.ibm.com>
11499         * doc/invoke.texi (Option Summary): Add power8 options.
11500         (RS/6000 and PowerPC Options): Likewise.
11502         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
11503         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
11504         wm, wn, wr documentation.
11506         * config/rs6000/constraints.md (wm): New constraint for VSX
11507         registers if direct move instructions are enabled.
11508         (wn): New constraint for no registers.
11509         (wq): New constraint for quad word even GPR registers.
11510         (wr): New constraint if 64-bit instructions are enabled.
11511         (wv): New constraint if power8 vector instructions are enabled.
11512         (wQ): New constraint for quad word memory locations.
11514         * config/rs6000/predicates.md (const_0_to_15_operand): New
11515         constraint for 0..15 for crypto instructions.
11516         (gpc_reg_operand): If VSX allow registers in VSX registers as well
11517         as GPR and floating point registers.
11518         (int_reg_operand): New predicate to match only GPR registers.
11519         (base_reg_operand): New predicate to match base registers.
11520         (quad_int_reg_operand): New predicate to match even GPR registers
11521         for quad memory operations.
11522         (vsx_reg_or_cint_operand): New predicate to allow vector logical
11523         operations in both GPR and VSX registers.
11524         (quad_memory_operand): New predicate for quad memory operations.
11525         (reg_or_indexed_operand): New predicate for direct move support.
11527         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
11528         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
11529         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
11530         (POWERPC_MASKS): Add power8 options.
11531         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
11532         various options.
11534         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
11535         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
11537         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
11538         (-mpower8-fusion): New power8 options.
11539         (-mpower8-fusion-sign): Likewise.
11540         (-mpower8-vector): Likewise.
11541         (-mcrypto): Likewise.
11542         (-mdirect-move): Likewise.
11543         (-mquad-memory): Likewise.
11545         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
11546         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
11547         registers.
11548         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
11549         (rs6000_debug_vector_unit): Add p8_vector.
11550         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
11551         definitions.  Also print fusion state.
11552         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
11553         (rs6000_builtin_mask_calculate): Add power8 builtin support.
11554         (rs6000_option_override_internal): Add support for power8.
11555         (rs6000_common_init_builtins): Add debugging for skipped builtins
11556         if -mdebug=builtin.
11557         (rs6000_adjust_cost): Add power8 support.
11558         (rs6000_issue_rate): Likewise.
11559         (insn_must_be_first_in_group): Likewise.
11560         (insn_must_be_last_in_group): Likewise.
11561         (force_new_group): Likewise.
11562         (rs6000_register_move_cost): Likewise.
11563         (rs6000_opt_masks): Likewise.
11565         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
11566         power8 capable assembler, default to power7 options.
11567         (TARGET_DIRECT_MOVE): Likewise.
11568         (TARGET_CRYPTO): Likewise.
11569         (TARGET_P8_VECTOR): Likewise.
11570         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
11571         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
11572         (VECTOR_MEM_P8_VECTOR_P): Likewise.
11573         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
11574         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
11575         (TARGET_XSCVDPSPN): Likewise.
11576         (TARGET_XSCVSPDPN): Likewsie.
11577         (TARGET_SYNC_HI_QI): Likewise.
11578         (TARGET_SYNC_TI): Likewise.
11579         (MASK_CRYPTO): Likewise.
11580         (MASK_DIRECT_MOVE): Likewise.
11581         (MASK_P8_FUSION): Likewise.
11582         (MASK_P8_VECTOR): Likewise.
11583         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
11584         temporary used by some of the direct move instructions to get two FP
11585         temporary registers does not force creation of a stack frame.
11586         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
11587         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
11588         that any VSX registers are tieable, even if they are also an
11589         Altivec vector mode.
11590         (r6000_reg_class_enum): Add wm, wr, wv constraints.
11591         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
11592         (RS6000_BTM_CRYPTO): Likewise.
11593         (RS6000_BTM_COMMON): Likewise.
11595         * config/rs6000/rs6000.md (cpu attribute): Add power8.
11596         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
11597         (enum rs6000_vector): Add power8 vector support.
11599 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11601         PR target/19599
11602         PR target/57340
11603         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
11604         (any_sibcall_could_use_r3): this and handle indirect calls.
11605         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
11607 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11609         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
11611 2013-05-22  Richard Biener  <rguenther@suse.de>
11613         PR middle-end/57349
11614         * profile.c (branch_prob): Do not split blocks that are
11615         abnormally receiving from ECF_RETURNS_TWICE functions.
11617 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
11619         * recog.c (offsettable_address_addr_space_p): Fix calculation of
11620         address mode.  Move pointer mode initialization to the same place.
11622 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
11624         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
11625         while it has any effect.
11627 2013-05-21  Easwaran Raman  <eraman@google.com>
11629         PR tree-optimization/57322
11630         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
11631         UID of the statement added to the BB to be 1.
11633 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
11635         PR tree-optimization/57331
11636         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
11637         of conversion from pointer type to integral type with integer.
11639 2013-05-21  Martin Jambor  <mjambor@suse.cz>
11641         PR lto/57289
11642         * ipa-prop.c (ipa_read_node_info): Process param_used and
11643         controlled_uses in the same order as when writing.
11645 2013-05-21  Magnus Granberg  <baldrick@free.fr>
11647         PR plugins/56754
11648         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
11650 2013-05-21  Richard Biener  <rguenther@suse.de>
11652         PR tree-optimization/57318
11653         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
11654         estimate stmts with side-effects as likely eliminated.
11656 2013-05-21  Richard Biener  <rguenther@suse.de>
11658         PR tree-optimization/57330
11659         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
11660         preserve the call stmts fntype.
11662 2013-05-21  Richard Biener  <rguenther@suse.de>
11664         PR tree-optimization/57303
11665         * tree-ssa-sink.c (statement_sink_location): Improve killing
11666         stmt detection and properly handle self-assignments.
11668 2013-05-21  Christian Bruel  <christian.bruel@st.com>
11670         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
11671         spanning registers. LEAF_REG_REMAP is supported only for contiguous
11672         registers. Set register size out of the PARALLEL loop.
11674 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
11676         PR target/56547
11677         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
11678         (*fmasf4, *fmasf4_media): New insns.
11680 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
11682         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
11683         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
11684         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
11685         (mips_idiv_insns): Update the comments to say that the returned
11686         instruction counts are in units of BASE_INSN_LENGTH.
11687         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
11688         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
11689         using 2 rather than 4 as the length of indirect MIPS16 and
11690         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
11691         length of a NOP.  Don't divide MIPS16 lengths by 2.
11692         (mips16_split_long_branches): Assume a branch is long if the
11693         length is greater than 4 rather than 8.
11694         * config/mips/mips.md (length): Give MIPS16 lengths directly,
11695         rather than multiplying them by 2.  Multiply instruction counts
11696         by BASE_INSN_LENGTH rather than 4.
11697         (*jump_mips16, tls_get_tp_mips16_<mode>)
11698         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
11700 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
11702         * config/mips/mips.md (extended_mips16): Remove branch case.
11703         (length): Remove duplicated extended_mips16 test.
11705 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
11707         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
11709 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
11711         * recog.h (Recog_data): Rename to...
11712         (recog_data_d): ...this.
11713         (recog_data): Update accordingly.
11714         * recog.c (recog_data): Likewise.
11715         * reload.c (save_recog_data): Likewise.
11716         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
11717         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
11719 2013-05-17  Julian Brown  <julian@codesourcery.com>
11721         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
11722         found in a REG_EQUAL note, invalidate it.
11724 2013-05-17  Easwaran Raman  <eraman@google.com>
11726         * tree-ssa-reassoc.c (find_insert_point): New function.
11727         (insert_stmt_after): Likewise.
11728         (get_def_stmt): Likewise.
11729         (ensure_ops_are_available): Likewise.
11730         (not_dominated_by): Likewise.
11731         (rewrite_expr_tree): Do not move statements beyond what is
11732         necessary. Remove call to swap_ops_for_binary_stmt...
11733         (reassociate_bb): ... and move it here.
11734         (build_and_add_sum): Assign UIDs for new statements.
11735         (linearize_expr): Likewise.
11736         (do_reassoc): Renumber gimple statement UIDs.
11738 2013-05-17  Jan Hubicka  <jh@suse.cz>
11740         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
11741         weakrefs.
11742         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
11743         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
11744         weakrefs.
11745         (output_weakrefs): Update.
11747 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
11748             Martin Jambor  <mjambor@suse.cz>
11750         PR middle-end/57276
11751         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
11752         value that corresponds to the given aggval is found in values vector.
11754 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
11756         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
11757         sse, sse2, sse3, ssse3 and sse4a flags to options.
11759 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
11761         * gengtype-state.c: (s_expr_writer): New class, to handle
11762         prettifying of output layout of s-expressions.
11763         (state_writer): New class, to write out gtype.state.
11764         (state_written_type_count): Move this variable into member data of
11765         state_writer.
11766         (s_expr_writer::s_expr_writer): New code: constructor for new class
11767         (state_writer::state_writer(): ditto
11768         (s_expr_writer::write_new_line): New function
11769         (s_expr_writer::write_any_indent): ditto
11770         (s_expr_writer::begin_s_expr): ditto
11771         (s_expr_writer::end_s_expr): ditto
11772         (write_state_fileloc): convert to method of state_writer...
11773         (state_writer:: write_state_fileloc): ...and use methods of
11774         s_expr_writer to write indentation into the gtype.state output file
11775         to visually represent the hierarchical structure of the list
11776         structures
11777         (write_state_fields): ditto, renaming to...
11778         (state_writer::write_state_fields)
11779         (write_state_a_string): ditto, renaming to...
11780         (state_writer::write_state_a_string)
11781         (write_state_string_option): ditto, renaming to...
11782         (state_writer::write_state_string_option)
11783         (write_state_type_option): ditto, renaming to...
11784         (state_writer::write_state_type_option)
11785         (write_state_nested_option): ditto, renaming to...
11786         (state_writer::write_state_nested_option)
11787         (write_state_option): ditto, renaming to...
11788         (state_writer::write_state_option)
11789         (write_state_options): ditto, renaming to...
11790         (state_writer::write_state_options)
11791         (write_state_lang_bitmap): ditto, renaming to...
11792         (state_writer::write_state_lang_bitmap)
11793         (write_state_version): ditto, renaming to...
11794         (state_writer::write_state_version)
11795         (write_state_scalar_type): ditto, renaming to...
11796         (state_writer::write_state_scalar_type)
11797         (write_state_string_type): ditto, renaming to...
11798         (state_writer::write_state_string_type)
11799         (write_state_undefined_type): ditto, renaming to...
11800         (state_writer::write_state_undefined_type)
11801         (write_state_struct_union_type): ditto, renaming to...
11802         (state_writer::write_state_struct_union_type)
11803         (write_state_struct_type): ditto, renaming to...
11804         (state_writer::write_state_struct_type)
11805         (write_state_user_struct_type): ditto, renaming to...
11806         (state_writer::write_state_user_struct_type)
11807         (write_state_lang_struct_type): ditto, renaming to...
11808         (state_writer::write_state_lang_struct_type)
11809         (write_state_param_struct_type): ditto, renaming to...
11810         (state_writer::write_state_param_struct_type)
11811         (write_state_pointer_type): ditto, renaming to...
11812         (state_writer::write_state_pointer_type)
11813         (write_state_array_type): ditto, renaming to...
11814         (state_writer::write_state_array_type)
11815         (write_state_gc_used): ditto, renaming to...
11816         (state_writer::write_state_gc_used)
11817         (write_state_common_type_content): ditto, renaming to...
11818         (state_writer::write_state_common_type_content)
11819         (write_state_type): ditto, renaming to...
11820         (state_writer::write_state_type)
11821         (write_state_pair_list): ditto, renaming to...
11822         (state_writer::write_state_pair_list)
11823         (write_state_pair): ditto, renaming to...
11824         (state_writer::write_state_pair)
11825         (write_state_typedefs): ditto, renaming to...
11826         (state_writer::write_state_typedefs)
11827         (write_state_structures): ditto, renaming to...
11828         (state_writer::write_state_structures)
11829         (write_state_param_structs): ditto, renaming to...
11830         (state_writer::write_state_param_structs)
11831         (write_state_variables): ditto, renaming to...
11832         (state_writer::write_state_variables)
11833         (write_state_srcdir): ditto, renaming to...
11834         (state_writer::write_state_srcdir)
11835         (write_state_files_list): ditto, renaming to...
11836         (state_writer::write_state_files_list)
11837         (write_state_languages): ditto, renaming to...
11838         (state_writer::write_state_languages)
11839         (write_state): create a state_writer instance and use it when
11840         writing out the state file
11842 2013-05-17  Mike Stump  <mikestump@comcast.net>
11844         PR rtl-optimization/57304
11845         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
11846         accessing DF_REF_REAL_LOC.
11848 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
11850         PR rtl-optimization/57281
11851         PR rtl-optimization/57300
11852         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
11853         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
11854         what the other splitter did if the registers are dead.
11856 2013-05-17  Richard Biener  <rguenther@suse.de>
11858         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
11859         MEM_REF offsets.
11861 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
11863         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
11864         linking.
11866 2013-05-17  Marek Polacek  <polacek@redhat.com>
11868         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
11869         length when doing non-zero store of storing '\0' to '\0'.
11871 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
11873         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
11874         vect_external_def oprnd1 with loop_vinfo, try to emit
11875         optional cast, negation and and stmts on the loop preheader
11876         edge instead of into the pattern def seq.
11878         PR tree-optimization/57051
11879         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
11880         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
11882 2013-05-16  Nick Clifton  <nickc@redhat.com>
11884         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
11885         (rl78_is_naked_func): New function.
11886         (rl78_expand_prologue): Skip prologue generation for naked functions.
11887         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
11888         * doc/extend.texi (naked): Add RL78 to the list of processors
11889         that supports this attribute.
11891 2013-05-16  Jeff Law  <law@redhat.com>
11893         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
11895 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
11897         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
11898         cache parameters using detect_caches_amd also for CYRIX,
11899         NSC and TM2 signatures.
11901 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
11902             Dzianis Kahanovich  <mahatma@eu.by>
11904         PR target/45359
11905         PR target/46396
11906         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
11907         VIA/Centaur processors and determine their cache parameters
11908         using detect_caches_amd.
11910 2013-05-16  Teresa Johnson  <tejohnson@google.com>
11912         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
11913         (rtl_verify_edges): New function.
11914         (rtl_verify_bb_insns): Ditto.
11915         (rtl_verify_bb_pointers): Ditto.
11916         (rtl_verify_bb_insn_chain): Ditto.
11917         (rtl_verify_fallthru): Ditto.
11918         (rtl_verify_bb_layout): Ditto.
11919         (rtl_verify_flow_info_1): Outline checks into new functions.
11920         (rtl_verify_flow_info): Ditto.
11922 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
11924         * cfghooks.c (copy_bbs): Add update_dominance argument.
11925         * cfghooks.h (copy_bbs): Update prototype.
11926         * tree-cfg.c (gimple_duplicate_sese_region):
11927         Add update_dominance argument.
11928         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
11929         * tree-ssa-loop-ch.c (copy_loop_headers): Update
11930         gimple_duplicate_sese_region call.
11931         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
11932         Update copy_bbs call.
11933         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
11934         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
11936 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
11938         * tree-vectorizer.h (NUM_PATTERNS): Increment.
11939         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
11940         vect_recog_rotate_pattern.
11941         (vect_recog_rotate_pattern): New function.
11943 2013-05-16  Jason Merrill  <jason@redhat.com>
11945         * Makefile.in (LLINKER): New variable.
11946         (mostlyclean): Remove link mutex.
11947         * configure.ac: Handle --enable-link-mutex.
11948         * lock-and-run.sh: New script.
11950 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11952         PR target/19599
11953         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
11954         for NULL decl.
11956 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11958         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
11960 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
11962         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
11963         * config/arm/arm.c (next_consecutive_mem): New function.
11964         (gen_movmem_ldrd_strd): Likewise.
11965         * config/arm/arm.md (movmemqi): Update condition and code.
11966         (unaligned_loaddi, unaligned_storedi): New patterns.
11968 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11970         * config.gcc: Obsolete *-*-solaris2.9*.
11971         * doc/install.texi (Specific, *-*-solaris2*): Document it.
11973 2013-05-16  Richard Biener  <rguenther@suse.de>
11975         * passes.c (init_optimization_passes): Move pass_parallelize_loops
11976         earlier, after GRAPHITE transforms and IV canonicalization.
11978 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
11980         * omp-low.c (extract_omp_for_data): For collapsed loops,
11981         if at least one of the loops is known at compile time to
11982         iterate zero times, set count to 0.
11983         (expand_omp_regimplify_p): New function.
11984         (expand_omp_for_generic): For collapsed loops, if at least
11985         one of the loops isn't known to iterate at least once,
11986         add runtime check with setting count to 0.
11987         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
11988         For unsigned types if it isn't known at compile time that
11989         the loop will iterate at least once, add runtime check to bypass
11990         the whole loop if initial condition isn't true.
11992 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
11994         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
11996 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
11998         PR middle-end/57286
11999         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
12000         transformations to avoid an infinite loop.
12002 2013-05-16  Marek Polacek  <polacek@redhat.com>
12004         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
12006 2013-05-15  Leif Ekblad  <leif@rdos.net>
12008         * config/i386/i386.c (ix86_decompose_address): Use
12009         DEFAULT_TLS_SEG_REG to access TLS segment register.
12010         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
12011         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
12012         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
12014 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
12016         PR target/57260
12017         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
12018         sibling calls to functions that would normally be lazily bound,
12019         unless $gp is call-clobbered.
12021 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
12023         * config/i386/i386.c (ix86_option_override_internal): Update
12024         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
12025         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
12026         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
12027         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
12028         of TARGET_3DNOW.
12029         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
12031 2013-05-15  Andreas Schwab  <schwab@suse.de>
12033         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
12034         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
12035         third operand.
12037 2013-05-15  Teresa Johnson  <tejohnson@google.com>
12039         * loop-unroll.c (report_unroll_peel): Check decision before
12040         emitting unroll/peel message.
12042 2013-05-15  Teresa Johnson  <tejohnson@google.com>
12044         * function.h (has_bb_partition): New rtl_data flag.
12045         (bb_reorder_complete): Ditto.
12046         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
12047         instead of flag_reorder_blocks_and_partition.
12048         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
12049         with some enhancements.
12050         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
12051         * bb-reorder.c (connect_traces): Check for has_bb_partition
12052         instead of flag_reorder_blocks_and_partition.
12053         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
12054         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
12055         verify_hot_cold_block_grouping.
12056         (partition_hot_cold_basic_blocks): Set has_bb_partition.
12058 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12060         PR target/19599
12061         * config/arm/predicates.md (call_insn_operand): New predicate.
12062         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
12063         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
12064         if insn is not a tail call.
12065         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
12066         registers.
12067         * config/arm/arm.h (enum reg_class): New caller save register class.
12068         (REG_CLASS_NAMES): Likewise.
12069         (REG_CLASS_CONTENTS): Likewise.
12070         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
12071         without decls.
12073 2013-05-15  Richard Biener  <rguenther@suse.de>
12075         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
12076         of MSG_OPTIMIZED_LOCATIONS.
12077         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
12078         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
12079         message.
12080         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
12081         of MSG_OPTIMIZED_LOCATIONS.
12082         (execute_vect_slp): Likewise.
12083         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
12084         (vect_create_cond_for_alias_checks): Likewise.
12085         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
12086         (vect_recog_widen_mult_pattern): Likewise.
12087         (vect_recog_widen_sum_pattern): Likewise.
12088         (vect_recog_over_widening_pattern): Likewise.
12089         (vect_recog_widen_shift_pattern): Likewise.
12090         (vect_recog_vector_vector_shift_pattern): Likewise.
12091         (vect_recog_divmod_pattern): Likewise.
12092         (vect_recog_mixed_size_cond_pattern): Likewise.
12093         (vect_recog_bool_pattern): Likewise.
12094         (vect_pattern_recog_1): Likewise.
12096 2013-05-15  Martin Jambor  <mjambor@suse.cz>
12098         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
12099         non-functions to builtin_unreachable.
12100         * ipa-inline-transform.c (inline_call): Do not assert estimates were
12101         correct when new direct edges were discovered.
12103 2013-05-15  Martin Jambor  <mjambor@suse.cz>
12105         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
12106         header, print symbol order instead of node uid, print more information
12107         about indirect edge targets.
12108         (ipa_make_edge_direct_to_target): Print symbol order instead of node
12109         uids.
12110         (ipa_make_edge_direct_to_target): Likewise.
12111         (remove_described_reference): Likewise.
12112         (propagate_controlled_uses): Likewise.
12113         (ipa_print_node_params): Also print symbol order.
12114         (ipcp_transform_function): Print symbol order instead of node uids.
12115         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
12116         (cgraph_get_create_real_symbol_node): Likewise.
12117         * ipa-cp.c (print_lattice): Likewise.
12118         (print_all_lattices): Likewise.
12119         (determine_versionability): Likewise.
12120         (initialize_node_lattices): Likewise.
12121         (estimate_local_effects): Likewise.
12122         (update_profiling_info): Likewise.
12123         (create_specialized_node): Likewise.
12124         (perhaps_add_new_callers): Likewise.
12125         (decide_about_value): Likewise.
12126         (decide_whether_version_node): Likewise.
12127         (identify_dead_nodes): Likewise.
12128         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
12129         (dump_inline_summary): Likewise.
12130         (estimate_node_size_and_time): Likewise.
12131         (inline_analyze_function): Likewise.
12132         * ipa-inline.c (report_inline_failed_reason): Likewise.
12133         (want_early_inline_function_p): Likewise.
12134         (edge_badness): Likewise.
12135         (update_edge_key): Likewise.
12136         (inline_small_functions): Likewise.  Add dumping of order to two other
12137         dumps.
12138         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
12139         instead of node uids.
12140         (propagate_pure_const): Likewise.
12141         (propagate_pure_const): Likewise.
12142         * ipa-utils.c (dump_cgraph_node_set): Likewise.
12143         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
12144         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
12145         of node uids.
12146         * tree-pretty-print.c (dump_function_header): Likewise.
12147         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
12148         Print symbol order instead of node uids.
12150 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12152         * config/s390/s390.c (s390_register_move_cost): Don't impose the
12153         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
12155 2013-05-15  Richard Biener  <rguenther@suse.de>
12157         PR tree-optimization/57275
12158         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
12159         return value for fail to do runtime alias checks for gather loads.
12161 2013-05-15  Jan Hubicka  <jh@suse.cz>
12163         PR lto/57038
12164         PR lto/47375
12165         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
12166         weakrefs are not external.
12167         (lto_symtab_merge_decls): Fix thinko when dealing with
12168         non-lto_symtab decls.
12169         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
12170         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
12171         * varpool.c (dump_varpool_node): Dump more flags.
12173 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
12175         * config/i386/i386.c (processor_alias_table): Add instruction
12176         FSGSBASE for AMD bdver3 architecture.
12178 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
12180         * tree.c (warn_deprecated_use): Print file:line using locus color.
12181         * diagnostic.c (diagnostic_report_current_module): Print file:line
12182         and file:line:column using locus color.
12184 2013-05-14  Mike Stump  <mikestump@comcast.net>
12186         * gdbinit.in: Add __null.
12188 2013-05-14  Mike Stump  <mikestump@comcast.net>
12190         * recog.h: Rename struct recog_data to Recog_data.
12191         * recog.c: Likewise.
12192         * reload.c (can_reload_into): Likewise.
12193         * config/picochip/picochip.c: Likewise.
12195 2013-05-14  Mike Stump  <mikestump@comcast.net>
12197         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
12199 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
12201         * resource.h (struct resources): Remove unch_memory member.
12202         (CLEAR_RESOURCE): Don't clear unch_memory.
12203         * resource.c (mark_referenced_resources): Don't set it.
12204         (mark_set_resources): Likewise.
12205         (mark_target_live_regs): Don't clear it.
12206         (init_resource_info): Likewise.
12207         * reorg.c (resource_conflicts_p): Don't compare it.
12208         (redundant_insn): Don't set it.
12210         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
12211         Remove prototypes.
12212         * emit-rtl.c (next_label): Remove unused function.
12213         (skip_consecutive_labels, link_cc0_insns): Move to ...
12214         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
12215         only place where these functions are used, and make them static.
12217 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
12219         * fold-const.c (fold_negate_expr): Handle vectors.
12220         (fold_truth_not_expr): Make it static.
12221         (fold_invert_truthvalue): New static function.
12222         (invert_truthvalue_loc): Handle vectors. Do not call
12223         fold_truth_not_expr directly.
12224         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
12225         <TRUTH_NOT_EXPR>: Do not cast to boolean.
12226         (fold_comparison): Handle vector constants.
12227         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
12228         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
12229         * tree.h (fold_truth_not_expr): Remove declaration.
12231 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
12233         * config/aarch64/aarch64-simd.md
12234         (aarch64_vcond_internal<mode>): Rename to...
12235         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
12236         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
12237         float modes. Clarify all iterator modes.
12238         (vcond<mode><mode>): Use new name for vcond expanders.
12239         (vcond<v_cmp_result><mode>): Likewise.
12240         (vcondu<mode><mode>: Likewise.
12241         * config/aarch64/iterators.md (VDQF_COND): New.
12243 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
12245         PR bootstrap/57266
12246         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
12247         variable for the shift amount. Check that we shift by non-negative
12248         amounts.
12250 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
12252         PR target/42017
12253         * config/arm/arm.h (EPILOGUE_USES): Only return true
12254         for LR_REGNUM after epilogue_completed.
12256 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
12258         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
12259         is error_mark_node.
12261 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
12263         PR target/57261
12264         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
12265         and Solaris 11+/x86 with gld.
12266         * configure: Regenerate.
12268 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
12270         * expmed.c (expand_shift_1): Canonicalize rotates by
12271         constant bitsize / 2 to bitsize - 1.
12272         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
12273         case ROTATERT>: Likewise.
12275         Revert:
12276         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
12278         * config/i386/i386.md (rotateinv): New code attr.
12279         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
12280         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
12281         roll $31, %eax, etc.
12283 2013-05-14  Richard Biener  <rguenther@suse.de>
12285         PR middle-end/57235
12286         * tree-eh.c (sink_clobbers): Give up for successors with
12287         multiple predecessors and no virtual uses.
12289 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
12291         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
12292         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
12294 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
12296         PR middle-end/57251
12297         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
12298         the case when both op0 and op1 have VOIDmode.
12300 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
12302         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
12303         in multiply-accumulate mode.
12305 2013-05-13  Guozhi Wei  <carrot@google.com>
12307         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
12309 2013-05-13  Kai Tietz  <ktietz@redhat.com>
12311         PR target/56975
12312         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
12313         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
12314         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
12315         * config/i386/i386.c (ix86_option_override_internal): Likewise.
12316         (ix86_expand_prologue): Likewise.
12317         (ix86_expand_split_stack_prologue): Likewise.
12318         (legitimate_pic_address_disp_p): Likewise.
12319         (legitimize_pic_address): Likewise.
12320         (legitimize_tls_address): Likewise.
12321         (legitimize_pe_coff_symbol): Likewise.
12322         (output_pic_addr_const): Likewise.
12323         (construct_plt_address): Likewise.
12324         (ix86_expand_call): Likewise.
12325         (x86_output_mi_thunk): Likewise.
12326         (x86_function_profiler): Likewise.
12328 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
12330         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
12331         similar switch cases.
12332         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
12333         (aarch64_simd_mov_to_<mode>low): Delete.
12334         (aarch64_simd_mov_to_<mode>high): Delete.
12335         (move_lo_quad_<mode>): Add w<-r alternative.
12336         (aarch64_simd_move_hi_quad_<mode>): Likewise.
12337         (aarch64_simd_mov_from_*): Update type attribute.
12338         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
12339         statement.
12341 2013-05-13  Jan Hubicka  <jh@suse.cz>
12343         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
12344         * config/i386/i386.c (ix86_compute_frame_layout,
12345         ix86_expand_epilogue, emit_i387_cw_initialization,
12346         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
12347         ix86_local_alignment): Fix use of size/speed predicates.
12349 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
12351         PR tree-optimization/45216
12352         PR tree-optimization/57157
12353         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
12354         the (-Y) & (B - 1) variant if OP is |.
12355         * expmed.c (expand_shift_1): For rotations by const0_rtx just
12356         return shifted.  Use (-op1) & (prec - 1) as other_amount
12357         instead of prec - op1.
12359 2013-05-13  Martin Jambor  <mjambor@suse.cz>
12361         PR middle-end/42371
12362         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
12363         (ipa_constant_data): New type.
12364         (ipa_jump_func): Use ipa_constant_data to hold information about
12365         constant jump functions.
12366         (ipa_get_jf_constant): Adjust to jump function type changes.
12367         (ipa_get_jf_constant_rdesc): New function.
12368         (ipa_param_descriptor): New field controlled_uses.
12369         (ipa_get_controlled_uses): New function.
12370         (ipa_set_controlled_uses): Likewise.
12371         * ipa-ref.h (ipa_find_reference): Declare.
12372         * ipa-prop.c (ipa_cst_ref_desc): New type.
12373         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
12374         changes.
12375         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
12376         New parameter cs.  Adjust all callers.
12377         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
12378         (remove_described_reference): New function.
12379         (jfunc_rdesc_usable): Likewise.
12380         (try_make_edge_direct_simple_call): Decrement controlled use count,
12381         attempt to remove reference if it hits zero.
12382         (combine_controlled_uses_counters): New function.
12383         (propagate_controlled_uses): Likewise.
12384         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
12385         (ipa_edge_duplication_hook): Duplicate reference descriptions.
12386         (ipa_print_node_params): Print described use counter.
12387         (ipa_write_jump_function): Adjust to jump function type changes.
12388         (ipa_read_jump_function): New parameter CS, pass it to
12389         ipa_set_jf_constant.  Adjust caller.
12390         (ipa_write_node_info): Stream controlled use count
12391         (ipa_read_node_info): Likewise.
12392         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
12393         asserting.
12394         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
12395         count.  Remove cloning-added reference if it reaches zero.
12396         * ipa-ref.c (ipa_find_reference): New function.
12398 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
12400         * config/i386/i386.c (processor_target_table): Modified default
12401         alignment values for AMD BD and BT architectures.
12403 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
12405         * tree-vect-generic.c (uniform_vector_p): Move ...
12406         * tree.c (uniform_vector_p): ... here.
12407         * tree.h (uniform_vector_p): Declare it.
12408         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
12409         into a scalar.
12411 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
12413         PR tree-optimization/57230
12414         * tree-ssa-strlen.c (handle_char_store): Record length for
12415         array store from STRING_CST.
12417         PR tree-optimization/57230
12418         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
12419         check.
12421 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
12423         * config/epiphany/epiphany.c (epiphany_init): Check size of
12424         NUM_MODES_FOR_MODE_SWITCHING.
12425         (epiphany_expand_prologue):
12426         Remove CONFIG_REGNUM initial value handling code.
12427         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
12428         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
12429         (emit_set_fp_mode, epiphany_mode_after): Likewise.
12430         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
12431         Don't return 1 for FP_MODE_NONE.
12432         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
12433         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
12434         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
12435         * config/epiphany/epiphany.md (save_config): New pattern.
12437 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
12439         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
12441 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
12443         * config/i386/i386.md (memory): Handle sseishft1.
12444         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
12445         (*vec_extractv2di_1): Ditto.
12447 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
12449         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
12450         saved registers.
12452 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
12454         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
12455         Add mthumb/march=armv7-a multilib.
12456         Add mthumb/march=armv7-r multilib.
12457         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
12459 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
12461         * config/v850/t-rtems: Add more multilibs.
12463 2013-05-10  Richard Biener  <rguenther@suse.de>
12465         PR tree-optimization/57214
12466         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
12467         not propagate from SSA names that occur in abnormal PHI nodes.
12469 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
12471         * stor-layout.c (element_precision): New function.
12472         * machmode.h (element_precision): Declare it.
12473         * tree.c (build_minus_one_cst): New function.
12474         (element_precision): Likewise.
12475         * tree.h (build_minus_one_cst): Declare new function.
12476         (element_precision): Likewise.
12477         * fold-const.c (operand_equal_p): Use element_precision.
12478         (fold_binary_loc): Handle vector types.
12479         * convert.c (convert_to_integer): Use element_precision.
12480         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
12481         separately.
12483 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
12485         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
12486         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
12487         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
12488         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
12489         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
12490         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
12491         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
12492         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
12493         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
12494         (Uuw8): New constraints.
12495         (Usb4): Move into alphabetical order.
12496         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
12497         (sd8_operand, ub8_operand, uw8_operand): New predicates.
12498         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
12499         previously unnamed patterns.
12500         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
12501         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
12502         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
12503         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
12504         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
12505         of specific lengths.
12507 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
12509         * config/i386/i386.md (rotateinv): New code attr.
12510         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
12511         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
12512         roll $31, %eax, etc.
12514         PR tree-optimization/45216
12515         PR tree-optimization/57157
12516         * tree-ssa-forwprop.c (simplify_rotate): New function.
12517         (ssa_forward_propagate_and_combine): Call it.
12519 2013-05-10  Richard Biener  <rguenther@suse.de>
12521         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
12522         disable peeling when we version for aliasing.
12523         (vector_alignment_reachable_p): Honor explicit user alignment.
12524         (vect_supportable_dr_alignment): Likewise.
12525         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
12526         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
12527         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
12528         then peeling to arrange for the cost-model check to come first.
12530 2013-05-10  Alan Modra  <amodra@gmail.com>
12532         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
12533         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
12534         * configure: Regenerate.
12536 2013-05-10  Alan Modra  <amodra@gmail.com>
12538         PR target/55033
12539         * varasm.c (default_elf_select_section): Move !DECL_P check..
12540         (get_named_section): ..to here before calling get_section_name.
12541         Adjust assertion.
12542         (default_section_type_flags): Add DECL_P check.
12543         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
12544         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
12546 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
12548         * config/epiphany/epiphany.c (epiphany_expand_prologue):
12549         When using gen_stack_adjust_str with a register offset, add a
12550         REG_FRAME_RELATED_EXPR note.
12552 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
12554         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
12555         (*vec_extractv4si_zext_mem): Ditto.
12556         (*vec_extractv2di): Add 0->x and x->x alternatives.
12557         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
12558         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
12560 2013-05-09  Jason Merrill  <jason@redhat.com>
12562         N3639 C++1y VLA support
12563         * gimplify.c (gimplify_vla_decl): Don't touch an existing
12564         DECL_VALUE_EXPR.
12566         * tree.c (build_constructor_va): New.
12567         * tree.h: Declare it.
12569 2013-05-09  Martin Jambor  <mjambor@suse.cz>
12571         PR lto/57084
12572         * gimple-fold.c (canonicalize_constructor_val): Call
12573         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
12575 2013-05-09  Jan Hubicka  <jh@suse.cz>
12576             Richard Biener  <rguenther@suse.de>
12578         PR lto/54095
12579         * symtab.c (symtab_make_decl_local): Do not add private names.
12581 2013-05-09  Jan Hubicka  <jh@suse.cz>
12583         PR lto/54095
12584         * symtab.c (insert_to_assembler_name_hash): Handle clones.
12585         (unlink_from_assembler_name_hash): Likewise.
12586         (symtab_prevail_in_asm_name_hash, symtab_register_node,
12587         symtab_unregister_node, symtab_initialize_asm_name_hash,
12588         change_decl_assembler_name): Update.
12590 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
12592         * config/aarch64/aarch64.md: New movtf split.
12593         (*movtf_aarch64): Update.
12594         (aarch64_movdi_tilow): Handle TF modes and rename to
12595         aarch64_movdi_<mode>low.
12596         (aarch64_movdi_tihigh): Handle TF modes and rename to
12597         aarch64_movdi_<mode>high
12598         (aarch64_movtihigh_di): Handle TF modes and rename to
12599         aarch64_mov<mode>high_di
12600         (aarch64_movtilow_di): Handle TF modes and rename to
12601         aarch64_mov<mode>low_di
12602         (aarch64_movtilow_tilow): Remove spurious whitespace.
12603         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
12604         splits.
12605         (aarch64_print_operand): Update.
12607 2013-05-09  Alan Modra  <amodra@gmail.com>
12609         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
12610         powerpc64le.
12611         * configure: Regenerate.
12613 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
12615         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
12616         splitter preparation statements.
12617         * config/i386/sse.md (*vec_extract* splitters): Ditto.
12618         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
12619         adjust_address_nv.
12621 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12623         * gimple-ssa-strength-reduction.c (count_candidates): Change
12624         return value to int.
12625         (analyze_candidates_and_replace): Change type of length to int.
12627 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
12629         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
12630         (*vec_extract<mode>): Use VI12_128 mode iterator.
12631         (*vec_extract<mode>_mem): Ditto.
12632         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
12633         attribute.
12635 2013-05-08  Diego Novillo  <dnovillo@google.com>
12637         PR bootstrap/54659
12639         Revert:
12640         2012-08-17  Diego Novillo  <dnovillo@google.com>
12642         PR bootstrap/54281
12643         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
12644         * config.in: Regenerate.
12645         * configure: Regenerate.
12646         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
12648 2013-05-08  Jan Hubicka  <jh@suse.cz>
12650         PR lto/54095
12651         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
12652         * cgraph.h (symtab_node_base): Add unique_name.
12653         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
12654         input_overwrite_node, input_varpool_node): Stream unique_name.
12655         * cgraphclones.c (cgraph_create_virtual_clone,
12656         cgraph_function_versioning): Set unique_name.
12657         * ipa.c (function_and_variable_visibility): Set unique_name.
12659 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12661         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
12662         (alloc_cand_and_find_basis): Restrict conditional candidate
12663         processing to CAND_MULTs.
12665 2013-05-08  Jan Hubicka  <jh@suse.cz>
12667         PR lto/54095
12668         lto-symtab.c (lto_symtab_symbol_p): New function.
12669         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
12670         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
12671         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
12672         Skip static symbols.
12674 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
12676         PR tree-optimization/57200
12677         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
12678         Only call inform if the preceding warning_at returns true.
12680 2013-05-07  Han Shen  <shenhan@google.com>
12682         * cfgexpand.c (record_or_union_type_has_array_p): New function.
12683         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
12684         * common.opt (fstack-protector-strong): New option.
12685         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
12686         * doc/invoke.texi (Optimization Options): Document
12687         "-fstack-protector-strong".
12688         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
12690 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
12692         * config/mips/mips.c (mips_machine_reorg2): Return 0.
12694 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
12696         * ira.c (update_equiv_regs): Add insn having equiv memory even if
12697         it is not lhs of the insn.
12698         (setup_reg_equiv): Remove insn having equiv memory which it is not
12699         lhs of the insn.
12700         * lra-constraints.c (process_address): Try to improve generation
12701         code for address base + disp.
12702         (lra_constraints): Make correct the code for checking insn setting
12703         up backward equivalence.  Remove insn only if it is in the init
12704         insn list.
12705         * lra-eliminations.c (update_reg_eliminate): Change return value.
12706         (lra_eliminate): Use the result.
12708 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
12710         * config/i386/sse.md (ssescalarnummask): New mode attribute.
12711         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
12712         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
12713         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
12714         register target operands.
12715         (*vec_extractv8hi_sse2): New pattern.
12716         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
12717         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
12718         (*vec_extract<mode>_mem): New insn and split pattern.
12720 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
12722         * config/arm/arm.c (arm_asan_shadow_offset): New function.
12723         (TARGET_ASAN_SHADOW_OFFSET): Define.
12724         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
12725         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
12727 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12729         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
12730         (incr_vec_index): Return -1 if increment not found.
12731         (create_add_on_incoming_edge): Assert if increment not found.
12732         (record_increment): Limit number of increments recorded.
12733         (all_phi_incrs_profitable): Return false if an increment not found.
12734         (replace_profitable_candidates): Don't process increments that were
12735         not recorded.
12736         (analyze_candidates_and_replace): Limit size of incr_vec.
12738 2013-05-07  Richard Biener  <rguenther@suse.de>
12740         * calls.c (special_function_p): setjmp-like functions are leaf.
12741         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
12742         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
12744 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
12746         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
12747         (aarch64_simd_mov<mode>): New expander.
12748         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
12749         (aarch64_simd_mov_to_<mode>high): Likewise.
12750         (aarch64_simd_mov_from_<mode>low): Likewise.
12751         (aarch64_simd_mov_from_<mode>high): Likewise.
12752         (aarch64_dup_lane<mode>): Update.
12753         (aarch64_dup_lanedi): New instruction pattern.
12754         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
12755         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
12757 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12759         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
12760         (replace_mult_candidate): Remove unnecessary argument; remove
12761         unnecessary parameter from call to introduce_cast_before_cand.
12762         (replace_unconditional_candidate): Remove unnecessary parameter
12763         from call to replace_mult_candidate.
12764         (replace_conditional_candidate): Likewise.
12765         (insert_initializers): Use make_temp_ssa_name.
12766         (introduce_cast_before_cand): Remove unnecessary argument; use
12767         make_temp_ssa_name.
12768         (replace_one_candidate): Remove unnecessary argument; remove
12769         unnecessary parameter from calls to introduce_cast_before_cand.
12770         (replace_profitable_candidates): Remove unnecessary parameters
12771         from calls to replace_one_candidate.
12773 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12775         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12776         phi def as possibly hiding a basis for a CAND_ADD whose operands
12777         have been commuted in the analysis.
12778         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
12780 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12782         * config/aarch64/aarch64.md
12783         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
12784         shift value between 0-4.
12786 2013-05-07  Richard Biener  <rguenther@suse.de>
12788         * double-int.h (rshift): New overload.
12789         * double-int.c (rshift): New function.
12790         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
12791         (create_reference_ops_from_ref): Remove.
12792         (vn_reference_insert): Use shared ops for constructing the
12793         reference and copy it.
12795 2013-05-07  Richard Biener  <rguenther@suse.de>
12797         PR middle-end/57190
12798         * tree-eh.c (sink_clobbers): Properly propagate
12799         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
12801 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
12803         PR tree-optimization/57149
12804         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
12805         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
12806         collect_phi_def_edges, execute_late_warn_uninitialized): Use
12807         uninit_undefined_value_p instead of ssa_undefined_value_p.
12809         PR debug/57184
12810         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
12811         for modifier == EXPAND_INITIALIZER.
12813 2013-05-07  Anton Blanchard  <anton@samba.org>
12815         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
12816         for powerpc64 little endian.
12817         * configure: Regenerate.
12819 2013-05-06  Graham Stott  <grahams@btinternet.com>
12821         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
12822         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
12823         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
12824         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
12826 2013-05-06  Graham Stott  <grahams@btinternet.com>
12828         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
12829         codes which allow non-lvalues.
12831 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
12833         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
12834         components are all 1s.
12835         (integer_minus_onep): New function.
12836         * tree.h (integer_minus_onep): Declare it.
12837         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
12838         integer_minus_onep instead of integer_all_onesp.
12840 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
12842         PR target/52933
12843         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
12844         variations of these patterns.
12846 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
12848         * config/i386/i386.md (isa): Add x64_sse4 member.
12849         (enabled): Handle x64_sse4.
12850         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
12851         instruction for 64bit SSE4_1 targets.  Update insn attributes.
12852         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
12853         instruction for SSE4_1 targets.  Update insn attributes.
12854         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
12855         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
12856         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
12857         const_1 selector.
12858         (*vec_extractv4si): Rename from *sse4_1_pextrd.
12859         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
12860         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
12862 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
12864         PR target/57108
12865         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
12867 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
12869         * final.c (do_assembler_dialects): Don't handle curly braces and
12870         vertical bar escaped by % as dialect delimiters.
12871         (output_asm_insn): Print curly braces and vertical bar if escaped
12872         by % and ASSEMBLER_DIALECT defined.
12873         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
12874         * doc/tm.texi: Regenerated.
12876 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
12878         * config/mips/mips.c: Include tree-pass.h.
12879         (mips_reorg): Split in pre- and post-dbr_schedule parts.
12880         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
12881         (pass_mips_machine_reorg2): New machine specific pass.
12882         (insert_pass_mips_machine_reorg2): New pass plugin definition.
12883         (mips_option_override): Register the new pass.
12884         * rtl.h (cleanup_barriers): Remove prototype.
12885         (dbr_schedule): Likewise.
12886         * jump.c (cleanup_barriers): Make static.
12887         * reorg.c (dbr_schedule): Likewise.
12889 2013-05-06  Richard Biener  <rguenther@suse.de>
12891         PR tree-optimization/57185
12892         * tree-parloops.c (add_field_for_reduction): Handle anonymous
12893         SSA names properly.
12895 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
12897         PR target/57106
12898         * config/i386/i386.c (add_parameter_dependencies): Add dependence
12899         between "first_arg" and "insn", not "last" and "insn".
12901 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12903         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
12904         (find_candidates_in_block): Re-enable slsr_process_phi.
12905         (create_phi_basis): Fix double counting of candidate adjustment.
12907 2013-05-06  Richard Biener  <rguenther@suse.de>
12909         PR middle-end/57147
12910         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
12911         the edge is also fallthru, preserve it and just clear the
12912         abnormal flag.
12913         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
12914         also complex, preserve that and just clear the fallthru flag.
12915         * tree-inline.c (update_ssa_across_abnormal_edges): Also
12916         update virtual operands.
12918 2013-05-06  Alan Modra  <amodra@gmail.com>
12920         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
12921         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
12922         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
12923         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
12924         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
12925         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
12927 2013-05-06  Alan Modra  <amodra@gmail.com>
12929         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
12930         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
12931         (DEFAULT_ASM_ENDIAN): Define.
12932         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
12933         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
12934         Update -K PIC clause from sysv4.h.
12935         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
12936         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
12938 2013-05-06  Alan Modra  <amodra@gmail.com>
12940         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
12941         twice for little-endian.
12942         (ashrdi3_no_power, ashrdi3): Support little-endian.
12944 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
12946         PR target/55303
12947         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
12948         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
12949         related expanders.
12950         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
12951         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
12952         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
12953         New predicates.
12955 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
12956             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
12958         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
12959         * config/pa/pa.opt: Make mbig-switch a no-op.
12960         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
12961         (CASE_VECTOR_MODE): Always return SImode.
12962         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
12963         for the !TARGET_BIG_SWITCH case.
12964         * config/pa/pa-linux.h: Likewise.
12965         * config/pa/pa-openbsd.h: Likewise.
12966         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
12967         * config/pa/pa.md (short_jump): Remove define_insn.
12968         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
12969         (casesi0): Remove define_insn.
12970         (type): Remove btable_branch.
12971         (pa_combine_type): Likewise.
12972         (in_nullified_branch_delay): Likewise.
12973         (in_call_delay): Likewise.
12974         (define_delay): Likewise.
12975         (define_insn_reservation "Z3"): Likewise.
12976         (define_insn_reservation "Z4"): Likewise.
12977         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
12978         (pa_adjust_insn_length): Remove adjustment for btable branches.
12979         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
12980         and mno-big-switch
12982 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
12984         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
12985         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
12986         Add m->r,x alternatives.
12987         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
12988         splitters using SWI48x mode iterator.
12989         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
12990         TARGET_64BIT.  Add m->x alternative.
12991         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
12992         Add o->x alternative.  Enable for TARGET_SSE.
12993         (sse_storeq): Remove expander.
12994         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
12995         with memory input operand.
12996         (*vec_extractv2di_1 splitter): New.
12997         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
12998         * config/i386/i386.md (ssevecmodelower): New mode attribute.
13000 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
13002         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
13003         (INT_LOWPART): Delete.
13004         (extract_MB): Adjust.
13005         (extract_ME): Adjust.
13006         (print_operand): Adjust.
13008 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
13010         * config/rs6000/predicates.md (reg_or_add_cint_operand,
13011         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
13012         (reg_or_logical_cint_operand, easy_fp_constant,
13013         logical_const_operand): Delete "CONST_DOUBLE" case.
13014         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
13015         "HOST_BITS_PER_WIDE_INT == 64" test.
13016         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
13017         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
13018         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
13019         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
13020         test.
13021         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
13022         CONST_DOUBLE DImode/VOIDmode case.
13023         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
13024         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
13025         CONST_DOUBLE VOIDmode case.
13026         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
13027         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
13028         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
13029         Delete CONST_DOUBLE case.
13030         (splitters for mov FMOVE64 const_double): Delete
13031         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
13032         "HOST_BITS_PER_WIDE_INT >= 64" test.
13033         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
13034         case.
13035         (mov DI const_double): Delete.
13037 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
13039         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
13040         on op shows all bits zero in mode of a lowpart subreg, return zero.
13042 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
13044         PR target/57150
13045         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
13046         to save TFmode registers and DImode to save TImode registers for
13047         caller save operations.
13048         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
13049         mark being partially clobbered since they only use the first
13050         double word.
13052         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
13053         and TDmode only use the upper 64-bits of each VSX register.
13055 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13057         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
13058         (find_candidates_in_block): Disable slsr_process_phi.
13060 2013-05-03  Guozhi Wei  <carrot@google.com>
13062         * coverage.c (coverage_obj_init): Move the construction of gcov
13063         constructor to ...
13064         (build_init_ctor): ... here.
13066 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13068         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
13069         (slsr_cand_d): Redefine def_phi.
13070         (stride_status, phi_adjust_status, count_phis_status): New enums.
13071         (find_phi_def): New.
13072         (find_basis_for_base_expr): New.
13073         (find_basis_for_candidate): Handle hidden bases.
13074         (alloc_cand_and_find_basis): Handle phi candidates.
13075         (slsr_process_phi): New.
13076         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
13077         (create_mul_imm_cand): Likewise.
13078         (create_add_ssa_cand): Exclude phi base candidates.
13079         (create_add_imm_cand): Likewise.
13080         (slsr_process_cast): Likewise.
13081         (slsr_process_copy): Likewise.
13082         (find_candidates_in_block): Handle phi candidates.
13083         (dump_candidate): Likewise.
13084         (unconditional_cands): Delete.
13085         (unconditional_cands_with_known_stride_p): Delete.
13086         (phi_dependent_cand_p): New.
13087         (cand_increment): Handle phi-dependent candidates.
13088         (replace_dependent): Delete.
13089         (replace_mult_candidate): New.
13090         (replace_unconditional_candidate): New.
13091         (incr_vec_index): Move to avoid forward reference.
13092         (create_add_on_incoming_edge): New.
13093         (create_phi_basis): New.
13094         (replace_dependents): Delete.
13095         (replace_conditional_candidate): New.
13096         (phi_add_costs): New.
13097         (replace_uncond_cands_and_profitable_phis): New.
13098         (record_increment): Handle phi adjustments.
13099         (record_phi_increments): New.
13100         (record_increments): Handle phi adjustments.
13101         (phi_incr_cost): New.
13102         (lowest_cost_path): Handle phis.
13103         (total_savings): Likewise.
13104         (analyze_increments): Likewise.
13105         (ncd_with_phi): New.
13106         (ncd_of_cand_and_phis): New.
13107         (nearest_common_dominator_for_cands): Handle phi increments.
13108         (all_phi_incrs_profitable): New.
13109         (replace_profitable_candidates): Handle phi-dependent candidates.
13110         (analyze_candidates_and_replace): Likewise.
13112 2013-05-03  Teresa Johnson  <tejohnson@google.com>
13114         PR bootstrap/57154
13115         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
13116         do not exceed REG_BR_PROB_BASE.
13118 2013-05-03  Jeff Law  <law@redhat.com>
13120         PR tree-optimization/57144
13121         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
13122         operand of the condition will bit into the new type when eliminating
13123         a cast feeding a condition.
13125 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
13127         PR rtl-optimization/57130
13128         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
13129         of COMPARE as in_code to the recursive call if needed.
13131 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
13133         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
13134         (enabled): Handle new members.
13135         * config/i386/sse.md (*vec_concatv2si): Merge from
13136         *vec_concatv2si_sse2 and vec_concatv2si_sse.
13137         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
13139 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
13141         PR tree-optimization/57027
13142         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
13143         for fnms opportunity, check we got the prerequisite kind
13144         of tree / gimple before using accessor functions.
13146 2013-05-03  Richard Biener  <rguenther@suse.de>
13148         * double-int.h (lshift): New overload without precision
13149         and arith argument.
13150         (operator *=, operator +=, operator -=): Move ...
13151         * double-int.c (operator *=, operator +=, operator -=): ... here
13152         and implement more efficiently.
13153         (mul_double_with_sign): Remove.
13154         (lshift_double): Adjust to take unsinged shift argument, push
13155         dispatching code to callers.
13156         (mul_double_wide_with_sign): Add early out for callers that
13157         are not interested in high parts or overflow.
13158         (lshift): New function.
13159         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
13160         dispatch code here.
13161         (lrotate, rrotate): Use logical shifts.
13162         * expr.c (get_inner_reference): Use lshift.
13163         * fixed-value.c (do_fixed_divide): Likewise.
13164         * tree-dfa.c (get_ref_base_and_extent): Likewise.
13165         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
13166         (indirect_refs_may_alias_p): Likewise.
13167         (stmt_kills_ref_p_1): Likewise.
13169 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
13171         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
13173 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
13175         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
13176         scalar form of FABD instruction.
13178 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
13180         * lra-constraints.c (process_alt_operands): Add checking alt
13181         number to choose the best alternative.
13183 2013-05-02  Richard Biener  <rguenther@suse.de>
13185         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
13186         bitmap and its handling.
13187         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
13189 2013-05-02  Richard Biener  <rguenther@suse.de>
13191         PR middle-end/57140
13192         * tree-inline.c (copy_loops): Properly handle removed loops.
13193         (copy_cfg_body): Mark destination loops for fixup if source
13194         loops needed fixup.
13196 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
13198         PR target/56732
13199         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
13200         generating simple_return for naked functions.
13202 2013-05-02  Martin Jambor  <mjambor@suse.cz>
13204         PR middle-end/56988
13205         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
13206         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
13207         flags match.
13208         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
13209         ipa_agg_replacement_value structures.
13210         (known_aggs_to_agg_replacement_list): Likewise.
13211         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
13212         (read_agg_replacement_chain): Likewise.
13213         (ipcp_transform_function): Also check that by_ref flags match.
13215 2013-05-02  Richard Biener  <rguenther@suse.de>
13217         * graphds.h (struct graph): Add obstack member.
13218         * graphds.c (new_graph): Initialize obstack and allocate
13219         vertices from it.
13220         (add_edge): Allocate edge from the obstack.
13221         (free_graph): Free the obstack instead of all edges and vertices.
13223 2013-05-02  Teresa Johnson  <tejohnson@google.com>
13225         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
13226         divides.
13227         * cfg.c (update_bb_profile_for_threading): Ditto.
13228         * tree-inline.c (copy_bb): Ditto.
13229         (copy_edges_for_bb): Ditto.
13230         (initialize_cfun): Ditto.
13231         (copy_cfg_body): Ditto.
13232         (expand_call_inline): Ditto.
13233         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
13234         (estimate_node_size_and_time): Ditto.
13235         (inline_merge_summary): Ditto.
13236         * cgraphclones.c (cgraph_clone_edge): Ditto.
13237         (cgraph_clone_node): Ditto.
13238         * sched-rgn.c (compute_dom_prob_ps): Ditto.
13239         (compute_trg_info): Ditto.
13241 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
13243         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
13244         S reg when fp attribute set.
13245         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
13247 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
13249         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
13250         New pattern.
13251         (*and_one_cmplsi3_compare0_uxtw): Likewise.
13252         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
13253         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
13255 2013-05-02  Richard Biener  <rguenther@suse.de>
13257         * tree-scalar-evolution.c (scev_info_hasher): Remove.
13258         (struct instantiate_cache_entry): New type.
13259         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
13260         (struct instantiate_cache_type): New type.
13261         (set_instantiated_value, get_instantiated_value): Remove.
13262         (get_instantiated_value_entry): New function.
13263         (instantiate_scev_name): Use the new cache and adjust.
13264         (instantiate_scev_poly): Adjust.
13265         (instantiate_scev_binary): Likewise.
13266         (instantiate_array_ref): Likewise.
13267         (instantiate_scev_convert): Likewise.
13268         (instantiate_scev_not): Likewise.
13269         (instantiate_scev_3): Likewise.
13270         (instantiate_scev_2): Likewise.
13271         (instantiate_scev_r): Likewise.
13272         (instantiate_scev): Likewise.
13273         (resolve_mixers): Likewise.
13275 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
13277         PR target/57091
13278         * lra-constraints.c (best_small_class_operands_num): Remove.
13279         (process_alt_operands): Remove small_class_operands_num.  Take
13280         small classes operands into losers and only if the operand is not
13281         matched.  Modify debugging output.
13282         (curr_insn_transform): Remove best_small_class_operands_num.
13283         Print insn name.
13285 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13287         * config/aarch64/aarch64-builtins.c
13288         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
13289         * config/aarch64/aarch64-simd-builtins.def
13290         (reduc_splus_): Add new modes.
13291         (reduc_uplus_): New.
13292         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
13293         (reduc_uplus_v4sf): Likewise.
13294         (reduc_splus_v4sf): Likewise.
13295         (aarch64_addv<mode>): Likewise.
13296         (reduc_uplus_<mode>): Likewise.
13297         (reduc_splus_<mode>): Likewise.
13298         (aarch64_addvv2di): Likewise.
13299         (reduc_uplus_v2di): Likewise.
13300         (reduc_splus_v2di): Likewise.
13301         (aarch64_addvv2si): Likewise.
13302         (reduc_uplus_v2si): Likewise.
13303         (reduc_splus_v2si): Likewise.
13304         (reduc_<sur>plus_<mode>): New.
13305         (reduc_<sur>plus_v2di): Likewise.
13306         (reduc_<sur>plus_v2si): Likewise.
13307         (reduc_<sur>plus_v4sf): Likewise.
13308         (aarch64_addpv4sf): Likewise.
13309         * config/aarch64/arm_neon.h
13310         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
13311         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
13312         add UNSPEC_SADDV, UNSPEC_UADDV.
13313         (SUADDV): New.
13314         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
13316 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13318         * config/aarch64/arm_neon.h
13319         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
13321 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13323         * config/aarch64/aarch64-builtins
13324         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
13326 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13328         * config/aarch64/aarch64-simd-builtins.def
13329         (reduc_smax_): New.
13330         (reduc_smin_): Likewise.
13331         (reduc_umax_): Likewise.
13332         (reduc_umin_): Likewise.
13333         (reduc_smax_nan_): Likewise.
13334         (reduc_smin_nan_): Likewise.
13335         (fmax): Remove.
13336         (fmin): Likewise.
13337         (smax): Update for V2SF, V4SF and V2DF modes.
13338         (smin): Likewise.
13339         (smax_nan): New.
13340         (smin_nan): Likewise.
13341         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
13342         (<su><maxmin><mode>3): ...This, refactor.
13343         (s<maxmin><mode>3): New.
13344         (<maxmin_uns><mode>3): Likewise.
13345         (reduc_<maxmin_uns>_<mode>): Refactor.
13346         (reduc_<maxmin_uns>_v4sf): Likewise.
13347         (reduc_<maxmin_uns>_v2si): Likewise.
13348         (aarch64_<fmaxmin><mode>: Remove.
13349         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
13350         new builtin names.
13351         (vmin<q>_f<32,64>): Likewise.
13352         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
13353         (FMAXMIN): New.
13354         (su): Add mappings for smax, smin, umax, umin.
13355         (maxmin): New.
13356         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
13357         (FMAXMIN): Rename as...
13358         (FMAXMIN_UNS): ...This.
13359         (maxminv): Remove.
13360         (fmaxminv): Likewise.
13361         (fmaxmin): Likewise.
13362         (maxmin_uns): New.
13363         (maxmin_uns_op): Likewise.
13365 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13367         * config/aarch64/arm_neon.h
13368         (vac<ge, gt><sd>_f<32, 64>): Rename to...
13369         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
13370         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
13372 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13374         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
13375         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
13377 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13379         * config/aarch64/aarch64-simd.md
13380         (vcond<mode>_internal): Handle special cases for constant masks.
13381         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
13382         (vcondu<mode><mode>): Likewise.
13383         (vcond<v_cmp_result><mode>): New.
13385 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13387         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
13388         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
13389         * config/aarch64/aarch64-simd-builtins.def
13390         (cmeq): Update to BUILTIN_VALLDI.
13391         (cmgt): Likewise.
13392         (cmge): Likewise.
13393         (cmle): Likewise.
13394         (cmlt): Likewise.
13395         * config/aarch64/arm_neon.h
13396         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
13397         to builtins or C as appropriate.
13399 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13401         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
13402         (cmgeu): ...This.
13403         (cmhi): Rename to...
13404         (cmgtu): ...This.
13405         * config/aarch64/aarch64-simd.md
13406         (simd_mode): Add SF.
13407         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
13408         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
13409         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
13410         (cstore<mode>_neg): ...This.
13411         * config/aarch64/iterators.md
13412         (VALLF): new.
13413         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
13414         (COMPARISONS): New.
13415         (UCOMPARISONS): Likewise.
13416         (optab): Add missing comparisons.
13417         (n_optab): New.
13418         (cmp_1): Likewise.
13419         (cmp_2): Likewise.
13420         (CMP): Likewise.
13421         (cmp): Remove.
13422         (VCMP_S): Likewise.
13423         (VCMP_U): Likewise.
13424         (V_cmp_result): Add DF, SF modes.
13425         (v_cmp_result): Likewise.
13426         (v): Likewise.
13427         (vmtype): Likewise.
13428         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
13430 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
13432         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
13433         define_insn to define_insn_and_split.
13434         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
13435         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
13436         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
13437         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
13438         (thumb2_negscc): Likewise.
13440 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
13442         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
13444 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
13446         * config/arm/thumb2.md: Remove trailing whitespaces.
13448 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
13450         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
13451         Use gen_int_mode rather than GEN_INT.
13453 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
13455         * value-prof.c (stream_in_histogram_value): Remove the strayed
13456         debug_gimple_stmt.
13458 2013-04-30  Richard Biener  <rguenther@suse.de>
13460         PR middle-end/57122
13461         * cfghooks.c (split_edge): Properly check for the loop latch edge.
13463 2013-04-30  Richard Biener  <rguenther@suse.de>
13465         PR middle-end/57107
13466         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
13468 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
13470         PR rtl-optimization/56957
13471         PR rtl-optimization/57105
13472         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
13473         variable.  Use just INSN_UID for determining whether an insn
13474         should be only disconnected from the insn stream.
13475         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
13477 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
13479         PR tree-optimization/57104
13480         * tsan.c (instrument_expr): Don't instrument accesses to
13481         DECL_HARD_REGISTER VAR_DECLs.
13483 2013-04-30  Richard Biener  <rguenther@suse.de>
13485         * function.h (loops_for_fn): New inline function.
13486         (set_loops_for_fn): Likewise.
13487         * cfgloop.h (place_new_loop): Add struct function parameter.
13488         (get_loop): Likewise.
13489         (get_loops): Likewise.
13490         (number_of_loops): Likewise.
13491         (fel_next): Adjust.
13492         (fel_init): Likewise.
13493         * cfg.c (get_loop_copy): Adjust.
13494         * cfgloop.c (flow_loops_dump): Likewise.
13495         (record_loop_exits): Likewise.
13496         (verify_loop_structure): Likewise.
13497         * cfgloopanal.c (mark_irreducible_loops): Likewise.
13498         (estimate_reg_pressure_cost): Likewise.
13499         (mark_loop_exit_edges): Likewise.
13500         * cfgloopmanip.c (place_new_loop): Likewise.
13501         (add_loop): Likewise.
13502         (duplicate_loop): Likewise.
13503         * graph.c (draw_cfg_nodes): Likewise.
13504         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
13505         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
13506         (extract_affine_chrec): Likewise.
13507         (build_scop_iteration_domain): Likewise.
13508         * graphite.c (graphite_initialize): Likewise.
13509         * ira-build.c (create_loop_tree_nodes): Likewise.
13510         (more_one_region_p): Likewise.
13511         (rebuild_regno_allocno_maps): Likewise.
13512         (mark_loops_for_removal): Likewise.
13513         (mark_all_loops_for_removal): Likewise.
13514         (remove_unnecessary_regions): Likewise.
13515         (ira_build): Likewise.
13516         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
13517         * loop-init.c (fix_loop_structure): Likewise.
13518         (gate_rtl_move_loop_invariants): Likewise.
13519         (gate_rtl_unswitch): Likewise.
13520         (gate_rtl_unroll_and_peel_loops): Likewise.
13521         (rtl_doloop): Likewise.
13522         * lto-streamer-in.c (input_cfg): Likewise.
13523         * lto-streamer-out.c (output_cfg): Likewise.
13524         * modulo-sched.c (sms_schedule): Likewise.
13525         * predict.c (tree_estimate_probability): Likewise.
13526         (tree_estimate_probability_driver): Likewise.
13527         (estimate_loops): Likewise.
13528         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
13529         (move_sese_region_to_fn): Likewise.
13530         (debug_loop_num): Likewise.
13531         * tree-chrec.c (chrec_evaluate): Likewise.
13532         (hide_evolution_in_other_loops_than_loop): Likewise.
13533         (chrec_component_in_loop_num): Likewise.
13534         (reset_evolution_in_loop): Likewise.
13535         (evolution_function_is_invariant_rec_p): Likewise.
13536         * tree-if-conv.c (main_tree_if_conversion): Likewise.
13537         * tree-inline.c (copy_loops): Likewise.
13538         (copy_cfg_body): Likewise.
13539         (tree_function_versioning): Likewise.
13540         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
13541         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
13542         Likewise.
13543         (add_to_evolution_1): Likewise.
13544         (scev_const_prop): Likewise.
13545         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
13546         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
13547         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
13548         (tree_ssa_lim_initialize): Likewise.
13549         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
13550         (verify_loop_closed_ssa): Likewise.
13551         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
13552         (tree_ssa_loop_im): Likewise.
13553         (tree_ssa_loop_unswitch): Likewise.
13554         (tree_vectorize): Likewise.
13555         (check_data_deps): Likewise.
13556         (tree_ssa_loop_ivcanon): Likewise.
13557         (tree_ssa_loop_bounds): Likewise.
13558         (tree_complete_unroll): Likewise.
13559         (tree_complete_unroll_inner): Likewise.
13560         (tree_parallelize_loops): Likewise.
13561         (tree_ssa_loop_prefetch): Likewise.
13562         (tree_ssa_loop_ivopts): Likewise.
13563         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
13564         * tree-vectorizer.c (vectorize_loops): Likewise.
13566 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
13568         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
13569         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
13570         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
13571         with EABI_LINK_SPEC.
13573 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
13575         PR target/44578
13576         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
13577         alternative.
13579 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
13581         PR target/57097
13582         * lra-constraints.c (process_alt_operands): Discourage a bit more
13583         using memory for pseudos.  Print cost dump for alternatives.
13584         Modify cost values for conflicts with early clobbers.
13585         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
13587 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
13589         PR target/57098
13590         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
13592 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
13594         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
13595         from/to S register.
13596         (movdi_aarch64): Support LDR/STR from/to D register.
13598 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
13600         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
13601         or higher by default.
13603 2013-04-29  Richard Biener  <rguenther@suse.de>
13605         PR middle-end/57075
13606         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
13607         even if not adding abnormal edges for calls that can make
13608         abnormal gotos.
13610 2013-04-29  Richard Biener  <rguenther@suse.de>
13612         PR middle-end/57103
13613         * tree-cfg.c (move_stmt_op): Fix condition under which to update
13614         TREE_BLOCK.
13615         (move_stmt_r): Remove redundant checking.
13617 2013-04-29  Teresa Johnson  <tejohnson@google.com>
13619         PR bootstrap/57077
13620         * basic-block.h (apply_scale): New function.
13621         (apply_probability): Use apply_scale.
13622         * gimple-streamer-in.c (input_bb): Ditto.
13623         * lto-streamer-in.c (input_cfg): Ditto.
13624         * lto-cgraph.c (merge_profile_summaries): Ditto.
13625         * tree-optimize.c (execute_fixup_cfg): Ditto.
13626         * tree-inline.c (copy_bb): Update comment to use apply_scale.
13627         (copy_edges_for_bb): Ditto.
13628         (copy_cfg_body): Ditto.
13630 2013-04-29  Tom de Vries  <tom@codesourcery.com>
13632         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
13633         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
13634         (tail_merge_optimize): Handle current_loops == NULL.
13636 2013-04-26  Jeff Law  <law@redhat.com>
13638         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
13639         (simplify_cond_using_ranges): Generalize code to simplify
13640         COND_EXPRs where one argument is a constant and the other
13641         is an SSA_NAME created by an integral type conversion.
13643 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13645         * config/arm/arm.md (store_minmaxsi): Use only when
13646         optimize_insn_for_size_p.
13648 2013-04-29  Christian Bruel  <christian.bruel@st.com>
13650         PR target/57108
13651         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
13653 2013-04-29  Richard Biener  <rguenther@suse.de>
13655         PR middle-end/57089
13656         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
13657         loop tree make sure to schedule a fixup for the child as well.
13658         (expand_omp_for_generic): Properly add loops.
13659         (expand_omp_for_static_nochunk): Likewise.
13660         (expand_omp_for_static_chunk): Likewise.
13661         (expand_omp_for): For the degenerate case fixup loops.
13662         (expand_omp_sections): Fix default bb placement in loops.
13663         (expand_omp_atomic_pipeline): Properly add loops.
13665 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13667         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
13669 2013-04-29  Tom de Vries  <tom@codesourcery.com>
13671         * tree-ssa-tail-merge.c: Update header comment.
13673 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13675         * config/aarch64/arm_neon.h
13676         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
13677         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
13678         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
13679         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
13680         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
13681         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
13682         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
13683         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
13685 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13687         * config/aarch64/aarch64-simd.md
13688         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
13689         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
13690         fix_trunc, fixuns_trunc.
13691         (ftrunc<VDQF:mode>2): New.
13692         * config/aarch64/iterators.md (optab): Add fix, fixuns.
13693         (fix_trunc_optab): New.
13695 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13697         * config/aarch64/aarch64-builtins.c
13698         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
13699         iceilf, lround, iroundf.
13701 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
13703         PR target/54349
13704         * config/i386/i386.h (enum ix86_tune_indices)
13705         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
13706         New, split from X86_TUNE_INTER_UNIT_MOVES.
13707         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
13708         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
13709         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
13710         (TARGET_INTER_UNIT_MOVES): Remove.
13711         * config/i386/i386.c (initial_ix86_tune_features): Update.
13712         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
13713         (ix86_expand_convert_uns_didf_sse): Use
13714         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
13715         (ix86_expand_vector_init_one_nonzero): Ditto.
13716         (ix86_expand_vector_init_interleave): Ditto.
13717         (inline_secondary_memory_needed): Return true for moves from SSE class
13718         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
13719         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
13720         * config/i386/constraints.md (Yi, Ym): Depend on
13721         TARGET_INTER_UNIT_MOVES_TO_VEC.
13722         (Yj, Yn): New constraints.
13723         * config/i386/i386.md (*movdi_internal): Change constraints of
13724         operand 1 from Yi to Yj and from Ym to Yn.
13725         (*movsi_internal): Ditto.
13726         (*movdf_internal): Ditto.
13727         (*movsf_internal): Ditto.
13728         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
13729         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
13730         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
13731         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
13732         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
13733         * config/i386/sse.md (movdi_to_sse): Ditto.
13734         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
13735         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
13736         TARGET_INTER_UNIT_MOVES.
13737         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
13738         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
13739         instead of TARGET_INTER_UNIT_MOVES.
13740         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
13741         operand 1 from Yi to Yj and from Ym to Yn.
13743 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13745         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
13746         (float_truncate_hi_): Likewise.
13747         (float_extend_lo_): Likewise.
13748         (float_truncate_lo_): Likewise.
13749         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
13750         (aarch64_float_extend_lo_v2df): Likewise.
13751         (vec_unpacks_hi_v4sf): Likewise.
13752         (aarch64_float_truncate_lo_v2sf): Likewise.
13753         (aarch64_float_truncate_hi_v4sf): Likewise.
13754         (vec_pack_trunc_v2df): Likewise.
13755         (vec_pack_trunc_df): Likewise.
13757 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13759         * config/aarch64/aarch64-builtins.c
13760         (aarch64_fold_builtin): Fold float conversions.
13761         * config/aarch64/aarch64-simd-builtins.def
13762         (floatv2si, floatv4si, floatv2di): New.
13763         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
13764         * config/aarch64/aarch64-simd.md
13765         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
13766         * config/aarch64/iterators.md (FLOATUORS): New.
13767         (optab): Add float, floatuns.
13768         (su_optab): Likewise.
13770 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13772         * config/aarch64/aarch64-builtins.c
13773         (aarch64_builtin_vectorized_function): Use new names for
13774         fcvt builtins.
13775         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
13776         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
13777         (fcvtzu): Split as...
13778         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
13779         (fcvtas): Split as...
13780         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
13781         (fcvtau): Split as...
13782         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
13783         (fcvtps): Split as...
13784         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
13785         (fcvtpu): Split as...
13786         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
13787         (fcvtms): Split as...
13788         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
13789         (fcvtmu): Split as...
13790         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
13791         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
13792         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
13793         (lfrintnusf, lfrintnudf): Likewise.
13794         * config/aarch64/aarch64-simd.md
13795         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
13796         define_insn.
13797         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
13798         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
13799         (fcvt_pattern): Likewise.
13801 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13803         * config/aarch64/aarch64-simd.md
13804         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
13805         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
13807 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13809         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
13810         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
13811         (vrnd<a,m,n,p>_f32): Implement using builtins.
13812         (vrnd<i,x><q>_f<32, 64>): New.
13814 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
13816         * config/aarch64/aarch64-builtins.c
13817         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
13818         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
13819         (frintz): Rename to...
13820         (btrunc): ...this.
13821         (frintp): Rename to...
13822         (ceil): ...this.
13823         (frintm): Rename to...
13824         (floor): ...this.
13825         (frinti): Rename to...
13826         (nearbyint): ...this.
13827         (frintx): Rename to...
13828         (rint): ...this.
13829         (frinta): Rename to...
13830         (round): ...this.
13831         * config/aarch64/aarch64-simd.md
13832         (aarch64_frint<frint_suffix><mode>): Delete.
13833         (<frint_pattern><mode>2): Convert to insn.
13834         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
13835         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
13836         (frint_pattern): Likewise.
13837         (frint_suffix): Likewise.
13839 2013-04-29  Richard Biener  <rguenther@suse.de>
13841         PR tree-optimization/57081
13842         * loop-init.c: Include tree-flow.h.
13843         (loop_optimizer_finalize): Free number of iteration estimates.
13844         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
13846 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
13848         PR tree-optimization/57083
13849         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
13850         non-singleton shift count range, zero extend low_bound for uns case.
13852         * config/i386/predicates.md (general_vector_operand): New predicate.
13853         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
13854         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
13855         if they aren't nonimmediate operands.  If their original values
13856         satisfy const_vector_equal_evenodd_p, don't shift them.
13857         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
13858         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
13859         if not nonimmediate_operand.
13860         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
13861         instead of register_operand.
13862         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
13864 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
13866         * stor-layout.c (finalize_size_functions): Allocate a structure and
13867         reset cfun before dumping the functions.
13869 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
13871         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
13873         PR target/56866
13874         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
13875         use xop_pmacsdqh if uns_p.
13876         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
13877         the immediate rotate count.
13879 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
13881         * rtl.h (struct rtx_def): Add comment for field jump.
13882         (LRA_SUBREG_P): New macro.
13883         * recog.c (register_operand): Check LRA_SUBREG_P.
13884         * lra.c (lra): Add note at the end of RTL code. Align non-empty
13885         stack frame.
13886         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
13887         (lra_final_code_change): Skip subreg change for operators.
13888         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
13889         if there are no operand changes.
13890         * lra-constraints.c (curr_insn_set): New.
13891         (match_reload): Set LRA_SUBREG_P.
13892         (emit_spill_move): Ditto.
13893         (check_and_process_move): Use curr_insn_set. Process only single
13894         set insns.  Don't initialize sec_mem_p and change_p.
13895         (simplify_operand_subreg): Use LRA_SUBREG_P.
13896         (reg_in_class_p): New function.
13897         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
13898         of #ifdef.  Add code to remove cycling.
13899         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
13900         non-null disp.  Reload inner instead of disp when base and index
13901         are null.  Try to put lo_sum into register.
13902         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
13903         (check_and_process_move): Move code for move cost check to
13904         simple_move_p.  Remove equiv_substitution.
13905         (simple_move_p): New function.
13906         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
13907         curr_insn_set.  Call check_and_process_move only for single set
13908         insns.  Use the new function.  Move call of check_and_process_move
13909         after operand equiv substitution and address process.
13911 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
13913         PR go/57045
13914         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
13915         with nonlocal goto receivers or returns twice calls, ignore
13916         unininitialized values from abnormal edges to nl goto receiver
13917         or returns twice call.
13919 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
13921         PR tree-optimization/57051
13922         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
13923         and VEC_RSHIFT_EXPR if shift count is a multiple of element
13924         bitsize.
13926 2013-04-26  Richard Biener  <rguenther@suse.de>
13928         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
13929         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
13930         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
13931         (fixup_loop_arrays_after_move): New function.
13932         (move_sese_region_to_fn): Properly outline the loop tree parts
13933         of the SESE region.
13935 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
13937         * config/i386/i386.md (type, unit): Fix long lines.
13939 2013-04-26  Richard Biener  <rguenther@suse.de>
13941         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
13942         (lto-streamer-out.o): Likewise.
13943         * cfgloop.c (init_loops_structure): Export, add struct function
13944         argument and adjust.
13945         (flow_loops_find): Adjust.
13946         * cfgloop.h (enum loop_estimation): Add EST_LAST.
13947         (init_loops_structure): Declare.
13948         * lto-streamer-in.c: Include cfgloop.h.
13949         (input_cfg): Input the loop tree.
13950         * lto-streamer-out.c: Include cfgloop.h.
13951         (output_cfg): Output the loop tree.
13952         (output_struct_function_base): Do not drop PROP_loops.
13954 2013-03-26  Richard Biener  <rguenther@suse.de>
13956         * tree-cfg.c (execute_build_cfg): Build the loop tree.
13957         (pass_build_cfg): Provide PROP_loops.
13958         (move_sese_region_to_fn): Remove loops that are outlined into fn
13959         for now.
13960         * tree-inline.c: Include cfgloop.h.
13961         (initialize_cfun): Do not drop PROP_loops.
13962         (copy_loops): New function.
13963         (copy_cfg_body): Copy loop structure.
13964         (tree_function_versioning): Initialize destination loop tree.
13965         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
13966         (pass_parallelize_loops): Do IL verification.
13967         * loop-init.c (loop_optimizer_init): Fixup loops if required.
13968         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
13969         the CFG make sure we fixup loops as well.
13970         * tree-ssa-tail-merge.c: Include cfgloop.h.
13971         (replace_block_by): When merging loop latches mark loops for fixup.
13972         * lto-streamer-out.c (output_struct_function_base): Drop
13973         PROP_loops for now.
13974         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
13975         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
13976         * ipa-split.c: Include cfgloop.h.
13977         (split_function): Add the new return block to the loop tree root.
13978         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
13979         whether we have removed the forwarder block.
13980         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
13981         * cfgloop.h (place_new_loop): Declare.
13982         * cfgloopmanip.c (place_new_loop): Export.
13983         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
13984         (tree-switch-conversion.o): Likewise.
13985         (tree-complex.o): Likewise.
13986         (tree-inline.o): Likewise.
13987         (tree-ssa-tailmerge.o): Likewise.
13988         (ipa-split.o): Likewise.
13989         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
13990         (tree-ssa-copy.o): Likewise.
13991         * tree-switch-conversion.c: Include cfgloop.h
13992         (process_switch): If we emit a bit-test cascade, schedule loops
13993         for fixup.
13994         * tree-complex.c: Include cfgloop.h.
13995         (expand_complex_div_wide): Properly add new basic-blocks to loops.
13996         * asan.c: Include cfgloop.h.
13997         (create_cond_insert_point): Properly add new basic-blocks to
13998         loops, schedule loop fixup.
13999         * cfgloop.c (verify_loop_structure): Check that looks are not
14000         marked for fixup.
14001         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
14002         to loops.
14003         (expand_omp_for_generic): Likewise.
14004         (expand_omp_sections): Likewise.
14005         (expand_omp_atomic_pipeline): Schedule loops for fixup.
14006         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
14007         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
14008         is initialized, not when loops are present.
14009         * tree-parloops.c (parallelize_loops): Remove checking here.
14010         * passes.c (init_optimization_passes): Schedule a copy-propagation
14011         pass before complete unrolling of inner loops.
14013 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
14015         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
14016         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
14017         (colorize_init): Add argument to _WIN32 version.
14018         * toplev.c: Include diagnostic-color.h.
14019         (process_options): Default to -fdiagnostics-color=auto if
14020         GCC_COLORS env var is in the environment.
14021         * common.opt (fdiagnostics-color=): Add Var and Init.
14022         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
14023         env var is in the environment, the default is auto rather than never.
14025         * diagnostic.h (file_name_as_prefix): Add context argument.
14026         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
14027         the string as locus.
14028         * langhooks.c (lhd_print_error_function): Adjust caller.
14030 2013-04-25  Lawrence Crowl  <crowl@google.com>
14032         * var-tracking.c (shared_hash_def::htab):
14033         Change type to hash_table.  Update dependent calls and types.
14035 2013-04-25  Lawrence Crowl  <crowl@google.com>
14037         * Makefile.in: Update as needed below.
14039         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
14040         Move declaration to after the type's method definitons.
14042         * attribs.c (htab_t scoped_attributes::attribute_hash):
14043         Change type to hash_table.  Update dependent calls and types.
14045         * bitmap.c (htab_t bitmap_desc_hash):
14046         Change type to hash_table.  Update dependent calls and types.
14048         * cselib.c (htab_t cselib_hash_table):
14049         Change type to hash_table.  Update dependent calls and types.
14051         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
14052         (hash_string_slot_node): Move implementation into lto-streamer.h
14053         struct string_slot_hasher.
14054         (eq_string_slot_node): Likewise.
14056         * data-streamer-out.c: Update output_block::string_hash_table
14057         dependent calls and types.
14059         * dwarf2cfi.c (htab_t trace_index):
14060         Change type to hash_table.  Update dependent calls and types.
14062         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
14063         Change type to hash_table.  Update dependent calls and types.
14064         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
14065         (htab_t optimize_external_refs::map): Likewise.
14066         (htab_t output_comp_unit::extern_map): Likewise.
14067         (htab_t output_comdat_type_unit::extern_map): Likewise.
14068         (htab_t output_macinfo::macinfo_htab): Likewise.
14069         (htab_t optimize_location_lists::htab): Likewise.
14070         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
14072         * except.c (htab_t ehspec_hash_type):
14073         Change type to hash_table.  Update dependent calls and types.
14074         (assign_filter_values::ttypes): Likewise.
14075         (assign_filter_values::ehspec): Likewise.
14076         (sjlj_assign_call_site_values::ar_hash): Likewise.
14077         (convert_to_eh_region_ranges::ar_hash): Likewise.
14079         * gcse.c (htab_t pre_ldst_table):
14080         Change type to hash_table.  Update dependent calls and types.
14082         * ggc-common.c (htab_t saving_htab):
14083         Change type to hash_table.  Update dependent calls and types.
14084         (htab_t loc_hash): Likewise.
14085         (htab_t ptr_hash): Likewise.
14086         (call_count): Rename ggc_call_count.
14087         (call_alloc): Rename ggc_call_alloc.
14088         (loc_descriptor): Rename make_loc_descriptor.
14089         (add_statistics): Rename ggc_add_statistics.
14091         * ggc-common.c (saving_htab):
14092         Change type to hash_table.  Update dependent calls and types.
14094         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
14095         (push_gimplify_context): Likewise.
14096         (pop_gimplify_context): Likewise.
14097         (struct gimple_temp_hash_elt): Added.
14098         (struct gimplify_hasher): Likewise.
14099         (struct gimplify_ctx.temp_htab):
14100         Change type to hash_table.  Update dependent calls and types.
14102         * gimple-fold.c: Include gimplify-ctx.h.
14104         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
14105         Change type to hash_table.  Update dependent calls and types.
14106         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
14107         avoid potential global name collision.
14109         * gimplify.c: Include gimplify-ctx.h.
14110         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
14111         (htab_t gimplify_ctx::temp_htab):
14112         Update dependent calls and types for new type hash_table.
14113         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
14114         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
14116         * gimplify-ctx.h: New.
14117         (struct gimple_temp_hash_elt): Move from gimplify.c.
14118         (class gimplify_hasher): New.
14119         (struct gimplify_ctx): Move from gimple.h.
14120         (htab_t gimplify_ctx::temp_htab):
14121         Change type to hash_table.  Update dependent calls and types.
14123         * graphite-clast-to-gimple.c: Include graphite-htab.h.
14124         (htab_t ivs_params::newivs_index):
14125         Change type to hash_table.  Update dependent calls and types.
14126         (htab_t ivs_params::params_index): Likewise.
14127         (htab_t print_generated_program::params_index): Likewise.
14128         (htab_t gloog::newivs_index): Likewise.
14129         (htab_t gloog::params_index): Likewise.
14131         * graphite.c: Include graphite-htab.h.
14132         4htab_t graphite_transform_loops::bb_pbb_mapping):
14133         Change type to hash_table.  Update dependent calls and types.
14135         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
14136         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
14137         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
14139         * graphite-dependences.c: Include graphite-htab.h.
14140         (loop_is_parallel_p): Change hash table type of parameter.
14142         * graphite-htab.h: New.
14143         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
14144         (extern find_pbb_via_hash): Move from graphite-poly.h.
14145         (extern loop_is_parallel_p): Move from graphite-poly.h.
14146         (extern get_loop_body_pbbs): Move from graphite-poly.h.
14148         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
14149         (extern loop_is_parallel_p): Move to graphite-htab.h.
14150         (extern get_loop_body_pbbs): Move to graphite-htab.h.
14152         * haifa-sched.c (htab_t delay_htab):
14153         Change type to hash_table.  Update dependent calls and types.
14154         (htab_t delay_htab_i2): Likewise.
14156         * ira-color.c (htab_t allocno_hard_regs_htab):
14157         Change type to hash_table.  Update dependent calls and types.
14159         * ira-costs.c (htab_t cost_classes_htab):
14160         Change type to hash_table.  Update dependent calls and types.
14162         * loop-invariant.c (htab_t merge_identical_invariants::eq):
14163         Change type to hash_table.  Update dependent calls and types.
14165         * loop-iv.c (htab_t bivs):
14166         Change type to hash_table.  Update dependent calls and types.
14168         * loop-unroll.c (htab_t opt_info::insns_to_split):
14169         Change type to hash_table.  Update dependent calls and types.
14170         (htab_t opt_info::insns_with_var_to_expand): Likewise.
14172         * lto-streamer.h (struct string_slot): Move from data-streamer.h
14173         (struct string_slot_hasher): New.
14174         (htab_t output_block::string_hash_table):
14175         Change type to hash_table.  Update dependent calls and types.
14177         * lto-streamer-in.c (freeing_string_slot_hasher): New.
14178         (htab_t file_name_hash_table):
14179         Change type to hash_table.  Update dependent calls and types.
14181         * lto-streamer-out.c: Update output_block::string_hash_table dependent
14182         calls and types.
14184         * lto-streamer.c (htab_t tree_htab):
14185         Change type to hash_table.  Update dependent calls and types.
14187         * omp-low.c: Include gimplify-ctx.h.
14189         * passes.c (htab_t name_to_pass_map):
14190         Change type to hash_table.  Update dependent calls and types.
14191         (pass_traverse): Rename to passes_pass_traverse.
14193         * plugin.c (htab_t event_tab):
14194         Change type to hash_table.  Update dependent calls and types.
14196         * postreload-gcse.c (htab_t expr_table):
14197         Change type to hash_table.  Update dependent calls and types.
14198         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
14200         * sese.c (debug_rename_map_1): Make extern.
14201         (htab_t copy_bb_and_scalar_dependences::rename_map):
14202         Change type to hash_table.  Update dependent calls and types.
14204         * sese.h (extern debug_rename_map): Move to .c file.
14206         * store-motion.c (htab_t store_motion_mems_table):
14207         Change type to hash_table.  Update dependent calls and types.
14209         * trans-mem.c (htab_t tm_new_mem_hash):
14210         Change type to hash_table.  Update dependent calls and types.
14212         * tree-browser.c (htab_t TB_up_ht):
14213         Change type to hash_table.  Update dependent calls and types.
14215         * tree-cfg.c (htab_t discriminator_per_locus):
14216         Change type to hash_table.  Update dependent calls and types.
14218         * tree-complex.c: Include tree-hasher.h
14219         (htab_t complex_variable_components):
14220         Change type to hash_table.  Update dependent calls and types.
14222         * tree-eh.c (htab_t finally_tree):
14223         Change type to hash_table.  Update dependent calls and types.
14225         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
14226         struct int_tree_hasher.
14227         (extern int_tree_map_eq): Likewise.
14228         (uid_decl_map_hash): Removed.
14229         (extern decl_tree_map_eq): Likewise.
14231         * tree-hasher.h: New.
14232         (struct int_tree_hasher): New.
14233         (typedef int_tree_htab_type): New.
14235         * tree-inline.c: Include gimplify-ctx.h.
14237         * tree-mudflap.c: Include gimplify-ctx.h.
14239         * tree-parloops.c: Include tree-hasher.h.
14240         (htab_t eliminate_local_variables_stmt::decl_address):
14241         Change type to hash_table.  Update dependent calls and types.
14242         (htab_t separate_decls_in_region::decl_copies): Likewise.
14244         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
14245         Change type to hash_table.  Update dependent calls and types.
14247         * tree-sra.c (candidates):
14248         Change type to hash_table.  Update dependent calls and types.
14250         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
14251         in tree-flow.h.
14252         (int_tree_map_hash): Likewise.
14254         * tree-ssa-dom.c (htab_t avail_exprs):
14255         Change type to hash_table.  Update dependent calls and types.
14257         * tree-ssa-live.c (var_map_base_init::tree_to_index):
14258         Change type to hash_table.  Update dependent calls and types.
14260         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
14261         Change type to hash_table.  Update dependent calls and types.
14263         * tree-ssa-phiopt.c (seen_ssa_names):
14264         Change type to hash_table.  Update dependent calls and types.
14266         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
14267         Change type to hash_table.  Update dependent calls and types.
14269         * tree-ssa-uncprop.c (equiv):
14270         Change type to hash_table.  Update dependent calls and types.
14272 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
14274         PR rtl-optimization/57003
14275         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
14276         call note_stores with kill_clobbered_value callback again after
14277         killing regs_invalidated_by_call.
14279 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
14281         * config/aarch64/aarch64-simd.md
14282         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
14283         (aarch64_simd_bsl<mode>): Likewise.
14284         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
14286 2013-04-25  Marek Polacek  <polacek@redhat.com>
14288         PR tree-optimization/57066
14289         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
14291 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
14293         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
14295 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
14297         * config/aarch64/aarch64-builtins.c
14298         (aarch64_fold_builtin): New.
14299         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
14300         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
14301         * config/aarch64/aarch64-simd-builtins.def (abs): New.
14302         * config/aarch64/arm_neon.h
14303         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
14305 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
14306             Tejas Belagod  <tejas.belagod@arm.com>
14308         * config/aarch64/aarch64-builtins.c
14309         (aarch64_gimple_fold_builtin): New.
14310         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
14311         * config/aarch64/aarch64-simd-builtins.def (addv): New.
14312         * config/aarch64/aarch64-simd.md (addpv4sf): New.
14313         (addvv4sf): Update.
14314         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
14316 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14318         * config/aarch64/aarch64.md
14319         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
14321 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14323         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
14324         (*ngcsi_uxtw): New pattern.
14326 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14327             Julian Brown  <julian@codesourcery.com>
14329         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
14330         (TB_DREG): Add T_V4HF.
14331         (v4hf_UP): New macro.
14332         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
14333         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
14334         Handle initialisation of V4HF. Adjust initialisation of reinterpret
14335         built-ins.
14336         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
14337         (arm_vector_mode_supported_p): Handle V4HF.
14338         (arm_mangle_map): Handle V4HFmode.
14339         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
14340         * config/arm/arm_neon_builtins.def: Add entries for
14341         vcvtv4hfv4sf, vcvtv4sfv4hf.
14342         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
14343         (neon_vcvtv4hfv4sf): Likewise.
14344         * config/arm/neon-gen.ml: Handle half-precision floating point
14345         features.
14346         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
14347         * config/arm/arm_neon.h: Regenerate.
14348         * config/arm/neon.ml (type elts): Add F16.
14349         (type vectype): Add T_float16x4, T_floatHF.
14350         (type vecmode): Add V4HF.
14351         (type features): Add Requires_FP_bit feature.
14352         (elt_width): Handle F16.
14353         (elt_class): Likewise.
14354         (elt_of_class_width): Likewise.
14355         (mode_of_elt): Refactor.
14356         (type_for_elt): Handle F16, fix error messages.
14357         (vectype_size): Handle T_float16x4.
14358         (vcvt_sh): New function.
14359         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
14360         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
14361         (string_of_mode): Handle V4HF.
14362         * doc/arm-neon-intrinsics.texi: Regenerate.
14364 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
14366         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
14367         format specifier in 'X' case.
14369 2013-04-25  Alan Modra  <amodra@gmail.com>
14371         PR target/57052
14372         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
14373         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
14374         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
14375         Repeat for many other rotate/shift and mask patterns using subregs.
14376         Name lshiftrt insns.
14377         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
14378         on WORDS_BIG_ENDIAN.
14380 2013-04-25  Alan Modra  <amodra@gmail.com>
14382         * config.gcc: Support little-endian powerpc-linux targets.
14383         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
14384         (LINK_OS_LINUX_SPEC): Define.
14385         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
14386         Preserve MASK_LITTLE_ENDIAN.
14387         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
14388         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
14389         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
14390         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
14391         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
14392         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
14393         Correct fp word order for little-endian.  Don't shift toc entries
14394         smaller than a word for little-endian.
14395         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
14396         (bswapdi2 splits): Correct low-part subreg for little-endian.
14397         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
14398         low/high where such is correct only for be.
14399         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
14400         little-endian for -mcall-aixdesc.
14402 2013-04-25  Alan Modra  <amodra@gmail.com>
14404         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
14405         replace_equiv_address_nv.
14407 2013-04-25  Alan Modra  <amodra@gmail.com>
14409         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
14411 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
14413         Revert:
14414         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
14415         * rtl.h (struct rtx_def): ...
14417 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
14419         PR rtl-optimizations/57046
14420         * lra-constraints (split_reg): Set up lra_risky_transformations_p
14421         for multi-reg splits.
14423 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
14425         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
14427 2013-04-24  Sterling Augustine  <saugustine@google.com>
14429         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
14430         (comp_dir_string, debug_str_dwo_section): New.
14431         (DEBUG_STR_DWO_SECTION): Rename to ...
14432         (DEBUG_DWO_STR_SECTION): ... this.
14433         (DEBUG_NORM_STR_SECTION): Delete.
14434         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
14435         (DEBUG_STR_DWO_SECTION_FLAGS): New.
14436         (find_AT_string): Move most logic to ...
14437         (find_AT_string_in_table): ... here.  New.
14438         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
14439         add_skeleton_AT_string.  Delete logic.
14440         (output_skeleton_debug_sections): Remove call to
14441         add_top_level_skeleton_die_attrs.
14442         (add_comp_dir_attribute): Move logic to comp_dir_string.
14443         (dwarf2out_init): Initialize debug_str_dwo_section.
14444         (output_indirect_string): Call find_string_form.
14445         (output_indirect_strings): Rewrite.
14446         (prune_unused_types): Empty skeleton_debug_str_hash.
14447         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
14448         (dwarf2out_finish):  Call output_indirect_strings.
14450 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
14452         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
14454 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
14456         * rtl.h (struct rtx_def): Add comment for field jump.
14457         (LRA_SUBREG_P): New macro.
14458         * recog.c (register_operand): Check LRA_SUBREG_P.
14459         * lra.c (lra): Add note at the end of RTL code. Align non-empty
14460         stack frame.
14461         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
14462         (lra_final_code_change): Skip subreg change for operators.
14463         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
14464         if there are no operand changes.
14465         * lra-constraints.c (curr_insn_set): New.
14466         (match_reload): Set LRA_SUBREG_P.
14467         (emit_spill_move): Ditto.
14468         (check_and_process_move): Use curr_insn_set. Process only single
14469         set insns.  Don't initialize sec_mem_p and change_p.
14470         (simplify_operand_subreg): Use LRA_SUBREG_P.
14471         (reg_in_class_p): New function.
14472         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
14473         of #ifdef.  Add code to remove cycling.
14474         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
14475         non-null disp.  Reload inner instead of disp when base and index
14476         are null.  Try to put lo_sum into register.
14477         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
14478         (check_and_process_move): Move code for move cost check to
14479         simple_move_p.  Remove equiv_substitution.
14480         (simple_move_p): New function.
14481         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
14482         curr_insn_set.  Call check_and_process_move only for single set
14483         insns.  Use the new function.  Move call of check_and_process_move
14484         after operand equiv substitution and address process.
14486 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
14488         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
14489         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
14490         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
14492 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
14494         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
14496 2013-04-24  Marek Polacek  <polacek@redhat.com>
14498         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
14499         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
14500         (select_loops_exit_conditions): Likewise.
14501         (number_of_iterations_for_all_loops): Likewise.
14502         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
14503         (scev_analysis): Likewise.
14505 2013-04-02  Catherine Moore  <clm@codesourcery.com>
14506             Chao-ying Fu  <fu@mips.com>
14508         * config/mips/micromips.md (jraddiusp): New pattern.
14509         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
14510         instruction if possible.
14512 2013-04-24  Alan Modra  <amodra@gmail.com>
14514         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
14516 2013-04-24  Julian Brown  <julian@codesourcery.com>
14517             Chung-Lin Tang  <cltang@codesourcery.com>
14519         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
14520         dependency behavior in enumeration type DIE generation. Add TODO note
14521         to comments about future DW_FORM_sdata/udata re-work of related code.
14523 2013-04-23  Lawrence Crowl  <crowl@google.com>
14525         * Makefile.in: Update as needed below.
14527         * hash-table.h (class hash_table):
14528         Correct many methods with parameter types compare_type to the correct
14529         value_type.  (Correct code was unlikely to notice the change.)
14530         (hash_table::elements_with_deleted) New.
14531         (class hashtable::iterator): New.
14532         (hashtable::begin()): New.
14533         (hashtable::end()): New.
14534         (FOR_EACH_HASH_TABLE_ELEMENT): New.
14536         * statistics.c (statistics_hashes):
14537         Change type to hash_table.  Update dependent calls and types.
14539         * tree-into-ssa.c (var_infos):
14540         Change type to hash_table.  Update dependent calls and types.
14542         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
14543         Change type to hash_table.  Update dependent calls and types.
14545         * tree-ssa-loop-im.c (struct mem_ref.refs):
14546         Change type to hash_table.  Update dependent calls and types.
14548         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
14549         Change type to hash_table.  Update dependent calls and types.
14551         * tree-ssa-sccvn.c (vn_tables_s::nary):
14552         Change type to hash_table.  Update dependent calls and types.
14553         (vn_tables_s::phis): Likewise.
14554         (vn_tables_s::references): Likewise.
14556         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
14557         (vn_reference_eq): Update parameter and return types.
14559         * tree-ssa-structalias.c (pointer_equiv_class_table):
14560         Change type to hash_table.  Update dependent calls and types.
14561         (location_equiv_class_table): Likewise.
14563         * tree-vect-data-refs.c: Consequential changes for making
14564         peeling a hash_table.
14566         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
14567         (destroy_loop_vec_info): Dependent hash_table update.
14569         * tree-vectorizer.h (peeling_htab):
14570         Change type to hash_table.  Update dependent calls and types.
14572 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
14574         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
14575         to check the register content is equal or not.
14576         * lra-constraints.c (match_reload): Use lra_assign_reg_val
14577         to assign register content record.
14578         * lra-eliminations.c (update_reg_eliminate): Use
14579         lra_update_reg_val_offset to update register content offset.
14580         * lra-int.h (struct lra_reg): Add offset member.
14581         (lra_reg_val_equal_p): New static inline function.
14582         (lra_update_reg_val_offset): New static inline function.
14583         (lra_assign_reg_val): New static inline function.
14584         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
14585         to assign register content record.
14586         (initialize_lra_reg_info_element): Initial offset to zero.
14588 2013-04-23  Catherine Moore  <clm@codesourcery.com>
14590         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
14591         operands.  Record compression.
14593 2013-04-23  Xinliang David Li  <davidxl@google.com>
14595         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
14597 2013-04-23  Richard Biener  <rguenther@suse.de>
14599         PR middle-end/57036
14600         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
14601         parameter, only add abnormal goto edges from the copied body
14602         if the call could perform abnormal gotos.
14603         (copy_cfg_body): Adjust.
14605 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
14607         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
14609 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
14611         * coretypes.h (gimple_stmt_iterator): Add struct to make
14612         compatible with C.
14614 2013-04-23  Richard Biener  <rguenther@suse.de>
14616         PR tree-optimization/57026
14617         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
14618         from SSA names occuring in abnormal PHI nodes.
14620 2013-04-22  Andi Kleen  <ak@linux.intel.com>
14622         * lto/lto.c (print_lto_report_1): Fix LTO report names.
14624 2013-04-22  Andi Kleen  <ak@linux.intel.com>
14626         * lto/lto.c (print_lto_report_1): Declare early.
14627         (read_cgraph_and_symbols): Call print_lto_report_1 early.
14629 2013-04-22  Andi Kleen  <ak@linux.intel.com>
14631         * common.opt (-flto-report-wpa): Add.
14632         * doc/invoke.texi (-flto-report-wpa): Add.
14633         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
14634         (lto_main): dito.
14636 2013-04-22  Xinliang David Li  <davidxl@google.com>
14638         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
14639         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
14640         * Makefile.in: New dependency
14642             David Daney <ddaney.cavm@gmail.com>
14644         * configure.ac (gcc_cv_as_micromips_support):  Use the
14645         --fatal-warnings option.
14646         * configure: Regenerate.
14648 2013-04-22  Marek Polacek  <polacek@redhat.com>
14650         PR sanitizer/56990
14651         * tsan.c (instrument_expr): Don't instrument expression
14652         in case its size is zero.
14654 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
14656         PR target/57032
14657         Revert:
14658         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
14660         * config/alpha/alpha.c (TARGET_LRA_P): New define.
14662 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
14664         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
14665         (gimple_stmt_iterator): New typedef.
14666         * gimple.h (gimple_stmt_iterator): Rename to...
14667         (gimple_stmt_iterator_d): ... This.
14668         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
14669         trees be valid for GIMPLE and GENERIC.
14670         (TARGET_GIMPLE_FOLD_BUILTIN): New.
14671         * gimple-fold.c (gimple_fold_call): Call target hook
14672         gimple_fold_builtin.
14673         * hooks.c (hook_bool_gsiptr_false): New.
14674         * hooks.h (hook_bool_gsiptr_false): New.
14675         * target.def (fold_stmt): New.
14676         * doc/tm.texi: Regenerate.
14678 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
14680         PR target/57018
14681         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
14682         a set sp if no stack realignment.
14684 2013-04-22  Nick Clifton  <nickc@redhat.com>
14686         * config.gcc (tilegx-linux): Extend extra_objs rather than
14687         overwriting it.
14688         (tilepro-linux): Likewise.
14690 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
14692         * config/aarch64/aarch64-builtins.c
14693         (CF): Remove.
14694         (CF0, CF1, CF2, CF3, CF4, CF10): New.
14695         (VAR<1-12>): Add MAP parameter.
14696         (BUILTIN_*): Likewise.
14697         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
14698         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
14699         (aarch64_ushl_n<mode>): Likewise.
14700         (aarch64_sshr_n<mode>): Likewise.
14701         (aarch64_ushr_n<mode>): Likewise.
14702         (aarch64_<maxmin><mode>): Likewise.
14703         (aarch64_sqrt<mode>): Likewise.
14704         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
14705         (vshr<q>_n_*): Likewise.
14707 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
14709         * config/aarch64/aarch64-builtins.c
14710         (aarch64_simd_builtin_type_mode): Handle SF types.
14711         (sf_UP): Define.
14712         (BUILTIN_GPF): Define.
14713         (aarch64_init_simd_builtins): Handle SF types.
14714         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
14715         (frecps): Likewise.
14716         (frecpx): Likewise.
14717         * config/aarch64/aarch64-simd.md
14718         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
14719         (aarch64_frecpe<mode>): New.
14720         (aarch64_frecps<mode>): Likewise.
14721         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
14722         (v8type): Add frecp<esx>.
14723         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
14724         (aarch64_frecps<mode>): Likewise.
14725         * config/aarch64/iterators.md (FRECP): New.
14726         (frecp_suffix): Likewise.
14727         * config/aarch64/arm_neon.h
14728         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
14730 2013-04-22  Christian Bruel  <christian.bruel@st.com>
14732         PR target/56995
14733         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
14734         (REG_CLASS_NAMES): Idem.
14735         (REG_CLASS_CONTENTS): Idem.
14736         (REGCLASS_HAS_FP_REG): Idem.
14737         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
14738         (sh_conditional_register_usage): Idem.
14740 2013-04-21  Jeff Law  <law@redhat.com>
14742         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
14743         (ssa_forward_propagate_and_combine): Use it.
14745 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
14747         * lra.c: Update the flow chart diagram.
14749 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
14751         PR rtl-optimization/56847
14752         * lra-constraints.c (process_alt_operands): Discourage alternative
14753         with non-matche doffsettable memory constraint fro memory with
14754         known offset.
14756 2013-04-19  Richard Biener  <rguenther@suse.de>
14758         PR tree-optimization/56982
14759         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
14760         function.
14761         * gimplify.c (gimplify_call_expr): Notice special calls.
14762         (gimplify_modify_expr): Likewise.
14763         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
14764         abnormal control flow receivers.
14765         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
14766         in the same way as cfun->has_nonlocal_labels.
14767         (gimple_purge_dead_abnormal_call_edges): Likewise.
14768         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
14769         receivers start a basic-block.
14771 2013-04-19  Richard Biener  <rguenther@suse.de>
14773         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
14774         member ...
14775         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
14776         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
14777         (SLP_TREE_LOAD_PERMUTATION): Add.
14778         (vect_transform_slp_perm_load): Adjust prototype.
14779         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
14780         (vect_free_slp_instance): Likewise.
14781         (vect_create_new_slp_node): Likewise.
14782         (vect_supported_slp_permutation_p): Remove.
14783         (vect_slp_rearrange_stmts): Adjust.
14784         (vect_supported_load_permutation_p): Likewise.  Inline
14785         vect_supported_slp_permutation_p here.
14786         (vect_analyze_slp_instance): Compute load permutations per
14787         slp node instead of per instance.
14788         (vect_get_slp_defs): Adjust.
14789         (vect_transform_slp_perm_load): Likewise.
14790         (vect_schedule_slp_instance): Remove redundant code.
14791         (vect_schedule_slp): Remove hack for PR56270, add it ...
14792         * tree-vect-stmts.c (vectorizable_load): ... here, do not
14793         CSE loads for SLP.  Adjust.
14795 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
14797         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
14798         spelling in two comments.
14800 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
14802         PR target/56797
14803         * config/arm/arm.c (load_multiple_sequence): Require SP
14804         as base register for loads if SP is in the register list.
14806 2013-04-19  Martin Jambor  <mjambor@suse.cz>
14808         PR tree-optimization/56718
14809         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
14810         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
14811         and made public.  Adjusted all callers.
14812         (ipa_intraprocedural_devirtualization): New function.
14813         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
14814         (ipa_intraprocedural_devirtualization): Likewise.
14815         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
14817 2013-04-19  Richard Biener  <rguenther@suse.de>
14819         PR tree-optimization/57000
14820         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
14822 2013-04-19  Terry Guo  <terry.guo@arm.com>
14824         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
14825         Replace with ...
14826         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
14827         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
14828         (cortex_m4_fmacs): Use new reservations.
14829         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
14831 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
14833         PR rtl-optimization/56999
14834         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
14835         related code.
14836         (lra_coalesce): Remove split_origin_bitmap and related code.
14837         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
14838         ranges if necessary.
14840 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
14842         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
14843         New array.
14844         (ix86_expand_call): Remove clobbered_registers array and use
14845         x86_64_ms_sysv_extra_clobbered_registers instead.
14846         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
14847         Declare here.
14848         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
14849         predicate.
14850         * config/i386/i386.md (*call_rex64_ms_sysv): Use
14851         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
14852         (*call_value_rex64_ms_sysv): Ditto.
14854 2013-04-18  Cary Coutant  <ccoutant@google.com>
14856         * dwarf2out.c (output_pubnames): Check die_perennial_p of
14857         parent instead of die_mark.
14859 2013-04-18  Diego Novillo  <dnovillo@google.com>
14861         * gimple.c (create_gimple_tmp): New.
14862         (get_expr_type): New.
14863         (build_assign): New.
14864         (build_type_cast): New.
14865         * gimple.h (enum ssa_mode): Define.
14866         (gimple_seq_set_location): New.
14867         * asan.c (build_check_stmt): Change some gimple_build_* calls
14868         to use build_assign and build_type_cast.
14870 2013-04-18  Richard Biener  <rguenther@suse.de>
14872         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
14873         handle negative step.  Remove redundant checks.
14874         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
14875         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
14876         for negative step and grouped loads fail to vectorize.
14878 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
14880         * emit-rtl.c (reset_insn_used_flags): New function.
14881         (reset_all_used_flags): Use it.
14882         (verify_insn_sharing): New function.
14883         (verify_rtl_sharing): Fix verification for SEQUENCEs.
14885 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
14887         PR tree-optimization/56984
14888         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
14889         and (x >> M) >= N don't register any assertion if N << M is the
14890         minimum value.
14892 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
14894         * lower-subreg.c (resolve_simple_move): If called self-recursive,
14895         do not delete_insn insns that have not yet been emitted, only
14896         unlink them with remove_insn.
14897         * df-scan.c (df_insn_delete): Revert r197492.
14899 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
14901         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
14902         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
14904 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
14906         * config/arm/arm.md (movsicc_insn): Convert define_insn into
14907         define_insn_and_split.
14908         (and_scc,ior_scc,negscc): Likewise.
14909         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
14911 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
14913         * config/arm/arm.c (use_return_insn): Return 0 for targets that
14914         can benefit from using a sequence of LDRD instructions in epilogue
14915         instead of a single LDM instruction.
14917 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
14919         PR 45688
14920         * doc/extend.texi: Fix typo.
14922 2013-04-17  Richard Biener  <rguenther@suse.de>
14924         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
14925         (vect_build_slp_tree): ... here.
14926         (vect_build_slp_tree_1): Compute which stmts of the SLP group
14927         match.  Remove special-casing of mismatched complex loads.
14928         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
14929         re-try the match with swapped commutative operands.
14930         (vect_supported_load_permutation_p): Remove special-casing of
14931         mismatched complex loads.
14932         (vect_analyze_slp_instance): Adjust.
14934 2013-04-17  Richard Biener  <rguenther@suse.de>
14936         PR rtl-optimization/56921
14937         * cfgloop.h (struct loop): Add simple_loop_desc member.
14938         (struct niter_desc): Mark with GTY(()).
14939         (simple_loop_desc): Do not use aux field but simple_loop_desc.
14940         * loop-iv.c (get_simple_loop_desc): Likewise.
14941         (free_simple_loop_desc): Likewise.
14943         Revert
14944         2013-04-16  Richard Biener  <rguenther@suse.de>
14946         PR rtl-optimization/56921
14947         * loop-init.c (pass_rtl_move_loop_invariants): Add
14948         TODO_do_not_ggc_collect to todo_flags_finish.
14949         (pass_rtl_unswitch): Same.
14950         (pass_rtl_unroll_and_peel_loops): Same.
14951         (pass_rtl_doloop): Same.
14953 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
14955         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
14956         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
14957         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
14958         references.
14959         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
14960         * tree-streamer.c (record_common_node): Adjust reference in comment.
14962 2013-04-17  Terry Guo  <terry.guo@arm.com>
14964         * config/arm/cortex-m4.md: Add a new bypass.
14966 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14968         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
14969         New pattern.
14970         (*subs_<optab><mode>_multp2): New pattern.
14971         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
14972         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
14974 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14976         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
14977         (*subs_mul_imm_<mode>): New pattern.
14979 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
14981         PR target/56948
14982         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
14983         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
14984         (vsx_movti_32bit): Change j->wa to O->wa.
14986 2013-04-16  Richard Biener  <rguenther@suse.de>
14988         PR rtl-optimization/56921
14989         * loop-init.c (pass_rtl_move_loop_invariants): Add
14990         TODO_do_not_ggc_collect to todo_flags_finish.
14991         (pass_rtl_unswitch): Same.
14992         (pass_rtl_unroll_and_peel_loops): Same.
14993         (pass_rtl_doloop): Same.
14995 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
14997         * config/arm/arm.c (emit_multi_reg_push): New declaration
14998         for an existing function.
14999         (arm_emit_strd_push): New function.
15000         (arm_expand_prologue): Used here.
15001         (arm_emit_ldrd_pop): New function.
15002         (arm_expand_epilogue): Used here.
15003         (arm_get_frame_offsets): Update condition.
15004         (arm_emit_multi_reg_pop): Add a special case for load of a single
15005         register with writeback.
15007 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
15009         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
15010         description.
15012 2013-04-16  Richard Biener  <rguenther@suse.de>
15014         PR tree-optimization/56756
15015         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
15016         (first_mem_ref_loc): New.
15017         (execute_sm): Place the load temporarily before a previous
15018         access instead of in the latch edge to ensure its SSA dependencies
15019         are defined at points dominating the load.
15021 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
15023         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
15024         correct fix by moving header and footer insn to the footer of
15025         the merged basic block.  Clear BB_END of the merged-away block.
15027         PR middle-end/43631
15028         * emit-rtl.c (make_note_raw): New function.
15029         (link_insn_into_chain): New static inline function.
15030         (add_insn): Use it.
15031         (add_insn_before, add_insn_after): Factor insn chain linking code...
15032         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
15033         using link_insn_into_chain.
15034         (note_outside_basic_block_p): New helper function for emit_note_after
15035         and emit_note_before.
15036         (emit_note_after): Use nobb variant of add_insn_after if the note
15037         should not be contained in a basic block.
15038         (emit_note_before): Use nobb variant of add_insn_before if the note
15039         should not be contained in a basic block.
15040         (emit_note_copy): Use make_note_raw.
15041         (emit_note): Likewise.
15042         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
15043         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
15044         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
15045         the moved barrier the tail of the basic block it follows.
15046         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
15048 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
15050         PR tree-optimization/56962
15051         * gimple-ssa-strength-reduction.c (record_increment): Only set
15052         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
15053         either rhs1 or rhs2 is equal to c->base_expr.
15055 2013-04-15  Richard Biener  <rguenther@suse.de>
15057         PR tree-optimization/56933
15058         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
15059         member.
15060         (GROUP_READ_WRITE_DEPENDENCE): Remove.
15061         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
15062         * tree-vect-data-refs.c (vect_analyze_group_access): Move
15063         dependence check ...
15064         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
15065         ... here.
15066         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
15067         GROUP_READ_WRITE_DEPENDENCE.
15069 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15071         * emit-rtl.c (reset_all_used_flags): New function.
15072         (verify_rtl_sharing): Call reset_all_used_flags before and after
15073         performing the checks.
15075 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15077         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
15078         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
15079         * config/arm/constraints.md (De): New constraint.
15080         * config/arm/neon.md (anddi3_neon): Delete.
15081         (neon_vand<mode>): Expand to standard anddi3 pattern.
15082         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
15083         Move earlier in the file.
15084         (neon_inv_logic_op2): Likewise.
15085         (arm_anddi_operand_neon): New predicate.
15087 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
15089         * configure.ac (gcc_cv_ld_as_needed): Set
15090         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
15091         Use -z ignore, -z record on *-*-solaris2*.
15092         (HAVE_LD_AS_NEEDED): Update comment.
15093         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
15094         * configure: Regenerate.
15095         * config.in: Regenerate.
15096         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
15097         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
15098         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
15099         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
15100         equivalents.  Fix markup.
15101         * doc/tm.texi: Regenerate.
15103 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
15105         * config/i386/i386.opt: New option mstack-protector-guard=.
15106         * config/i386/i386-opts.h: Add enum stack_protector_guard.
15107         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
15108         TARGET_SSP_TLS_GUARD.
15109         * config/i386/i386.c (ix86_option_override_internal): Set
15110         ix86_stack_protector_guard.
15111         * config/i386/i386.md (stack_protect_set): Enable for
15112         TARGET_SSP_TLS_GUARD only.
15113         (stack_protect_set_<mode>): Ditto.
15114         (stack_protect_test): Ditto.
15115         (stack_protect_test_<mode>): Ditto.
15116         * doc/invoke.texi (i386 Option): Document.
15118 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
15120         PR target/56890
15121         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
15122         (S_MODES): Set H_MODE bit.
15123         (SF_MODES): Set only S_MODE and SF_MODE bits.
15124         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
15125         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
15126         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
15127         <MODE_FLOAT>: Likewise.
15129 2013-04-15  Joey Ye  <joey.ye@arm.com>
15131         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
15133 2013-04-15  Joey Ye  <joey.ye@arm.com>
15135         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
15136         for real far jump.
15137         (thumb_far_jump_used_p): Count instruction size and set
15138         far_jump_used.
15140 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
15142         * reorg.c (fill_simple_delay_slots): Reindent block of code.
15143         * resource.c (mark_target_live_regs): Reformat conditional block.
15145 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
15147         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
15148         notes, they are emitted only just before final.
15149         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
15151 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
15153         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
15154         * cfgrtl.c (delete_insn): Call it here instead.
15155         * lra-spills.c (lra_final_code_change): Use delete_insn.
15156         * haifa-sched.c (sched_remove_insn): Likewise.
15157         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
15158         returning to the nop pool.
15159         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
15160         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
15162 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
15164         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
15165         * doc/tm.texi: Regenerated.
15167 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
15169         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
15170         QImode checks.
15172 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
15174         * df-core.c (df_find_def): Compare register numbers.
15175         (df_find_use): Likewise.
15177 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
15179         PR target/56903
15180         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
15181         lra_in_progress for return.
15183 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
15185         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
15186         define_insn into define_insn_and_split and emit movsicc patterns.
15188 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
15190         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
15192 2013-04-12  Richard Biener  <rguenther@suse.de>
15194         * tree-pass.h (TODO_do_not_ggc_collect): New.
15195         * passes.c (execute_one_ipa_transform_pass): Honor
15196         TODO_do_not_ggc_collect.
15197         (execute_one_pass): Likewise.
15199         Revert
15200         2013-04-10  Richard Biener  <rguenther@suse.de>
15202         * passes.c (init_optimization_passes): Remove reload pass.
15203         * ira.c (do_reload): Merge into ...
15204         (ira): ... this.
15205         (rest_of_handle_reload): Remove.
15206         (pass_reload): Likewise.
15207         * config/i386/i386.c (ix86_option_override): Refer to ira instead
15208         of reload for vzeroupper pass placement.
15210 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
15212         PR tree-optimization/56918
15213         PR tree-optimization/56920
15214         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
15215         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
15216         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
15217         use wide_mul_with_sign method.
15219 2013-04-12  Richard Biener  <rguenther@suse.de>
15221         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
15222         not be considered a gimple constant.
15224 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
15226         * fold-const.c (const_binop): Handle vector shifts by a scalar.
15227         (fold_binary_loc): Call const_binop also for mixed vector-scalar
15228         operations.
15230 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
15231             Jakub Jelinek  <jakub@redhat.com>
15233         * opts.c: Include diagnostic-color.h.
15234         (common_handle_option): Handle OPT_fdiagnostics_color_.
15235         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
15236         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
15237         (diagnostic-color.o): New.
15238         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
15239         (diagnostic_color_rule): New enum.
15240         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
15241         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
15242         the location string.
15243         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
15244         either NULL, or color kind.
15245         * diagnostic-color.c: New file.
15246         * diagnostic-color.h: New file.
15247         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
15248         arguments.
15249         * doc/invoke.texi (-fdiagnostics-color): Document.
15250         * pretty-print.h (pp_show_color): Define.
15251         (struct pretty_print_info): Add show_color field.
15252         * diagnostic.c: Include diagnostic-color.h.
15253         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
15254         macros.  Colorize error:, warning: etc. strings and also the location
15255         string.
15256         (diagnostic_show_locus): Colorize the caret line.
15257         * pretty-print.c: Include diagnostic-color.h.
15258         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
15259         inside of %< %> quotes or quoted through q format modifier.
15261 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15263         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
15265 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
15267         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
15268         code in CC_NZ mode.
15269         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
15270         pattern.
15272 2013-04-11  Marek Polacek  <polacek@redhat.com>
15274         PR tree-optimization/48184
15275         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
15277 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
15279         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
15280         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
15281         (skip_simple_arithmetic): Tidy up.
15282         * tree.h (skip_simple_constant_arithmetic): Declare.
15284 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
15286         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
15288 2013-04-11  Richard Biener  <rguenther@suse.de>
15290         * tree-vect-loop.c (get_initial_def_for_induction): Properly
15291         generate vector constants.
15293 2013-04-11  Richard Biener  <rguenther@suse.de>
15295         PR tree-optimization/56878
15296         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
15297         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
15298         New function.
15299         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
15300         Prefer to align the DR with the most invariant base address.
15302 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
15304         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
15305         comment.
15307 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
15309         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
15310         floating-point vector comparisons against 0.
15312 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
15314         PR tree-optimization/56899
15315         * fold-const.c (extract_muldiv_1): Apply distributive law
15316         only if TYPE_OVERFLOW_WRAPS (ctype).
15318 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
15320         PR target/56124
15321         * ira-costs.c (scan_one_insn): Check whether the source rtx of
15322         loading has side effect.
15324 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
15326         * config/sparc/sparc.c: Include tree-pass.h.
15327         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
15328         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
15329         head of file.  Change return type.  Split off gate function.
15330         (sparc_gate_work_around_errata): New function.
15331         (pass_work_around_errata): New pass definition.
15332         (insert_pass_work_around_errata) New pass insert definition to
15333         insert pass_work_around_errata just after delayed-branch scheduling.
15334         (sparc_option_override): Insert the pass.
15335         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
15337 2013-04-10  David S. Miller  <davem@davemloft.net>
15339         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
15340         or -mcpu=hypersparc.
15342         * target.def (cstore_mode): New hook.
15343         * target.h: Include insn-codes.h
15344         * targhooks.c: Likewise.
15345         (default_cstore_mode): New function.
15346         * targhooks.h: Declare it.
15347         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
15348         * doc/tm.texi: Rebuild.
15349         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
15350         target hook, rather than inspecting the insn_data.
15351         * config/sparc/sparc.c (sparc_cstore_mode): New function.
15352         (TARGET_CSTORE_MODE): Redefine.
15353         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
15354         result patterns.
15355         * config/sparc/predicates.md (cstore_result_operand): New special
15356         predicate.
15357         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
15358         Use it for operand 0.
15359         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
15360         (*snesi_special): Likewise.
15361         (*snesi_zero): Likewise.
15362         (*seqsi_zero): Likewise.
15363         (*sltu_insn): Likewise.
15364         (*sgeu_insn): Likewise.
15365         (*seqdi_special): Make operand 0 and comparison operation be of
15366         DImode.
15367         (*snedi_special): Likewise.
15368         (*snedi_special_vis3): Likewise.
15369         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
15370         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
15371         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
15372         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
15373         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
15374         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
15375         (*sltu_extend_sp64): Likewise.
15376         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
15377         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
15378         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
15379         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
15380         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
15382 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
15384         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
15385         (aarch64_start_file): Use the new function.
15387 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
15388             Jason Merrill  <jason@redhat.com>
15390         * common.opt: Add -gdwarf.
15391         * opts.c (common_handle_option): Handle it.
15392         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
15394 2013-04-10  Richard Biener  <rguenther@suse.de>
15396         * passes.c (execute_todo): Do not call ggc_collect conditional here.
15397         (execute_one_ipa_transform_pass): But unconditionally here.
15398         (execute_one_pass): And here.
15399         (init_optimization_passes): Remove reload pass.
15400         * tree-pass.h (TODO_ggc_collect): Remove.
15401         (pass_reload): Likewise.
15402         * ira.c (do_reload): Merge into ...
15403         (ira): ... this.
15404         (rest_of_handle_reload): Remove.
15405         (pass_reload): Likewise.
15406         * config/i386/i386.c (ix86_option_override): Refer to ira instead
15407         of reload for vzeroupper pass placement.
15408         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
15409         and todo_flags_finish of all passes.
15411 2013-04-10  Richard Biener  <rguenther@suse.de>
15413         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
15414         first_const_oprnd field, rename first_def_type to first_op_type.
15415         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
15416         (vect_get_and_check_slp_defs): Always use the type of the
15417         operand.  Allow mixed vect_external_def, vect_constant_def types.
15418         (vect_get_constant_vectors): Handle mixed vect_external_def,
15419         vect_constant_def types.
15421 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
15423         PR tree-optimization/55524
15424         * tree-ssa-math-opts.c
15425         (convert_mult_to_fma): Don't use an fms construct
15426         when we don't have an fms operation, but fnma, and it looks
15427         likely that we'll be able to use the latter.
15429 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
15431         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
15432         function.
15433         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
15434         inline fail caused by overwritable functions.
15436 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
15438         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
15439         unnecessary bits in the constant power of two case.
15441 2013-04-10  Richard Biener  <rguenther@suse.de>
15443         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
15444         broken code swapping operands.
15445         (vect_build_slp_tree): Do not compute load permutations here.
15446         (vect_analyze_slp_instance): Compute load permutations here,
15447         after building the SLP tree.
15449 2013-04-09  Christian Bruel  <christian.bruel@st.com>
15451         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
15452         of next/prev_real_insn.
15454 2013-04-09  Jan Hubicka  <jh@suse.cz>
15456         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
15457         Drop aliased parameter.
15458         (function_and_variable_visibility): Do not handle alias pairs.
15459         * cgraph.c (varpool_externally_visible_p): Update prototype.
15460         * varpool.c (varpool_add_new_variable): Update.
15462 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15464         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
15466 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
15468         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
15470         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
15472 2013-04-09  Marek Polacek  <polacek@redhat.com>
15474         PR tree-optimization/48762
15475         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
15477 2013-04-09  Richard Biener  <rguenther@suse.de>
15479         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
15480         dealing with cost.
15481         (vect_build_slp_tree): Likewise.
15482         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
15483         calculating the cost of a SLP instance.
15484         (vect_analyze_slp_instance): Use it from here, after building
15485         the SLP tree.
15487 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
15489         PR middle-end/56883
15490         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
15491         expand_omp_for_static_chunk): Use simple_p = true in
15492         force_gimple_operand_gsi calls when assigning to addressable decls.
15494 2013-04-09  Jeff Law  <law@redhat.com>
15496         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
15497         when the boolean was created by converting a wider object which
15498         had a boolean range.
15500 2013-04-09  Richard Biener  <rguenther@suse.de>
15502         * tree-vectorizer.h (slp_void_p): Remove.
15503         (slp_tree): Typedef before _slp_tree declaration.
15504         (struct _slp_tree): Use a vector of slp_tree as children.
15505         (vect_get_place_in_interleaving_chain): Remove.
15506         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
15507         Move ...
15508         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
15509         and make static.
15510         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
15511         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
15512         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
15513         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
15514         Use slp_node instead of slp_void_p and adjust.
15516 2013-04-09  Richard Biener  <rguenther@suse.de>
15518         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
15519         work that is not necessary.
15521 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
15523         PR tree-optimization/56854
15524         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
15525         forward into clobber stmts if it would change MEM_REF lhs into
15526         non-MEM_REF.
15528 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
15530         * tree.c (type_hash_lookup, type_hash_add): Make static.
15531         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
15533 2013-04-09  Richard Biener  <rguenther@suse.de>
15535         * tree.h (unsave_expr_now): Remove.
15536         * tree-inline.c (mark_local_for_remap_r): Remove.
15537         (unsave_expr_1): Likewise.
15538         (unsave_r): Likewise.
15539         (unsave_expr_now): Likewise.
15540         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
15541         (propagate_tree_value): Likewise.
15543 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
15545         * doc/rtl.texi (sequence): Rewrite documentation to match the
15546         current use of SEQUENCE rtl objects.
15547         * rtl.def (SEQUENCE): Likewise.
15549         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
15550         Update documentation.
15551         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
15552         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
15554         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
15556 2013-04-08  Teresa Johnson  <tejohnson@google.com>
15558         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
15559         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
15560         methods.
15561         (estimate_edge_size_and_time): Add comment to suggest using rounding
15562         methods.
15563         (estimate_node_size_and_time): Ditto.
15564         (remap_edge_change_prob): Use helper rounding divide methods.
15565         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
15566         (gimple_mod_pow2_value_transform): Ditto.
15567         (gimple_mod_subtract_transform): Ditto.
15568         (gimple_ic_transform): Ditto.
15569         (gimple_stringops_transform): Ditto.
15570         * stmt.c (conditional_probability): Ditto.
15571         (emit_case_dispatch_table): Ditto.
15572         * lto-cgraph.c (merge_profile_summaries): Ditto.
15573         * tree-optimize.c (execute_fixup_cfg): Ditto.
15574         * cfgcleanup.c (try_forward_edges): Ditto.
15575         * cfgloopmanip.c (scale_loop_profile): Ditto.
15576         (loopify): Ditto.
15577         (duplicate_loop_to_header_edge): Ditto.
15578         (lv_adjust_loop_entry_edge): Ditto.
15579         * tree-vect-loop.c (vect_transform_loop): Ditto.
15580         * profile.c (compute_branch_probabilities): Ditto.
15581         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
15582         * lto-streamer-in.c (input_cfg): Ditto.
15583         * gimple-streamer-in.c (input_bb): Ditto.
15584         * ipa-cp.c (update_profiling_info): Ditto.
15585         (update_specialized_profile): Ditto.
15586         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
15587         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
15588         rounding methods.
15589         * sched-rgn.c (compute_dom_prob_ps): Ditto.
15590         (compute_trg_info): Ditto.
15591         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
15592         (purge_dead_edges): Ditto.
15593         * loop-unswitch.c (unswitch_loop): Ditto.
15594         * cgraphclones.c (cgraph_clone_edge): Ditto.
15595         (cgraph_clone_node): Ditto.
15596         * tree-inline.c (copy_bb): Ditto.
15597         (copy_edges_for_bb): Ditto.
15598         (initialize_cfun): Ditto.
15599         (copy_cfg_body): Ditto.
15600         (expand_call_inline): Ditto.
15602 2013-04-08  Kai Tietz  <ktietz@redhat.com>
15604         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
15605         TARGET_CYGWIN64 by TARGET_64BIT.
15607 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
15609         * config/epiphany/epiphany.md (GPR_1): New constant.
15610         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
15611         * config/epiphany/epiphany.c (gen_compare_reg):
15612         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
15613         is already in place.
15614         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
15615         Don't require being called during rtl expansion; If y operlaps r0,
15616         return 0.
15617         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
15618         (epiphany_expand_epilogue): Likewise.
15620         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
15621         Don't use CC_FPmode for ORDERED / UNORDERED.
15622         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
15624         * config/epiphany/constraints.md (CnL): New constraint.
15625         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
15626         * config/epiphany/predicates.md (add_operand): Allow 1024.
15628         * config/epiphany/epiphany.md (logical_op): New code iterator.
15629         (op_mnc): New code attribute.
15630         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
15631         (mov_f+1, mov_f+2): New peephole2 patterns.
15633         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
15634         (cstoresi4): Also allow re-use of zero result when doing a NE
15635         comparison to a non-zero operand.
15636         Use (clobber (scratch)) for first insn if the gpr output is not needed.
15638         * config/epiphany/epiphany.md (<insn_opname>v2si3):
15639         Use gen_addsi3_i / gen_subsi3_i.
15641 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
15643         PR c++/34949
15644         PR c++/50243
15645         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
15646         contain anything but clobbers, at most one __builtin_stack_restore,
15647         optionally debug stmts and final resx, and if it has at least one
15648         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
15649         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
15650         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
15651         which isn't defaut definition, remove them.
15652         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
15653         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
15654         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
15655         with MEM_REF LHS with SSA_NAME address.
15657 2013-04-08  Jeff Law  <law@redhat.com>
15659         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
15661 2013-04-08  Richard Biener  <rguenther@suse.de>
15663         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
15664         extra newline.
15665         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
15666         determined vector type.
15667         (vect_analyze_data_refs): Likewise.
15668         (vect_get_new_vect_var): Adjust.
15669         (vect_create_destination_var): Preserve SSA name versions.
15670         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
15671         not dump anything here.
15673 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
15675         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
15676         Add member lr_slot_known.
15677         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
15678         if necessary.
15679         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
15680         Remove code that sets lr_slot_offset according to what a previous
15681         version of epiphany_emit_save_restore used to do.
15682         (epiphany_emit_save_restore): When doing an lr save or restore,
15683         set/verify lr_slot_known and lr_slot_offset.
15685 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
15687         PR target/54338
15688         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
15689         in ALL_REGS.
15691 2013-04-08  Richard Biener  <rguenther@suse.de>
15693         * alias.c (find_base_term): Fix thinko in previous change.
15695 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
15697         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
15698         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
15699         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
15700         if possible to compute val.
15701         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
15702         For QImode integers don't require anything about precision.  Use
15703         const_with_all_bytes_same to find out if the constant doesn't have
15704         repeated bytes in it.
15706 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15708         * config/s390/s390.c (s390_expand_insv): Only accept insertions
15709         within mode size.
15711 2013-04-08  Marek Polacek  <polacek@redhat.com>
15713         PR rtl-optimization/48182
15714         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
15715         value to 1.
15717 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
15719         PR target/55487
15720         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
15721         nuses, make sure we have a label.
15723 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
15725         PR target/56843
15726         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
15727         (rs6000_emit_swdiv_low_precision): Remove.
15728         (rs6000_emit_swdiv): Rewrite to handle between one and four
15729         iterations of Newton-Raphson generally; modify required number of
15730         iterations for some cases.
15731         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
15733 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
15735         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
15736         set-but-unused variable.
15738         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
15739         basic blocks of released function bodies garbage-collectable.
15741         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
15742         (struct rtl_opt_pass): Add TODO_df_finish.
15744         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
15746 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15748         * config/arm/constraints.md (q): New constraint.
15749         * config/arm/ldrdstrd.md: New file.
15750         * config/arm/arm.md (ldrdstrd.md) New include.
15751         (arm_movdi): Use "q" instead of "r" constraint
15752         for double-word memory access.
15753         (movdf_soft_insn): Likewise.
15754         * config/arm/vfp.md (movdi_vfp): Likewise.
15755         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
15756         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
15757         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
15758         (mem_ok_for_ldrd_strd): Likewise.
15759         (output_move_double): Update assertion.
15761 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15763         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
15765 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15767         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
15768         define_insn_and_split.
15769         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
15771 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15773         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
15774         define_insn_and_split.
15775         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
15776         (shiftsi3_compare): New pattern.
15777         (rrx): New pattern.
15778         * config/arm/unspecs.md (UNSPEC_RRX): New.
15780 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15782         * config/arm/arm.md (negdi_extendsidi): New pattern.
15783         (negdi_zero_extendsidi): Likewise.
15785 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15787         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
15788         define_insn_and_split.
15789         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
15790         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
15792 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15794         * config/arm/arm.md (arm_subdi3): Convert define_insn into
15795         define_insn_and_split.
15796         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
15797         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
15799 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15801         * config/arm/arm.md (subsi3_carryin): New pattern.
15802         (subsi3_carryin_const): Likewise.
15803         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
15804         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
15806 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15808         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
15810 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
15812         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
15813         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
15815 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15817         * config/arm/arm.c (arm_expand_builtin): Change fcode
15818         type to unsigned int.
15820 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
15822         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
15824 2013-04-04  Ian Lance Taylor  <iant@google.com>
15826         * doc/standards.texi (Standards): The Go frontend supports the Go 1
15827         language standard.
15829 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
15831         PR middle-end/56729
15832         * df-scan.c (df_insn_delete): Disable failing assert.
15834 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15836         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
15837         New function prototype.
15838         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
15839         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
15840         (arm_builtin_vectorized_function): New function.
15842 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15844         * config/arm/arm_neon_builtins.def: New file.
15845         * config/arm/arm.c (neon_builtin_data): Move contents to
15846         arm_neon_builtins.def.
15847         (enum arm_builtins): Include neon builtin definitions.
15848         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
15849         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
15851 2013-04-04  Marek Polacek  <polacek@redhat.com>
15853         PR tree-optimization/48186
15854         * predict.c (maybe_hot_frequency_p): Return false if
15855         HOT_BB_FREQUENCY_FRACTION is 0.
15856         (cgraph_maybe_hot_edge_p): Likewise.
15858 2013-04-04  Richard Biener  <rguenther@suse.de>
15860         PR tree-optimization/56826
15861         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
15862         more accurately.
15864 2013-04-04  Richard Biener  <rguenther@suse.de>
15866         PR tree-optimization/56213
15867         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
15868         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
15870 2013-04-04  Richard Biener  <rguenther@suse.de>
15872         PR tree-optimization/56837
15873         * tree-loop-distribution.c (classify_partition): For non-zero
15874         values require that the value has the same precision as its
15875         mode to be useful as memset value.
15877 2013-04-03  Nick Clifton  <nickc@redhat.com>
15879         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
15880         (fmssf4): Use fmsf.s on E3V5 architectures.
15881         (fnmasf4): Use fnmaf.s on E3V5 architectures.
15882         (fnmssf4): Use fnmsf.s on E3V5 architectures.
15884 2013-04-03  Jeff Law  <law@redhat.com>
15886         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
15887         (lra-eliminations.o): Likewise.
15889 2013-04-03  Teresa Johnson  <tejohnson@google.com>
15891         * gcov-io.c (compute_working_sets): Moved most of body of old
15892         compute_working_sets here from profile.c.
15893         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
15894         (gcov_working_set_t): Moved typedef here from basic-block.h
15895         (compute_working_set): Declare.
15896         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
15897         (get_working_sets): Renamed from compute_working_set,
15898         replace most of body with call to new compute_working_sets.
15899         (get_exec_counts): Replace call to compute_working_sets
15900         to get_working_sets.
15901         * profile.h (get_working_sets): Renamed from compute_working_set.
15902         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
15903         to get_working_sets.
15904         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
15905         * gcov-dump.c (dump_working_sets): New function.
15907 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
15909         * hwint.c (sext_hwi, zext_hwi): New functions.
15910         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
15911         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
15912         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
15913         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
15914         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
15915         (sext_hwi, zext_hwi): New functions.
15917 2013-04-03  Jeff Law  <law@redhat.com>
15919         PR tree-optimization/56799
15920         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
15921         back test for widening conversion erroneously dropped in prior change.
15923 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15925         PR target/56809
15926         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
15927         instead of next_real_insn.
15929 2013-04-03  Marek Polacek  <polacek@redhat.com>
15931         PR sanitizer/55702
15932         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
15934 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15936         PR target/56809
15937         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
15938         next_real_insn.
15939         (thumb1_output_casesi): Likewise.
15940         (thumb2_output_casesi): Likewise.
15942 2013-04-03  Richard Biener  <rguenther@suse.de>
15944         PR tree-optimization/56817
15945         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
15946         Split out ...
15947         (tree_unroll_loops_completely_1): ... new function to manually
15948         walk the loop tree, properly defering outer loops of unrolled
15949         loops to later iterations.
15951 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
15953         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
15954         (vectorizable_load): Likewise.
15955         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
15956         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
15958 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
15960         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
15961         BIT_FIELD_REF.
15963 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
15965         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
15967 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
15969         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
15971 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
15973         PR tree-optimization/56790
15974         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
15975         folding.
15977 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
15979         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
15980         Handle VEC_MERGE.
15981         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
15982         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
15983         equal arguments.
15985 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
15987         PR c/19449
15988         * tree.h (force_folding_builtin_constant_p): New decl.
15989         * builtins.c (force_folding_builtin_constant_p): New variable.
15990         (fold_builtin_constant_p): Fold immediately also if
15991         force_folding_builtin_constant_p.
15993 2013-04-03  Richard Biener  <rguenther@suse.de>
15995         PR tree-optimization/56812
15996         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
15997         DRs of the same interleaving chain are independent.
15999 2013-04-02  Jason Merrill  <jason@redhat.com>
16001         * gdbinit.in (pbb): Use debug fn.
16003 2013-04-02  Lawrence Crowl  <crowl@google.com>
16005         * sese.h (struct ivtype_map_elt_s): Remove unused.
16006         (extern debug_ivtype_map): Remove unused.
16007         (extern eq_ivtype_map_elts): Remove unused.
16008         * sese.c (debug_ivtype_map): Removed unused.
16009         (debug_ivtype_map_1): Removed unused.
16010         (debug_ivtype_elt): Remove unused.
16011         (eq_ivtype_map_elts): Remove unused.
16014 2013-04-02  Kai Tietz  <ktietz@redhat.com>
16016         PR target/52790
16017         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
16018         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
16019         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
16020         function.
16021         (legitimize_pe_coff_symbol): Likewise.
16022         (is_imported_p): New helper-function.
16023         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
16024         for Windows x64 targets.
16025         (ix86_expand_prologue): Optimize for pe-coff targets.
16026         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
16027         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
16028         medium/large code-model.
16029         (legitimize_pic_address): Likewise.
16030         (legitimize_tls_address): Likewise.
16031         (ix86_expand_call): Likewise.
16032         (x86_output_mi_thunk): Likewise.
16033         (get_dllimport_decl): Add new beimport argument.
16034         (construct_plt_address): Don't assert for x64 pe-coff targets.
16035         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
16036         targets.
16037         (SYMBOL_FLAG_STUBVAR): New macro.
16038         (SYMBOL_REF_STUBVAR_P): Likewise.
16039         * config/i386/winnt.c (stub_list): New structure.
16040         (stub_head): New local variable.
16041         (i386_pe_record_stub): New function.
16042         (i386_pe_file_end): Emit refptr-stubs.
16044 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
16046         PR rtl-optimization/56745
16047         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
16048         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
16050         PR c++/34949
16051         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
16052         and both of them are MEM_REFs, just compare first argument for
16053         equality and attempt to deal even with differing offsets.
16055         PR c++/34949
16056         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
16057         of gimple_clobber_p to be MEM_REF.
16058         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
16059         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
16060         after gimplification.
16061         * asan.c (get_mem_ref_of_assignment): Don't instrument
16062         gimple_clobber_p stmts.
16063         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
16064         gimple_clobber_p stmt if they have MEM_REF lhs and
16065         are dead because of another gimple_clobber_p stmt.
16066         * tree-ssa-live.c (clear_unused_block_pointer): Treat
16067         gimple_clobber_p stmts like debug stmts.
16068         (remove_unused_locals): Remove clobbers with MEM_REF lhs
16069         that refer to unused VAR_DECLs or uninitialized values.
16070         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
16071         gimple_clobber_p stmts if they refer to removed parameters.
16072         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
16073         formatting.
16075 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
16077         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
16078         using SWI48 mode attribute.
16080 2013-04-02  Wei Mi  <wmi@google.com>
16082         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
16083         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
16084         *<rotate_insn><mode>3_mask in i386.md.
16086 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
16088         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
16090 2013-04-02  Richard Biener  <rguenther@suse.de>
16092         PR tree-optimization/56778
16093         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
16094         Runtime alias tests are not supported for gather loads.
16095         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
16096         stmts referenced from SSA operands before updating SSA form.
16098 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
16099             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16101         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
16102         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
16103         * config/arm/cortex-a53.md: New file.
16104         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
16105         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
16106         * config/arm/arm.c (arm_issue_rate): Likewise.
16107         * config/arm/arm-tune.md: Regenerate
16108         * config/arm/arm-tables.opt: Regenerate.
16109         * config/arm/arm-cores.def: Add cortex-a53.
16111 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
16113         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
16114         non-static link.
16116 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
16118         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
16119         scalar load/store operations using B/H registers.
16120         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
16122 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
16124         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
16125         scalar move.
16126         * config/aarch64/aarch64.c
16127         (aarch64_simd_scalar_immediate_valid_for_move): New.
16128         * config/aarch64/aarch64-protos.h
16129         (aarch64_simd_scalar_immediate_valid_for_move): New.
16130         * config/aarch64/constraints.md (Dh, Dq): New.
16131         * config/aarch64/iterators.md (hq): New.
16133 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
16135         * reorg.c (get_branch_condition): Deal with conditional returns.
16136         (fill_simple_delay_slots): Remove dead code dealing with jumps.
16138 2013-04-01  Wei Mi  <wmi@google.com>
16140         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
16141         Truncate operand 2 using %b asm operand modifier.
16142         (*<shift_insn><mode>3_mask): Ditto.
16143         (*<rotate_insn><mode>3_mask): Ditto.
16145 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
16147         PR middle-end/56798
16148         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
16150 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
16152         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
16153         of next_real_insn.
16154         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
16156 2013-03-30  Lawrence Crowl  <crowl@google.com>
16158         * dse.c (clear_alias_sets): Remove never set.
16159         (disqualified_clear_alias_sets): Remove never set.
16160         (clear_alias_mode_pool): Remove never set.
16161         (dse_step0): Remove condition that is never true.
16162         (canon_address): Remove condition that is never true.
16163         (dse_step7): Remove condition that is never true.
16164         (rest_of_handle_dse): Remove condition that is never true.
16165         (rest_of_handle_dse::did_global): Remove never read from above.
16166         (dse_step2_spill): Remove never called from above.
16167         (dse_step5_spill): Remove never called from above.
16169 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
16171         * doc/md.texi (Standard Names) <casesi>: Update documentation for
16172         JUMP_TABLE_DATA changes.
16173         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
16174         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
16175         (Insns) <jump_table_data>: New entry.
16176         * doc/tm.texi: Regenerate.
16178         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
16180         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
16181         for table jump at the end of a basic block using tablejump_p.
16182         * targhooks.c (default_invalid_within_doloop): Likewise.
16183         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
16184         target hook implementation that is identical to the default hook.
16185         (rs6000_invalid_within_doloop): Remove.
16187         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
16188         unused variable from tablejump_p call.
16190         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
16191         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
16192         (INSN_DELETED_P): Likewise.
16193         (emit_jump_table_data): New prototype.
16194         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
16195         after 4th as unused.
16196         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
16197         * sched-vis.c (print_insn): Likewise.
16198         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
16199         insn for compatibility with back ends that use next_active_insn to
16200         identify jump table data.
16201         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
16202         (remove_insn): Likewise.
16203         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
16204         to be emitted.
16205         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
16206         (emit_jump_table_data): New function.
16208         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
16209         basic block, a JUMP_TABLE_DATA never is.
16210         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
16211         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
16212         off from code handling real insns.
16213         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
16214         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
16215         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
16216         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
16217         is not a NONDEBUG_INSN_P.
16218         * ira-costs.c (scan_one_insn): Likewise.
16219         * jump.c (mark_all_labels): Likewise.
16220         (mark_jump_label_1): Likewise.
16221         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
16222         * lra.c (get_insn_freq): Expect all insns reaching here to be in
16223         a basic block.
16224         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
16225         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
16226         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
16227         JUMP_TABLE_DATA_P insns.
16228         (calculate_elim_costs_all_insns): Likewise.
16229         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
16230         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
16231         (delete_output_reload): Code style fixups.
16232         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
16233         insn flags on this non-insn.
16234         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
16235         as scheduling barriers, for pre-change compatibility.
16236         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
16237         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
16239         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
16240         redundant JUMP_TABLE_DATA_P test.
16241         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
16242         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
16243         (frv_for_each_packet): Likewise.
16244         * config/i386/i386.c (min_insn_size): Likewise.
16245         (ix86_avoid_jump_mispredicts): Likewise.
16246         * config/m32r/m32r.c (m32r_is_insn): Likewise.
16247         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
16248         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
16249         (mips16_insn_length): Robustify.
16250         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
16251         (mips16_split_long_branches): Likewise.
16252         * config/pa/pa.c (pa_combine_instructions): Likewise.
16253         * config/rs6000/rs6000.c (get_next_active_insn): Treat
16254         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
16255         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
16256         as contributing to pool range lengths.
16257         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
16258         Remove redundant JUMP_TABLE_DATA_P test.
16259         (sh_loop_align): Likewise.
16260         (split_branches): Likewise.
16261         (sh_insn_length_adjustment): Likewise.
16262         * config/spu/spu.c (get_branch_target): Likewise.
16264 2013-03-29  Jan Hubicka  <jh@suse.cz>
16266         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
16267         gcov streaming; stream hot bb threshold to ltrans.
16268         * predict.c (get_hot_bb_threshold): Break out from ....
16269         (maybe_hot_count_p): ... here.
16270         (set_hot_bb_threshold): New function.
16271         * lto-section-in.c (lto_section_name): Add profile.
16272         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
16273         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
16274         and data-streamer.h
16275         (histogram_entry): New structure.
16276         (histogram, histogram_pool): New global vars.
16277         (histogram_hash): New structure.
16278         (histogram_hash::hash): New method.
16279         (histogram_hash::equal): Likewise.
16280         (account_time_size): New function.
16281         (cmp_counts): New function.
16282         (dump_histogram): New function.
16283         (ipa_profile_generate_summary): New function.
16284         (ipa_profile_write_summary): New function.
16285         (ipa_profile_read_summary): New function.
16286         (ipa_profile): Decide on threshold.
16287         (pass_ipa_profile): Add ipa_profile_write_summary and
16288         ipa_profile_read_summary.
16289         * Makefile.in (ipa.o): Update dependencies.
16290         * lto-streamer.h (LTO_section_ipa_profile): New section.
16292 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
16294         * tree.h (VAR_P): New.
16296 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
16298         PR lto/56777
16299         * doc/invoke.texi ([-fwhole-program]): Fix typo.
16301 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
16303         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
16304         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
16305         (control_flow_insn_p): Likewise.
16306         * cfgrtl.c (duplicate_insn_chain): Likewise.
16307         * final.c (get_attr_length_1): Likewise.
16308         (shorten_branches): Likewise.
16309         (final_scan_insn): Likewise.
16310         * function.c (instantiate_virtual_regs): Likewise.
16311         * gcse.c (insert_insn_end_basic_block): Likewise.
16312         * ira-costs.c (scan_one_insn): Likewise.
16313         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
16314         * lra.c (check_rtl): Likewise.
16315         * reload1.c (elimination_costs_in_insn): Likewise.
16316         * reorg.c (follow_jumps): Likewise.
16318         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
16319         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
16320         (thumb_far_jump_used_p): Likewise.
16321         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
16322         (workaround_speculation): Likewise.
16323         (add_sched_insns_for_speculation): Likewise.
16324         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
16325         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
16326         (frv_for_each_packet): Likewise.
16327         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
16328         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
16329         (final_emit_insn_group_barriers): Likewise.
16330         * config/m32r/m32r.c (m32r_is_insn): Likewise.
16331         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
16332         (mips16_insn_length): Likewise.
16333         * config/pa/pa.c (pa_reorg): Likewise.
16334         (pa_combine_instructions): Likewise.
16335         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
16336         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
16337         (sh_reorg): Likewise.
16338         (split_branches): Likewise.
16339         * config/spu/spu.c (get_branch_target): Likewise.
16341         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
16342         JUMP_TABLE_DATA_P.
16344 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
16346         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
16347         Fix declaration name.
16349 2013-03-28  Lawrence Crowl  <crowl@google.com>
16351         * graphds.h (struct graph.indicies): Remove unused.
16352         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
16353         (SCOP_ORIGINAL_PDDRS): Remove unused.
16354         * sese.h (extern insert_loop_close_phis): Removed unused.
16355         (extern insert_guard_phis): Removed unused.
16356         (extern ivtype_map_elt_info): Removed unused.
16357         (new_ivtype_map_elt): Removed unused.
16358         * sese.c (ivtype_map_elt_info): Removed unused.
16360 2013-03-28  Lawrence Crowl  <crowl@google.com>
16362         * Makefile.in: Add several missing include dependences.
16363         (DUMPFILE_H): New.
16364         (test-dump.o): New.  This object is not added to any executable,
16365         but is present for ad-hoc testing.
16366         * bitmap.c
16367         (debug (const bitmap_head_def &)): New.
16368         (debug (const bitmap_head_def *)): New.
16369         * bitmap.h
16370         (extern debug (const bitmap_head_def &)): New.
16371         (extern debug (const bitmap_head_def *)): New.
16372         * cfg.c
16373         (debug (edge_def &)): New.
16374         (debug (edge_def *)): New.
16375         * cfghooks.c
16376         (debug (basic_block_def &)): New.
16377         (debug (basic_block_def *)): New.
16378         * dumpfile.h
16379         (dump_node (const_tree, int, FILE *)): Correct source file.
16380         * dwarf2out.c
16381         (debug (die_struct &)): New.
16382         (debug (die_struct *)): New.
16383         * dwarf2out.h
16384         (extern debug (die_struct &)): New.
16385         (extern debug (die_struct *)): New.
16386         * gimple-pretty-print.c
16387         (debug (gimple_statement_d &)): New.
16388         (debug (gimple_statement_d *)): New.
16389         * gimple-pretty-print.h
16390         (extern debug (gimple_statement_d &)): New.
16391         (extern debug (gimple_statement_d *)): New.
16392         * ira-build.c
16393         (debug (ira_allocno_copy &)): New.
16394         (debug (ira_allocno_copy *)): New.
16395         (debug (ira_allocno &)): New.
16396         (debug (ira_allocno *)): New.
16397         * ira-int.h
16398         (extern debug (ira_allocno_copy &)): New.
16399         (extern debug (ira_allocno_copy *)): New.
16400         (extern debug (ira_allocno &)): New.
16401         (extern debug (ira_allocno *)): New.
16402         * ira-lives.c
16403         (debug (live_range &)): New.
16404         (debug (live_range *)): New.
16405         * lra-int.h
16406         (debug (lra_live_range &)): New.
16407         (debug (lra_live_range *)): New.
16408         * lra-lives.c
16409         (debug (lra_live_range &)): New.
16410         (debug (lra_live_range *)): New.
16411         * omega.c
16412         (debug (omega_pb_d &)): New.
16413         (debug (omega_pb_d *)): New.
16414         * omega.h
16415         (extern debug (omega_pb_d &)): New.
16416         (extern debug (omega_pb_d *)): New.
16417         * print-rtl.c
16418         (debug (const rtx_def &)): New.
16419         (debug (const rtx_def *)): New.
16420         * print-tree.c
16421         (debug_tree (tree): Move within file.
16422         (debug_raw (const tree_node &)): New.
16423         (debug_raw (const tree_node *)): New.
16424         (dump_tree_via_hooks (const tree_node *, int)): New.
16425         (debug (const tree_node &)): New.
16426         (debug (const tree_node *)): New.
16427         (debug_verbose (const tree_node &)): New.
16428         (debug_verbose (const tree_node *)): New.
16429         (debug_head (const tree_node &)): New.
16430         (debug_head (const tree_node *)): New.
16431         (debug_body (const tree_node &)): New.
16432         (debug_body (const tree_node *)): New.
16433         (debug_vec_tree (tree): Move and reimplement in terms of dump.
16434         (debug (vec<tree, va_gc> &)): New.
16435         (debug (vec<tree, va_gc> *)): New.
16436         * rtl.h
16437         (extern debug (const rtx_def &)): New.
16438         (extern debug (const rtx_def *)): New.
16439         * sbitmap.c
16440         (debug_raw (simple_bitmap_def &)): New.
16441         (debug_raw (simple_bitmap_def *)): New.
16442         (debug (simple_bitmap_def &)): New.
16443         (debug (simple_bitmap_def *)): New.
16444         * sbitmap.h
16445         (extern debug (simple_bitmap_def &)): New.
16446         (extern debug (simple_bitmap_def *)): New.
16447         (extern debug_raw (simple_bitmap_def &)): New.
16448         (extern debug_raw (simple_bitmap_def *)): New.
16449         * sel-sched-dump.c
16450         (debug (vinsn_def &)): New.
16451         (debug (vinsn_def *)): New.
16452         (debug_verbose (vinsn_def &)): New.
16453         (debug_verbose (vinsn_def *)): New.
16454         (debug (expr_def &)): New.
16455         (debug (expr_def *)): New.
16456         (debug_verbose (expr_def &)): New.
16457         (debug_verbose (expr_def *)): New.
16458         (debug (vec<rtx> &)): New.
16459         (debug (vec<rtx> *)): New.
16460         * sel-sched-dump.h
16461         (extern debug (vinsn_def &)): New.
16462         (extern debug (vinsn_def *)): New.
16463         (extern debug_verbose (vinsn_def &)): New.
16464         (extern debug_verbose (vinsn_def *)): New.
16465         (extern debug (expr_def &)): New.
16466         (extern debug (expr_def *)): New.
16467         (extern debug_verbose (expr_def &)): New.
16468         (extern debug_verbose (expr_def *)): New.
16469         (extern debug (vec<rtx> &)): New.
16470         (extern debug (vec<rtx> *)): New.
16471         * sel-sched-ir.h
16472         (_list_iter_cond_expr): Make inline instead of static.
16473         * sreal.c
16474         (debug (sreal &)): New.
16475         (debug (sreal *)): New.
16476         * sreal.h
16477         (extern debug (sreal &)): New.
16478         (extern debug (sreal *)): New.
16479         * tree.h
16480         (extern debug_raw (const tree_node &)): New.
16481         (extern debug_raw (const tree_node *)): New.
16482         (extern debug (const tree_node &)): New.
16483         (extern debug (const tree_node *)): New.
16484         (extern debug_verbose (const tree_node &)): New.
16485         (extern debug_verbose (const tree_node *)): New.
16486         (extern debug_head (const tree_node &)): New.
16487         (extern debug_head (const tree_node *)): New.
16488         (extern debug_body (const tree_node &)): New.
16489         (extern debug_body (const tree_node *)): New.
16490         (extern debug (vec<tree, va_gc> &)): New.
16491         (extern debug (vec<tree, va_gc> *)): New.
16492         * tree-cfg.c
16493         (debug (struct loop &)): New.
16494         (debug (struct loop *)): New.
16495         (debug_verbose (struct loop &)): New.
16496         (debug_verbose (struct loop *)): New.
16497         * tree-dump.c: Add header dependence.
16498         * tree-flow.h
16499         (extern debug (struct loop &)): New.
16500         (extern debug (struct loop *)): New.
16501         (extern debug_verbose (struct loop &)): New.
16502         (extern debug_verbose (struct loop *)): New.
16503         * tree-data-ref.c
16504         (debug (data_reference &)): New.
16505         (debug (data_reference *)): New.
16506         (debug (vec<data_reference_p> &)): New.
16507         (debug (vec<data_reference_p> *)): New.
16508         (debug (vec<ddr_p> &)): New.
16509         (debug (vec<ddr_p> *)): New.
16510         * tree-data-ref.h
16511         (extern debug (data_reference &)): New.
16512         (extern debug (data_reference *)): New.
16513         (extern debug (vec<data_reference_p> &)): New.
16514         (extern debug (vec<data_reference_p> *)): New.
16515         (extern debug (vec<ddr_p> &)): New.
16516         (extern debug (vec<ddr_p> *)): New.
16517         * tree-ssa-alias.c
16518         (debug (pt_solution &)): New.
16519         (debug (pt_solution *)): New.
16520         * tree-ssa-alias.h
16521         (extern debug (pt_solution &)): New.
16522         (extern debug (pt_solution *)): New.
16523         * tree-ssa-alias.c
16524         (debug (_var_map &)): New.
16525         (debug (_var_map *)): New.
16526         (debug (tree_live_info_d &)): New.
16527         (debug (tree_live_info_d *)): New.
16528         * tree-ssa-alias.h
16529         (extern debug (_var_map &)): New.
16530         (extern debug (_var_map *)): New.
16531         (extern debug (tree_live_info_d &)): New.
16532         (extern debug (tree_live_info_d *)): New.
16534 2013-03-28  Jan Hubicka  <jh@suse.cz>
16536         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
16538 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
16540         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
16541         record only when desired or required.
16543 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
16545         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
16546         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
16548 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16550         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
16551         (*andsi3_compare0_uxtw): New pattern.
16552         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
16553         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
16555 2013-03-28  Jan Hubicka  <jh@suse.cz>
16557         * data-streamer-in.c (streamer_read_gcov_count): New function.
16558         * gimple-streamer-out.c: Include value-prof.h.
16559         (output_gimple_stmt): Output histogram.
16560         (output_bb): Use streamer_write_gcov_count.
16561         * value-prof.c: Include data-streamer.h
16562         (dump_histogram_value): Add HIST_TYPE_MAX.
16563         (stream_out_histogram_value): New function.
16564         (stream_in_histogram_value): New function.
16565         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
16566         (stream_out_histogram_value, stream_in_histogram_value): Declare.
16567         * data-streamer-out.c (streamer_write_gcov_count): New function.
16568         (streamer_write_gcov_count_stream): New function.
16569         * lto-cgraph.c (lto_output_edge): Update counter streaming.
16570         (lto_output_node): Likewise.
16571         (input_node, input_edge): Likewise.
16572         * lto-streamer-out.c (output_cfg): Update streaming.
16573         * lto-streamer-in.c (input_cfg): Likewise.
16574         * data-streamer.h (streamer_write_gcov_count,
16575         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
16576         * gimple-streamer-in.c: Include value-prof.h
16577         (input_gimple_stmt): Input histograms.
16578         (input_bb): Update profile streaming.
16580 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
16582         * genmodes.c (emit_max_int): New function.
16583         (emit_insn_modes_h): Added call to emit_max_function.
16584         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
16585         Added doc.
16586         * machmode.def: Fixed comment.
16588 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
16590         * combine.c (try_combine): Removed useless assert.
16591         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
16593 2013-03-28  Marek Polacek  <polacek@redhat.com>
16594             Richard Biener  <rguenther@suse.de>
16596         PR tree-optimization/56695
16597         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
16598         build signed result of a vector comparison.
16599         * tree-cfg.c (verify_gimple_comparison): Check that a result
16600         of a vector comparison has signed type.
16602 2013-03-28  Richard Biener  <rguenther@suse.de>
16604         PR tree-optimization/37021
16605         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
16606         do not restrict gaps between groups.
16607         * tree-vect-stmts.c (vectorizable_load): Properly account for
16608         a gap between groups.
16610 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
16612         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
16613         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
16614         is not enabled.
16616 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
16618         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
16619         * doc/extend.texi (Named Address Spaces): Ditto.
16620         (Variable Attributes): Ditto.
16622 2013-03-27  Kai Tietz  <ktietz@redhat.com>
16624         * config.build: Add support for cygwin x64 target.
16625         * config.gcc: Likewise.
16626         * config.host: Likewise.
16627         * configure.ac: Likewise
16628         * configure: Regenerated.
16630 2013-03-27  Kai Tietz  <ktietz@redhat.com>
16632         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
16633         * config/i386/t-cygwin-w64: New file.
16634         * config/i386/cygwin-w64.h: New file.
16635         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
16636         and add support for x64-cygwin target.
16637         (CPP_SPEC): Likewise.
16638         (CXX_WRAP_SPEC_LIST): Undefine before define.
16639         (LIBGCJ_SONAME): Use 15 as version.
16641 2013-03-27  Richard Biener  <rguenther@suse.de>
16643         PR tree-optimization/56716
16644         * tree-ssa-structalias.c (perform_var_substitution): Adjust
16645         dumping for ref nodes.
16647 2013-03-27  Martin Jambor  <mjambor@suse.cz>
16649         PR tree-optimization/55334
16650         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
16651         restricted pointers to arrays.
16653 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
16655         * Makefile.in (.SUFFIXES): Add .cc.
16656         (.c.o): Apply same recipe for implicit rule .cc.o.
16658 2013-03-27  Richard Biener  <rguenther@suse.de>
16660         PR tree-optimization/37021
16661         * tree-vect-data-refs.c (vect_check_strided_load): Allow
16662         REALPART/IMAGPART_EXPRs around the supported refs.
16663         * tree-ssa-structalias.c (find_func_aliases): Assume that
16664         floating-point values are not used to transfer pointers.
16666 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
16668         * target.def (TARGET_HAS_IFUNC_P): New target hook.
16669         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
16670         * doc/tm.texi: Regenerate.
16671         * targhooks.h (default_has_ifunc_p): New.
16672         * targhooks.c (default_has_ifunc_p): Ditto.
16673         * config/linux-protos.h: New file.
16674         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
16675         hook for linux which disables support of indirect functions in android.
16676         * config/linux-android.c: New file.
16677         * config/t-linux-android.c: Ditto.
16678         * config.gcc: Added new object file linux-android.o.
16679         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
16680         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
16681         * varasm.c (do_assemble_alias): Likewise.
16682         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
16683         doesn't support indirect functions.
16684         * configure: Regenerate.
16686 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
16688         PR target/56102
16689         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
16690         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
16691         mult-word mode.
16693 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16695         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
16697 2013-03-27  Terry Guo  <terry.guo@arm.com>
16699         * config/arm/arm-cores.def: Added core cortex-r7.
16700         * config/arm/arm-tune.md: Regenerated.
16701         * config/arm/arm-tables.opt: Regenerated.
16702         * doc/invoke.texi: Added entry for core cortex-r7.
16704 2013-03-27  Walter Lee  <walt@tilera.com>
16706         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
16707         double-decrement of next_scratch_regno.
16709 2013-03-27  Walter Lee  <walt@tilera.com>
16711         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
16712         input operands.
16713         (insn_v1mulus): Ditto.
16714         (insn_v2muls): Ditto.
16716 2013-03-27  Walter Lee  <walt@tilera.com>
16718         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
16719         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
16721 2013-03-27  Walter Lee  <walt@tilera.com>
16723         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
16724         (*sibcall_value): Ditto.
16726 2013-03-27  Walter Lee  <walt@tilera.com>
16728         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
16729         (insn_mnz_v8qi): ... this ...
16730         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
16731         vector equivalent.
16732         (insn_v<n>mnz): Replaced by ...
16733         (insn_v1mnz): ... this ...
16734         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
16735         equivalent.
16736         (insn_mz_<mode>): Replaced by ...
16737         (insn_mz_v8qi): ... this ...
16738         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
16739         vector equivalent.
16740         (insn_v<n>mz): Replaced by ...
16741         (insn_v1mz): ... this ...
16742         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
16743         equivalent.
16745 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
16747         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
16749 2013-03-26  Roland McGrath  <mcgrathr@google.com>
16751         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
16752         than fprintf with a non-constant, non-format string.
16754 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
16756         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
16757         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
16758         operand 0 predicate.
16759         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
16760         attribute.  Use general_x64nomem_operand as operand 1 predicate.
16761         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
16762         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
16763         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
16764         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
16765         (mov<mode>_insv_1): Remove expander.  Merge insn with
16766         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
16767         Use general_x64nomem_operand as operand 1 predicate.
16768         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
16769         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
16770         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
16771         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
16772         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
16773         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
16774         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
16775         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
16776         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
16777         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
16778         (general_x64nomem_operand): Ditto.
16780 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
16782         * config/rtems.opt: Add -pthread option.
16784 2013-03-26  Richard Biener  <rguenther@suse.de>
16786         * alias.c (find_base_term): Avoid redundant and not used recursion.
16787         (base_alias_check): Get the initial base term from the caller.
16788         (true_dependence_1): Compute and pass base terms to base_alias_check.
16789         (write_dependence_p): Likewise.
16790         (may_alias_p): Likewise.
16792 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
16794         * config/aarch64/aarch64.c (aarch64_classify_address): Support
16795         PC-relative load in SI modes and above only.
16797 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
16799         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
16800         * config/arm/iwmmxt.md (WCGR0): Update.
16801         (WCGR1, WCGR2, WCGR3): Likewise.
16803 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
16805         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
16806         Use x64 and nox64 isa attributes.
16808 2013-03-26  Richard Biener  <rguenther@suse.de>
16810         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
16811         alignment computations and rely on get_object_alignment_1
16812         for the !TYPE_P case.
16813         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
16815 2013-03-26  Walter Lee  <walt@tilera.com>
16817         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
16818         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
16820 2013-03-25  Jeff Law  <law@redhat.com>
16822         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
16823         check for INTEGRAL_TYPE_P that was missing due to checking in
16824         wrong version of prior patch.
16826 2013-03-25  Walter Lee  <walt@tilera.com>
16828         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
16829         TILEGX_INSN_SHUFFLEBYTES1.
16830         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
16831         shufflebytes1.
16832         (tilegx_builtins): Ditto.
16833         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
16835 2013-03-25  Walter Lee  <walt@tilera.com>
16837         * config/tilegx/tilegx.md (floatsisf2): New pattern.
16838         (floatunssisf2): New pattern.
16839         (floatsidf2): New pattern.
16840         (floatunssidf2): New pattern.
16842 2013-03-25  Walter Lee  <walt@tilera.com>
16844         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
16845         tests for constraint J, K, N, P.
16847 2013-03-25  Walter Lee  <walt@tilera.com>
16849         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
16850         Use indirect/pcrel encoding.
16851         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
16852         Ditto.
16854 2013-03-25  Steve Ellcey  <sellcey@mips.com>
16856         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
16857         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
16858         * config/mips/mips.c (mips_option_override): Set IMADD default.
16859         * config/mips/mips.h (PTF_AVOID_IMADD): New.
16860         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
16861         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
16862         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
16863         * doc/invoke.texi (-mimadd/-mno-imadd): New.
16865 2013-03-25  Jeff Law  <law@redhat.com>
16867         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
16868         slightly to avoid creating and folding useless trees.  Simplify
16869         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
16871 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
16873         * config/i386/i386.md (*zero_extendsidi2): Merge with
16874         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
16875         * config/i386/predicates.md (x86_64_zext_operand): Rename from
16876         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
16877         targets.  Clarify comment.
16879 2013-03-25  Martin Jambor  <mjambor@suse.cz>
16881         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
16882         pass-through jump functions differently.
16883         (ipa_read_jump_function): Likewise.  Also use setter functions to set
16884         up jump functions.
16886 2013-03-25  Martin Jambor  <mjambor@suse.cz>
16888         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
16889         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
16890         process it.
16891         (ipa_get_indirect_edge_target): New function.
16892         (devirtualization_time_bonus): New parameter known_aggs, pass it to
16893         ipa_get_indirect_edge_target.  Update all callers.
16894         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
16895         ipa_get_indirect_edge_target_1 instead of calling
16896         ipa_get_indirect_edge_target.
16897         (create_specialized_node): Pass aggvlas to
16898         ipcp_discover_new_direct_edges.
16900 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16902         * config/arm/arm.md (f_sels, f_seld): New types.
16903         (*cmov<mode>): New pattern.
16904         * config/arm/predicates.md (arm_vsel_comparison_operator): New
16905         predicate.
16907 2013-03-25  Kai Tietz  <ktietz@redhat.com>
16909         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
16910         POSIX-printf for mingw-hosted  builds.
16912 2013-03-25  Richard Biener  <rguenther@suse.de>
16914         PR middle-end/56694
16915         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
16916         must-not-throw stmt location.
16918 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16920         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
16921         Emit load-acquire versions when acq is true.
16922         (arm_emit_store_exclusive): Add rel parameter.
16923         Emit store-release versions when rel is true.
16924         (arm_split_compare_and_swap): Use acquire-release instructions
16925         instead.
16926         of barriers when appropriate.
16927         (arm_split_atomic_op): Likewise.
16928         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
16929         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
16930         (VUNSPEC_SLX): Likewise.
16931         (VUNSPEC_LDA): Likewise.
16932         (VUNSPEC_STL): Likewise.
16933         * config/arm/sync.md (atomic_load<mode>): New pattern.
16934         (atomic_store<mode>): Likewise.
16935         (arm_load_acquire_exclusive<mode>): Likewise.
16936         (arm_load_acquire_exclusivesi): Likewise.
16937         (arm_load_acquire_exclusivedi): Likewise.
16938         (arm_store_release_exclusive<mode>): Likewise.
16940 2013-03-25  Catherine Moore  <clm@codesourcery.com>
16942         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
16943         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
16944         * config/mip/predicates.md (lwsp_swsp_operand,
16945         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
16946         sb16_operand, db4_operand, db7_operand, ib3_operand,
16947         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
16948         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
16949         andi16_operand): New predicates.
16950         * config/mips/mips.md (compression): New attribute.
16951         (enabled): New attribute.
16952         (length): Consider compression in computing length.
16953         (shift_compression): New code attribute.
16954         (*add<mode>3): New operands. Record compression.
16955         (sub<mode>3): Likewise.
16956         (one_cmpl<mode>2): Likewise.
16957         (*and<mode>3): Likewise.
16958         (*ior<mode>3): Likewise.
16959         (unnamed pattern for xor): Likewise.
16960         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
16961         (*<optab><mode>3): Likewise.
16962         (*mov<mode>_internal: Likewise.
16963         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
16964         (mips_unsigned_immediate_p): New.
16965         (umips_lwsp_swsp_address_p): New.
16966         (m16_based_address_p): New.
16967         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
16968         (mips_unsigned_immediate_p): New prototype.
16969         (lwsp_swsp_address_p): New prototype.
16970         (m16_based_address_p): New prototype.
16971         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
16972         (mips_signed_immediate_p): New function.
16973         (m16_based_address_p): New function.
16974         (lwsp_swsp_address_p): New function.
16975         (mips_print_operand_punctuation): Recognize short delay slot insns
16976         for microMIPS.add<mode>3"
16978 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16980         PR target/56720
16981         * config/arm/iterators.md (v_cmp_result): New mode attribute.
16982         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
16984 2013-03-25  Richard Biener  <rguenther@suse.de>
16986         PR tree-optimization/56689
16987         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
16988         any edge.
16990 2013-03-25  Richard Biener  <rguenther@suse.de>
16992         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
16993         of bitmap.
16994         (memory_references): Likewise.
16995         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
16996         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
16997         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
16998         (gather_mem_refs_in_loops): Fold into ...
16999         (analyze_memory_references): ... this.  Move initialization
17000         to tree_ssa_lim_initialize.
17001         (fill_always_executed_in): Rename to ...
17002         (fill_always_executed_in_1): ... this.
17003         (fill_always_executed_in): Move contains_call computation to
17004         this new function from ...
17005         (tree_ssa_lim_initialize): ... here.
17006         (tree_ssa_lim): Call fill_always_executed_in.
17008 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
17010         * postreload.c (reload_combine): Fix code detecting returns.
17012 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
17014         * function.c (emit_use_return_register_into_block): On cc0 targets,
17015         do not emit the sequence between cc0 setter and user.
17017 2013-03-25  Kai Tietz  <ktietz@redhat.com>
17019         * config/i386/predicates.md (local_symbolic_operand): Interpret
17020         dll-imported symbols as none-local.
17022 2013-03-25  Richard Biener  <rguenther@suse.de>
17024         * tree-ssa-loop-im.c (struct depend): Remove.
17025         (struct lim_aux_data): Make depends a vec of gimples.
17026         (free_lim_aux_data): Adjust.
17027         (add_dependency): Likewise.
17028         (set_level): Likewise.
17030 2013-03-25  Richard Biener  <rguenther@suse.de>
17032         PR middle-end/56434
17033         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
17034         the pointer returned by calls with ECF_MALLOC set.
17036 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
17038         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
17040 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
17042         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
17043         using MMXMODE mode iterator.
17044         (*move<mode>_internal): Merge with *movv2sf_internal and
17045         *movv2sf_internal_rex64 using MMXMODE mode iterator.
17047 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
17049         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
17050         (record_last_mem_set_info): Likewise.
17052         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
17053         of XNEWVEC followed by memset.
17054         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
17056 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
17058         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
17059         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
17060         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
17061         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
17062         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
17063         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
17064         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
17065         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
17066         BARRIER_P instead of GET_CODE.
17068 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
17070         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
17071         inaccuracy in the probing code.
17073         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
17074         (ctrapdi4): Likewise.
17076 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
17078         * calls.c (expand_call): Add missing guard to code handling return
17079         of non-BLKmode structures in MSB.
17080         * function.c (expand_function_end): Likewise.
17082 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
17084         * combine.c (try_combine): Adjust comment.  Do not add the set of
17085         insn #0 if the destination indirectly is set or dies in insn #2.
17086         Tidy up code to distribute a new note.
17088 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
17090         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
17091         also for alternatives 16 and 17.
17093 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
17095         * config/i386/sse.md (*mov<mode>_internal): Merge with
17096         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
17097         Emit insn template depending on type attribute.  Use
17098         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
17099         movd instead of movq mnemonic for interunit moves.  Rewrite mode
17100         attribute calculation.  Remove unit attribute calculation.
17101         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
17102         Set prefix_data16 attribute for DImode ssemov types.
17103         Use Ym instead of y for SSE-MMX conversion alternatives.
17104         Reorder operand constraints.
17106 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
17108         * df.h (df_insn_delete): Adjust prototype.
17109         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
17110         and let it decide whether mark the basic block dirty.
17111         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
17112         * df-scan.c (df_insn_info_delete): New helper function, split
17113         off from df_insn_delete.
17114         (df_scan_free_bb_info): Use it.
17115         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
17116         Likewise.
17117         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
17118         that the insn is actually an insn and it has a non-NULL basic block.
17119         Do not mark basic block dirty if only deleting a DEBUG_INSN.
17121 2013-03-22  Richard Biener  <rguenther@suse.de>
17123         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
17124         dep_ref members.
17125         (mem_ref_alloc): Do not allocate them.
17126         (refs_independent_p): Do not query or maintain a cache.
17128 2013-03-22  Richard Biener  <rguenther@suse.de>
17130         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
17131         (gather_mem_refs_in_loops): Do not compute it.
17132         (analyze_memory_references): Do not allocate it.
17133         (tree_ssa_lim_finalize): Do not free it.
17134         (for_all_locs_in_loop): Do not query all_refs_in_loop.
17136 2013-03-22  Richard Biener  <rguenther@suse.de>
17138         * is-a.h (as_a): Use gcc_checking_assert.
17140 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
17142         * config/aarch64/aarch64.c (aarch64_print_operand): New
17143         format specifier for printing a constant in hex.
17144         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
17145         format specifier for printing second operand.
17147 2013-03-22  Richard Biener  <rguenther@suse.de>
17149         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
17150         bitmaps.
17151         (gather_mem_refs_in_loops): Perform store accumulation here.
17152         (create_vop_ref_mapping_loop): Remove.
17153         (create_vop_ref_mapping): Likewise.
17154         (analyze_memory_references): Initialize refs_stored_in_loop.
17155         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
17156         (record_indep_loop): Remove.
17157         (record_dep_loop): New function.
17158         (ref_indep_loop_p_1): Adjust to only walk over references
17159         in the loop, not its subloops.
17160         (ref_indep_loop_p): Rename to ...
17161         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
17162         maintaining a more fine-grained cache.
17163         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
17164         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
17166 2013-03-22  Richard Biener  <rguenther@suse.de>
17168         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
17169         (struct mem_ref): Make accesses_in_loop a vec of a vec of
17170         aggregate mem_ref_loc.
17171         (free_mem_ref_locs): Inline into ...
17172         (memref_free): ... this and adjust.
17173         (mem_ref_alloc): Adjust.
17174         (mem_ref_locs_alloc): Remove.
17175         (record_mem_ref_loc): Adjust.
17176         (get_all_locs_in_loop): Rewrite into ...
17177         (for_all_locs_in_loop): ... this iterator.
17178         (rewrite_mem_ref_loc): New functor.
17179         (rewrite_mem_refs): Use for_all_locs_in_loop.
17180         (sm_set_flag_if_changed): New functor.
17181         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
17182         (ref_always_accessed): New functor.
17183         (ref_always_accessed_p): Use for_all_locs_in_loop.
17185 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
17187         * tree-pass.h (PROP_gimple_lvec): New.
17188         * passes.c (dump_properties): Handle PROP_gimple_lvec.
17189         (init_optimization_passes): Move pass_lower_vector.
17190         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
17191         PROP_gimple_lvec.
17192         (pass_lower_vector): Provide PROP_gimple_lvec.
17193         (pass_lower_vector_ssa): Likewise.
17194         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
17196 2013-03-21  Mark Wielaard  <mjw@redhat.com>
17198         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
17200 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
17202         * config/i386/i386.md (*movdi_internal): Disparage slightly
17203         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
17204         conversion alternatives.
17206 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
17208         PR middle-end/48087
17209         * diagnostic.def (DK_WERROR): New kind.
17210         * diagnostic.h (werrorcount): Define.
17211         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
17212         promoted to DK_ERROR, increment DK_WERROR counter instead of
17213         DK_ERROR counter.
17214         * toplev.c (toplev_main): Call print_ignored_options even if
17215         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
17216         even if just werrorcount is non-zero.
17218         PR debug/55608
17219         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
17220         on failure.
17221         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
17222         (string_cst_pool_decl): New function.
17223         (optimize_one_addr_into_implicit_ptr): New function.
17224         (resolve_addr_in_expr): Optimize DWARF location expression
17225         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
17226         which doesn't live in memory, but has DW_AT_location or
17227         DW_AT_const_value, or refers to a string literal, into
17228         DW_OP_GNU_implicit_pointer.
17229         (optimize_location_into_implicit_ptr): New function.
17230         (resolve_addr): If removing DW_AT_location of a variable because
17231         it was DW_OP_addr of address of the variable, but the variable doesn't
17232         live in memory, try to emit const value attribute for the initializer.
17234 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
17236         * tree.h (VECTOR_TYPE_P): New macro.
17237         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
17238         TYPE_MODE): Use it.
17239         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
17240         VEC_COND_EXPR cannot be lvalues.
17241         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
17243 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
17245         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
17246         Restrict the transformation to equal modes.
17248 2013-03-21  Richard Biener  <rguenther@suse.de>
17250         PR tree-optimization/39326
17251         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
17252         (MEM_ANALYZABLE): Adjust.
17253         (record_mem_ref_loc): Move bitmap ops ...
17254         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
17255         unanalyzable refs, do not record locations for it.
17256         (analyze_memory_references): Allocate ref zero as shared
17257         unanalyzable ref.
17258         (refs_independent_p): Do not test for unanalyzed mems here.
17259         (ref_indep_loop_p_1): Special-case disambiguation against
17260         the unanalyzed ref.
17261         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
17263 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
17265         * config/arm/arm-protos.h (tune_params): Add
17266         prefer_neon_for_64bits field.
17267         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
17268         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
17269         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
17270         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
17271         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
17272         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
17273         (arm_option_override): Handle -mneon-for-64bits new option.
17274         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
17275         (prefer_neon_for_64bits): Declare new variable.
17276         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
17277         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
17278         (arch_enabled): Handle new arch types. Remove support for onlya8
17279         and nota8.
17280         (one_cmpldi2): Use new arch names.
17281         (zero_extend<mode>di2, extend<mode>di2): Ditto.
17282         * config/arm/arm.opt (mneon-for-64bits): Add option.
17283         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
17284         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
17285         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
17286         of onlya8.
17287         * doc/invoke.texi (-mneon-for-64bits): Document.
17289 2013-03-21  Richard Biener  <rguenther@suse.de>
17291         PR tree-optimization/39326
17292         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
17293         (sort_bbs_in_loop_postorder_cmp): New function.
17294         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
17296 2013-03-21  Richard Biener  <rguenther@suse.de>
17298         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
17299         (vect_insert_into_interleaving_chain): Likewise.
17300         (vect_drs_dependent_in_basic_block): Inline ...
17301         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
17302         split out from ...
17303         (vect_analyze_data_ref_dependence): ... here.  Simplify.
17304         (vect_check_interleaving): Simplify.
17305         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
17306         (vect_slp_analyze_data_ref_dependences): ... this new function.
17307         (dr_group_sort_cmp): New function.
17308         (vect_analyze_data_ref_accesses): Compute data-reference groups
17309         here instead of in vect_analyze_data_ref_dependence.  Use
17310         a more efficient algorithm.
17311         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
17312         vect_slp_analyze_data_ref_dependences.  Call
17313         vect_analyze_data_ref_accesses earlier.
17314         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
17315         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
17316         (vect_slp_analyze_data_ref_dependences): New prototype.
17318 2013-03-21  Richard Biener  <rguenther@suse.de>
17320         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
17321         ref is stored in the loop.
17322         (find_refs_for_sm): Walk only over all stores.
17323         (store_motion_loop): Allocate from lim_bitmap_obstack.
17324         (store_motion): Likewise.
17326 2013-03-21  Richard Biener  <rguenther@suse.de>
17328         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
17329         Update virtual SSA form.
17331 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17333         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
17334         * configure: Regenerate.
17335         * config.in: Regenerate.
17336         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
17337         if !HAVE_LD_EH_FRAME_CIEV3.
17339 2013-03-21  Richard Biener  <rguenther@suse.de>
17341         * tree-cfg.c (verify_expr_no_block): New function.
17342         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
17343         nor DECL_VALUE_EXPR have locations with associated blocks.
17344         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
17345         (clear_unused_block_pointer): Remove code dealing with
17346         blocks in DECL_DEBUG_EXPR locations.
17348 2013-03-21  Richard Biener  <rguenther@suse.de>
17350         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
17351         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
17352         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
17353         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
17354         instead of DECL_DEBUG_EXPR_IS_FROM.
17355         * gimplify.c (gimplify_modify_expr): Likewise.
17356         * tree-cfg.c (verify_expr_location_1): Likewise.
17357         * tree-complex.c (create_one_component_var): Likewise.
17358         * tree-sra.c (create_access_replacement): Likewise.
17359         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
17360         (clear_unused_block_pointer): Likewise.
17361         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
17362         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
17363         * var-tracking.c (var_debug_decl): Likewise.
17364         (track_expr_p): Likewise.
17365         * tree-inline.c (add_local_variables): Likewise.  Set
17366         DECL_HAS_DEBUG_EXPR_P after copying it.
17367         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
17368         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
17370 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
17372         PR bootstrap/56656
17373         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
17374         * configure: Regenerate.
17375         * config.in: Regenerate.
17376         * config/i386/i386.md (*movdf_internal): Use
17377         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
17378         movd instead of movq mnemonic for interunit moves.
17379         (*movdi_internal): Ditto.
17381 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17383         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
17384         (abd<mode>_3): New pattern.
17385         (aba<mode>_3): New pattern.
17386         (fabd<mode>_3): New pattern.
17388 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17390         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
17391         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
17392         occurrence of REGISTER_PREFIX as its empty string.
17394 2013-03-20  Jeff Law  <law@redhat.com>
17396         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
17397         addititional equivalences for equality comparisons between an SSA_NAME
17398         and a constant where the SSA_NAME was set from a widening conversion.
17400 2013-03-20  Walter Lee  <walt@tilera.com>
17402         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
17404 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
17406         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
17407         depending on type attribute.
17408         (*movti_internal): Ditto.
17409         (*movtf_internal): Ditto.
17410         (*movxf_internal): Ditto.
17411         (*movdf_internal): Ditto.
17412         (*movsf_internal): Ditto.
17414 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
17416         * config/i386/i386.md (*movti_internal): Set prefix attribute to
17417         maybe_vex for sselog1 and ssemov types.
17418         (*movdi_internal): Reorder operand constraints.
17419         (*movsi_internal): Ditto.  Set prefix attribute to
17420         maybe_vex for sselog1 and ssemov types.
17421         (*movtf_internal): Set prefix attribute to maybe_vex
17422         for sselog1 and ssemov types.
17423         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
17424         DImode ssemov types.  Reorder operand constraints.
17425         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
17426         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
17427         attribute for SImode ssemov types.  Reorder operand constraints.
17429 2013-03-20  Martin Jambor  <mjambor@suse.cz>
17431         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
17432         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
17434 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
17436         * config/rs6000/predicates.md (indexed_address, update_address_mem
17437         update_indexed_address_mem): New predicates.
17438         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
17439         attribute for load/store instructions.
17440         * config/rs6000/dfp.md (movsd_store): Likewise.
17441         (movsd_load): Likewise.
17442         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
17443         (unnamed HI->DI extend define_insn): Likewise.
17444         (unnamed SI->DI extend define_insn): Likewise.
17445         (unnamed QI->SI extend define_insn): Likewise.
17446         (unnamed QI->HI extend define_insn): Likewise.
17447         (unnamed HI->SI extend define_insn): Likewise.
17448         (unnamed HI->SI extend define_insn): Likewise.
17449         (extendsfdf2_fpr): Likewise.
17450         (movsi_internal1): Likewise.
17451         (movsi_internal1_single): Likewise.
17452         (movhi_internal): Likewise.
17453         (movqi_internal): Likewise.
17454         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
17455         attribute for load/store instructions.
17456         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
17457         instructions.
17458         (mov<mode>_softfloat): Likewise.
17459         (mov<mode>_hardfloat32): Likewise.
17460         (mov<mode>_hardfloat64): Likewise.
17461         (mov<mode>_softfloat64): Likewise.
17462         (movdi_internal32): Likewise.
17463         (movdi_internal64): Likewise.
17464         (probe_stack_<mode>): Likewise.
17466 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
17468         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
17469         floating point, and decimal floating point to reload iterator.
17471         * config/rs6000/constraints.md (wl constraint): New constraints to
17472         return FLOAT_REGS if certain options are used to reduce the number
17473         of separate patterns that exist in the file.
17474         (wx constraint): Likewise.
17475         (wz constraint): Likewise.
17477         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
17478         -mdebug=reg, print wg, wl, wx, and wz constraints.
17479         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
17480         Initialize the reload functions for 64-bit binary/decimal floating
17481         point types.
17482         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
17483         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
17484         create the buffer on the stack to overcome not having a 32-bit
17485         load and store.
17486         (rs6000_emit_move): Likewise.
17487         (rs6000_secondary_memory_needed_rtx): Likewise.
17488         (rs6000_alloc_sdmode_stack_slot): Likewise.
17489         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
17490         via xxlxor, just like DFmode 0.0.
17492         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
17493         define as 1 if we are running on a power7 or newer.
17494         (enum r6000_reg_class_enum): Add new constraints.
17496         * config/rs6000/dfp.md (movsd): Delete, combine with binary
17497         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
17498         with other moves by using conditional constraits (wg).  Use LFIWZX
17499         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
17500         (movsd splitter): Likewise.
17501         (movsd_hardfloat): Likewise.
17502         (movsd_softfloat): Likewise.
17504         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
17505         binary and decimal floating point moves.
17506         (fmove_ok): New attributes to combine binary and decimal floating
17507         point moves, and to combine power6x (mfpgpr) moves along normal
17508         floating moves.
17509         (real_value_to_target): Likewise.
17510         (f32_lr): Likewise.
17511         (f32_lm): Likewise.
17512         (f32_li): Likewise.
17513         (f32_sr): Likewise.
17514         (f32_sm): Likewise.
17515         (f32_si): Likewise.
17516         (movsf): Combine binary and decimal floating point moves.  Combine
17517         power6x (mfpgpr) moves with other moves by using conditional
17518         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
17519         (mov<mode> for SFmode/SDmode); Likewise.
17520         (SFmode/SDmode splitters): Likewise.
17521         (movsf_hardfloat): Likewise.
17522         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
17523         (movsf_softfloat): Likewise.
17524         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
17526         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
17527         wx and wz constraints.
17529         * config/rs6000/constraints.md (wg constraint): New constraint to
17530         return FLOAT_REGS if -mmfpgpr (power6x) was used.
17532         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
17533         constraint.
17535         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
17536         -mdebug=reg, print wg, wl, wx, and wz constraints.
17537         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
17538         Initialize the reload functions for 64-bit binary/decimal floating
17539         point types.
17540         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
17541         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
17542         create the buffer on the stack to overcome not having a 32-bit
17543         load and store.
17544         (rs6000_emit_move): Likewise.
17545         (rs6000_secondary_memory_needed_rtx): Likewise.
17546         (rs6000_alloc_sdmode_stack_slot): Likewise.
17547         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
17548         via xxlxor, just like DFmode 0.0.
17550         * config/rs6000/dfp.md (movdd): Delete, combine with binary
17551         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
17552         with other moves by using conditional constraits (wg).  Use LFIWZX
17553         and STFIWX for loading SDmode on power7.
17554         (movdd splitters): Likewise.
17555         (movdd_hardfloat32): Likewise.
17556         (movdd_softfloat32): Likewise.
17557         (movdd_hardfloat64_mfpgpr): Likewise.
17558         (movdd_hardfloat64): Likewise.
17559         (movdd_softfloat64): Likewise.
17561         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
17562         64-bit binary and decimal floating point moves.
17563         (FMOVE64X): Likewise.
17564         (movdf): Combine 64-bit binary and decimal floating point moves.
17565         Combine power6x (mfpgpr) moves with other moves by using
17566         conditional constraits (wg).
17567         (mov<mode> for DFmode/DDmode): Likewise.
17568         (DFmode/DDmode splitters): Likewise.
17569         (movdf_hardfloat32): Likewise.
17570         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
17571         (movdf_softfloat32): Likewise.
17572         (movdf_hardfloat64_mfpgpr): Likewise.
17573         (movdf_hardfloat64): Likewise.
17574         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
17575         (movdf_softfloat64): Likewise.
17576         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
17577         (reload_<mode>_load): Move to later in the file so they aren't in
17578         the middle of the floating point move insns.
17579         (reload_<mode>_store): Likewise.
17581         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
17582         constraint.
17584         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
17585         constraint if -mdebug=reg.
17586         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
17587         Enable using dd reload support if needed.
17589         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
17590         binary and decimal floating point moves in rs6000.md.
17591         (movtd_internal): Likewise.
17593         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
17594         decimal floating point moves.
17595         (movtf): Likewise.
17596         (movtf_internal): Likewise.
17597         (mov<mode>_internal, TDmode/TFmode): Likewise.
17598         (movtf_softfloat): Likewise.
17599         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
17601         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
17602         movdi_internal64, using wg constraint for move direct operations.
17603         (movdi_internal64): Likewise.
17605         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
17606         MODES_TIEABLE_P for selected modes.  Print the numerical value of
17607         the various virtual registers. Use GPR/FPR first/last values,
17608         instead of hard coding the register numbers.  Print which modes
17609         have reload functions registered.
17610         (rs6000_option_override_internal): If -mdebug=reg, trace the options
17611         settings before/after setting cpu, target and subtarget settings.
17612         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
17613         and for secondary reload failures in rs6000_secondary_reload_inner.
17614         (rs6000_secondary_reload_fail): Likewise.
17615         (rs6000_secondary_reload_inner): Likewise.
17617         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
17618         macros for first/last GPR and FPR registers.
17619         (LAST_GPR_REGNO): Likewise.
17620         (FIRST_FPR_REGNO): Likewise.
17621         (LAST_FPR_REGNO): Likewise.
17623         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
17624         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
17625         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
17626         (vcond<mode><mode>): Likewise.
17627         (vcondu<mode><mode>): Likewise.
17628         (vector_gtu<mode>): Likewise.
17629         (vector_gte<mode>): Likewise.
17630         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
17631         to prevent the compiler from converting DImode operations to TImode.
17632         (ior<mode>3): Likewise.
17633         (and<mode>3): Likewise.
17634         (one_cmpl<mode>2): Likewise.
17635         (nor<mode>3): Likewise.
17636         (andc<mode>3): Likewise.
17638         * config/rs6000/constraints.md (wt constraint): New constraint
17639         that returns VSX_REGS if TImode is allowed in VSX registers.
17641         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
17642         constant under VSX.
17644         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
17645         similar to TImode, but it is restricted to being in the GPRs.
17647         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
17648         TImode to occupy a single VSX register.
17650         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
17651         -mvsx-timode for power7/power8.
17652         (power7 cpu): Likewise.
17653         (power8 cpu): Likewise.
17655         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
17656         sure that TFmode/TDmode take up two registers if they are ever
17657         allowed in the upper VSX registers.
17658         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
17659         registers.
17660         (rs6000_init_hard_regno_mode_ok): Likewise.
17661         (rs6000_debug_reg_global): Add debugging for PTImode and wt
17662         constraint.  Print if LRA is turned on.
17663         (rs6000_option_override_internal): Give an error if -mvsx-timode
17664         and VSX is not enabled.
17665         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
17666         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
17667         to reg+offset addressing.  Use PTImode when checking offset
17668         addresses for validity.
17669         (reg_offset_addressing_ok_p): Likewise.
17670         (rs6000_legitimate_offset_address_p): Likewise.
17671         (rs6000_legitimize_address): Likewise.
17672         (rs6000_legitimize_reload_address): Likewise.
17673         (rs6000_legitimate_address_p): Likewise.
17674         (rs6000_eliminate_indexed_memrefs): Likewise.
17675         (rs6000_emit_move): Likewise.
17676         (rs6000_secondary_reload): Likewise.
17677         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
17678         reloads to fpr registers to continue to use reg+offset addressing,
17679         but 64-bit reloads to altivec registers need reg+reg addressing.
17680         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
17681         it.  Treat LO_SUM like a PLUS operation.
17682         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
17683         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
17684         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
17685         registers to share a register with a smaller sized type, since VSX
17686         puts scalars in the upper 64-bits.
17687         (print_operand): Add support for PTImode.
17688         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
17689         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
17690         registers, but don't have arithmetic support.
17691         (rs6000_memory_move_cost): Add test for VSX.
17692         (rs6000_opt_masks): Add -mvsx-timode.
17694         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
17695         for TImode.
17696         (VSs): Likewise.
17697         (VSr): Use wt constraint for TImode.
17698         (VSv): Drop TImode support.
17699         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
17700         (vsx_movti_64bit): Likewise.
17701         (vsx_movti_32bit): Likewise.
17702         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
17703         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
17704         one '?' on the appropriate output constraint.  Do not allow TImode
17705         logical operations on 32-bit systems.
17706         (vsx_ior<mode>3): Likewise.
17707         (vsx_xor<mode>3): Likewise.
17708         (vsx_one_cmpl<mode>2): Likewise.
17709         (vsx_nor<mode>3): Likewise.
17710         (vsx_andc<mode>3): Likewise.
17711         (vsx_concat_<mode>): Likewise.
17712         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
17714         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
17715         OPTION_MASK_VSX_TIMODE.
17716         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
17717         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
17719         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
17720         (TI2 iterator): New iterator for TImode, PTImode.
17721         (wd mode attribute): Add values for vector types.
17722         (movti_string): Replace TI move operations with operations for TImode
17723         and PTImode.  Add support for TImode being allowed in VSX registers.
17724         (mov<mode>_string, TImode/PTImode): Likewise.
17725         (movti_ppc64): Likewise.
17726         (mov<mode>_ppc64, TImode/PTImode): Likewise.
17727         (TI mode splitters): Likewise.
17729         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
17730         constraint.
17732 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
17734         PR tree-optimization/56355
17735         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
17736         Also handle integers with undefined overflow.
17738 2013-03-20  Catherine Moore  <clm@codesourcery.com>
17739             Maciej W. Rozycki  <macro@codesourcery.com>
17740             Tom de Vries  <tom@codesourcery.com>
17741             Nathan Sidwell  <nathan@codesourcery.com>
17742             Iain Sandoe  <iain@codesourcery.com>
17743             Nathan Froyd  <froydnj@codesourcery.com>
17744             Chao-ying Fu  <fu@mips.com>
17746         * doc/extend.texi: (micromips, nomicromips, nocompression):
17747         Document new function attributes.
17748         * doc/invoke.texi (minterlink-compressed, mmicromips,
17749         m14k, m14ke, m14kec): Document new options.
17750         (minterlink-mips16): Update documentation.
17751         * doc/md.texi (ZC, ZD): Document new constraints.
17752         * configure.ac (gcc_cv_as_micromips): Check if linker
17753         supports the .set micromips directive.
17754         * configure: Regenerate.
17755         * config.in: Regenerate.
17756         * config/mips/mips-tables.opt: Regenerate.
17757         * config/mips/micromips.md: New file.
17758         * constraints.md (ZC, ZD): New constraints.
17759         * config/mips/predicates.md (movep_src_register): New predicate.
17760         (movep_src_operand): New predicate.
17761         (non_volatile_mem_operand): New predicate.
17762         * config/mips/mips.md (multimem): New type.
17763         (length): Differentiate between 17-bit and 18-bit branch offsets.
17764         (MOVEP1, MOVEP2): New mode iterator.
17765         (mov_<load>l): Use ZC constraint.
17766         (mov_<load>r): Likewise.
17767         (mov_<store>l): Likewise.
17768         (mov_<store>r): Likewise.
17769         (*branch_equality<mode>_inverted): Add microMIPS support.
17770         (*branch_equality<mode>): Likewise.
17771         (*jump_absolute): Likewise.
17772         (indirect_jump_<mode>): Likewise.
17773         (tablejump_<mode>): Likewise.
17774         (<optab>_internal): Likewise.
17775         (sibcall_internal): Likewise.
17776         (sibcall_value_internal): Likewise.
17777         (prefetch): Use constraint ZD.
17778         * config/mips/mips.opt (minterlink-compressed): New option.
17779         (minterlink-mips16): Now an alias for minterlink-compressed.
17780         (mmicromips): New option.
17781         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
17782         (compare_and_swap_12): Likewise.
17783         (sync_add<mode>): Likewise.
17784         (sync_<optab>_12): Likewise.
17785         (sync_old_<optab>_12): Likewise.
17786         (sync_new_<optab>_12): Likewise.
17787         (sync_nand_12): Likewise.
17788         (sync_old_nand_12): Likewise.
17789         (sync_new_nand_12): Likewise.
17790         (sync_sub<mode>): Likewise.
17791         (sync_old_add<mode>): Likewise.
17792         (sync_old_sub<mode>): Likewise.
17793         (sync_new_add<mode>): Likewise.
17794         (sync_new_sub<mode>): Likewise.
17795         (sync_<optab><mode>): Likewise.
17796         (sync_old_<optab><mode>): Likewise.
17797         (sync_new_<optab><mode>): Likewise.
17798         (sync_nand<mode>): Likewise.
17799         (sync_old_nand<mode>): Likewise.
17800         (sync_new_nand<mode>): Likewise.
17801         (sync_lock_test_and_set<mode>): Likewise.
17802         (test_and_set_12): Likewise.
17803         (atomic_compare_and_swap<mode>): Likewise.
17804         (atomic_exchange<mode>_llsc): Likewise.
17805         (atomic_fetch_add<mode>_llsc): Likewise.
17806         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
17807         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
17808         (umips_save_restore_pattern_p): Likewise.
17809         (umips_load_store_pair_p): Likewise.
17810         (umips_output_load_store_pair): Likewise.
17811         (umips_movep_target_p): Likewise.
17812         (umips_12bit_offset_address_p): Likewise.
17813         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
17814         (mips_base_mips16): Rename this...
17815         (mips_base_compression_flags): ...to this. Update all uses.
17816         (mips_attribute_table): Add micromips, nomicromips and nocompression.
17817         (mips_mips16_decl_p): Delete.
17818         (mips_nomips16_decl_p): Delete.
17819         (mips_get_compress_on_flags): New function.
17820         (mips_get_compress_off_flags): New function.
17821         (mips_get_compress_mode): New function.
17822         (mips_get_compress_on_name): New function.
17823         (mips_get_compress_off_name): New function.
17824         (mips_insert_attributes): Support multiple compression types.
17825         (mips_merge_decl_attributes): Likewise.
17826         (umips_12bit_offset_address_p): New function.
17827         (mips_start_function_definition): Emit .set micromips directive.
17828         (mips_call_may_need_jalx_p): New function.
17829         (mips_function_ok_for_sibcall): Add microMIPS support.
17830         (mips_print_operand_punctuation): Support short delay slots and
17831         compact jumps.
17832         (umips_swm_mask, umips_swm_encoding): New.
17833         (umips_build_save_restore): New function.
17834         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
17835         (was_mips16_p): Remove.
17836         (old_compression_mode): New.
17837         (mips_set_compression_mode): New function.
17838         (mips_set_current_function): Add microMIPS support.
17839         (mips_option_override): Likewise.
17840         (umips_save_restore_pattern_p): New function.
17841         (umips_output_save_restore): New function.
17842         (umips_load_store_pair_p_1): New function.
17843         (umips_load_store_pair_p): New function.
17844         (umips_output_load_store_pair_1): New function.
17845         (umips_output_load_store_pair): New function.
17846         (umips_movep_target_p) New function.
17847         (mips_prepare_pch_save): Add microMIPS support.
17848         * config/mips/mips.h (TARGET_COMPRESSION): New.
17849         (TARGET_CPU_CPP_BUILTINS): Update macro
17850         to use new compression flags and to support microMIPS.
17851         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
17852         (MIPS_ARCH_FLOAT_SPEC): Likewise.
17853         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
17854         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
17855         (ASM_SPEC): Support mmicromips and mno-micromips.
17856         (M16STORE_REG_P): New macro.
17857         (MIPS_CALL): Support TARGET_MICROMIPS.
17858         (MICROMIPS_J): New macro.
17859         (mips_base_mips16): Rename this...
17860         (mips_base_compression_flags): ...to this.
17861         (UMIPS_12BIT_OFFSET_P): New macro.
17862         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
17863         (MULTILIB_DIRNAMES): Likewise.
17864 2013-03-20  Richard Biener  <rguenther@suse.de>
17866         PR tree-optimization/56661
17867         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
17868         the result does not have to be distinct.
17870 2013-03-20  Richard Biener  <rguenther@suse.de>
17872         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
17873         remap_gimple_op_r.
17875 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17876             Steven Bosscher  <steven@gcc.gnu.org>
17878         PR rtl-optimization/56605
17879         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
17881 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
17883         PR bootstrap/56656
17884         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
17885         that require movd instead of movq.
17887 2013-03-20  Richard Biener  <rguenther@suse.de>
17889         * tree-ssa-structalias.c (struct variable_info): Add pointer
17890         to the first field of an aggregate with sub-vars.  Make
17891         this and the pointer to the next subfield its ID.
17892         (vi_next): New function.
17893         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
17894         storedanything_id, integer_id): Increment by one.
17895         (new_var_info, get_call_vi, lookup_call_clobber_vi,
17896         get_call_clobber_vi): Adjust.
17897         (solution_set_expand): Simplify and speedup.
17898         (solution_set_add): Inline into ...
17899         (set_union_with_increment): ... this.  Adjust accordingly.
17900         (do_sd_constraint): Likewise.
17901         (do_ds_constraint): Likewise.
17902         (do_complex_constraint): Simplify.
17903         (build_pred_graph): Adjust.
17904         (solve_graph): Likewise.  Simplify and speedup.
17905         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
17906         get_constraint_for_component_ref, get_constraint_for_1,
17907         first_vi_for_offset, first_or_preceding_vi_for_offset,
17908         create_function_info_for, create_variable_info_for_1,
17909         create_variable_info_for, intra_create_variable_infos): Adjust.
17910         (init_base_vars): Push NULL for ID zero.
17911         (compute_points_to_sets): Adjust.
17913 2013-03-20  Richard Biener  <rguenther@suse.de>
17915         * cfgloop.c (verify_loop_structure): Streamline and avoid
17916         ICEing on corrupt loop tree.
17917         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
17918         loop tree.
17920 2013-03-20  Richard Biener  <rguenther@suse.de>
17922         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
17923         check whether an SSA update is needed.
17925 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
17927         * config/mips/constraints.md (T): Rename to...
17928         (Yf): ...this.
17929         (U): Rename to...
17930         (Yd): ...this.
17931         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
17932         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
17934 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
17936         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
17937         (*subsi3_carryin_uxtw): Likewise.
17939 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
17941         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
17942         (*rorsi3_insn_uxtw): Likewise.
17944 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
17946         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
17947         (*extrsi5_insn_uxtw): Likewise.
17949 2013-03-19  Richard Biener  <rguenther@suse.de>
17951         PR tree-optimization/56273
17952         * passes.c (init_optimization_passes): Move second VRP after DOM.
17954 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
17956         * config/i386/i386.md (*movti_internal): Merge from
17957         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
17958         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
17959         nox64 isa attributes.
17961 2013-03-18  Richard Biener  <rguenther@suse.de>
17963         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
17964         (unite): Likewise.
17965         (merge_node_constraints): Likewise.
17966         (build_succ_graph): Likewise.
17967         (valid_graph_edge): Inline into single caller.
17968         (unify_nodes): Likewise.  Use bitmap_set_bit return value
17969         and cache varinfo.
17970         (scc_visit): Fix formatting and variable use.
17971         (do_sd_constraint): Use gcc_checking_assert.
17972         (do_ds_constraint): Likewise.
17973         (do_complex_constraint): Likewise.
17974         (condense_visit): Likewise.  Cleanup.
17975         (dump_pred_graph): New function.
17976         (perform_var_substitution): Dump the pred-graph before
17977         variable substitution.
17978         (find_equivalent_node): Use gcc_checking_assert.
17979         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
17981 2013-03-18  Richard Biener  <rguenther@suse.de>
17983         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
17984         Remove cond_expr_stmt_list argument and do not gimplify the
17985         built expression.
17986         (vect_loop_versioning): Adjust.
17987         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
17988         Cleanup to use less temporaries.
17989         (vect_create_data_ref_ptr): Cleanup.
17991 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
17993         PR tree-optimization/56635
17994         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
17995         require types_compatible_p types.
17997 2013-03-18  Nick Clifton  <nickc@redhat.com>
17999         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
18000         spurious backslash.
18002         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
18003         Add missing line to comment describing function.
18005 2013-03-18  Richard Biener  <rguenther@suse.de>
18007         PR tree-optimization/56210
18008         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18009         Handle string / character search functions.
18010         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
18012 2013-03-18  Richard Biener  <rguenther@suse.de>
18014         PR middle-end/56483
18015         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
18016         and implement properly.
18017         * gimple.h (gimple_cond_single_var_p): Remove.
18019 2013-03-18  Richard Biener  <rguenther@suse.de>
18021         * tree-data-ref.h (find_data_references_in_loop): Declare.
18022         * tree-data-ref.c (get_references_in_stmt): Use a stack
18023         vector pre-allocated in the callers.
18024         (find_data_references_in_stmt): Adjust.
18025         (graphite_find_data_references_in_stmt): Likewise.
18026         (create_rdg_vertices): Likewise.
18027         (find_data_references_in_loop): Export.
18028         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
18029         Compute dependences here...
18030         (vect_analyze_data_refs): ...not here.  When we encounter
18031         a non-vectorizable data reference in basic-block vectorization
18032         truncate the data reference vector.  Do not bother to
18033         fixup data-dependence information for gather loads.
18034         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
18035         of data references, as reported.
18037 2013-03-18  Richard Biener  <rguenther@suse.de>
18039         PR tree-optimization/3713
18040         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
18041         has_constants and expr.
18042         (stmt_has_constants): Properly valueize SSA names when deciding
18043         whether the stmt has constants.
18045 2013-03-18  Richard Biener  <rguenther@suse.de>
18047         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
18048         whole function when there is nothing to do.
18049         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
18050         * tree-vectorizer.c (vectorize_loops): Update virtual and
18051         loop-closed SSA once.
18052         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
18054 2013-03-18  Richard Biener  <rguenther@suse.de>
18056         PR middle-end/56113
18057         * domwalk.c (bb_postorder): New global static.
18058         (cmp_bb_postorder): New function.
18059         (walk_dominator_tree): Replace scheme imposing an order for
18060         visiting dominator sons by one sorting them at the time they
18061         are pushed on the stack.
18063 2013-03-18  Richard Biener  <rguenther@suse.de>
18065         PR tree-optimization/39326
18066         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
18067         (struct mem_ref): Replace mem member with ao_ref typed member.
18068         (MEM_ANALYZABLE): Adjust.
18069         (memref_eq): Likewise.
18070         (mem_ref_alloc): Likewise.
18071         (gather_mem_refs_stmt): Likewise.
18072         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
18073         (execute_sm_if_changed_flag_set): Adjust.
18074         (execute_sm): Likewise.
18075         (ref_always_accessed_p): Likewise.
18076         (refs_independent_p): Likewise.
18077         (can_sm_ref_p): Likewise.
18079 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
18081         PR c/56566
18082         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
18083         return 1 even for !unsignedp.
18085 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
18087         * config/i386/i386.md (isa): Add x64 and nox64.
18088         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
18089         (*pushtf): Enable *roF alternative for x64 isa only.
18090         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
18091         mode attribute of integer alternatives to DImode for TARGET_64BIT.
18092         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
18093         (*movtf_internal): Merge from *movtf_internal_rex64 and
18094         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
18095         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
18096         nox64 isa attributes.
18097         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
18098         nox64 isa attributes.
18099         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
18101 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
18103         * config/alpha/alpha.c (TARGET_LRA_P): New define.
18105 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
18107         PR target/56640
18108         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
18109         class names.  Remove trailing comma after "ALL_REGS".
18111 2013-03-16  Jan Hubicka  <jh@suse.cz>
18113         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
18114         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
18115         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
18116         of cgraph_get_create_node.
18117         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
18119 2013-03-16  Jason Merrill  <jason@redhat.com>
18121         PR debug/49090
18122         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
18123         with DW_AT_default_value.
18125 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
18127         * BASE-VER: Set to 4.9.0.
18129 2013-03-14  Andi Kleen  <ak@linux.intel.com>
18131         PR target/56619
18132         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
18133         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
18134         Document _x* TSX intrinsics.
18136 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
18137             David Holsgrove  <david.holsgrove@xilinx.com>
18139         * configure.ac: Add MicroBlaze TLS support detection.
18140         * configure: Regenerate.
18141         * config/microblaze/microblaze-protos.h
18142         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
18143         symbol_mentioned_p, label_mentioned_p): Add prototypes.
18144         * config/microblaze/microblaze.c (microblaze_address_type): Add
18145         ADDRESS_TLS and tls_reloc address types.
18146         (microblaze_address_info): Add tls_reloc.
18147         (TARGET_HAVE_TLS): Define.
18148         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
18149          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
18150          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
18151         load_tls_operand,  microblaze_call_tls_get_addr,
18152         microblaze_legitimize_tls_address): New functions.
18153         (microblaze_classify_unspec): Handle UNSPEC_TLS.
18154         (get_base_reg): Use microblaze_tls_symbol_p.
18155         (microblaze_classify_address): Handle TLS.
18156         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
18157         label_mentioned_p and microblaze_tls_referenced_p.
18158         (microblaze_legitimize_address): Handle TLS.
18159         (microblaze_address_insns): Handle ADDRESS_TLS.
18160         (pic_address_needs_scratch): Handle TLS.
18161         (print_operand_address): Handle TLS.
18162         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
18163         (microblaze_expand_move): Handle TLS.
18164         (microblaze_legitimate_constant_p): Check
18165         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
18166         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
18167         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
18168         (PIC_OFFSET_TABLE_REGNUM): Set.
18169         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
18170         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
18171         (addsi3, movsi_internal2, movdf_internal): Update constraints
18172         * config/microblaze/predicates.md (arith_plus_operand): Define
18173         (move_operand): Redefine as move_src_operand,
18174         check microblaze_tls_referenced_p.
18176 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
18178         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
18179         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
18181 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
18183         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
18184         CC mode for AND.
18186 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
18188         PR tree-optimization/53265
18189         * common.opt (Waggressive-loop-optimizations): New option.
18190         * tree-ssa-loop-niter.c: Include tree-pass.h.
18191         (do_warn_aggressive_loop_optimizations): New function.
18192         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
18193         if number_of_latch_executions returned constant.
18194         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
18195         early.  If number_of_latch_executions returned constant, set
18196         nb_iterations_upper_bound back to it.
18197         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
18198         field.
18199         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
18200         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
18202         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
18203         (MULTILIB_OSDIRNAMES): Set.
18204         * genmultilib: If defaultosdirname doesn't start with :: , set
18205         defaultosdirname2 instead, clear it and emit two . multilib_raw
18206         entries instead of just one.
18208 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
18210         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
18211         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
18212         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
18213         (SUBTARGET_OVERRIDE_OPTIONS): New.
18215 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
18217         PR target/49880
18218         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
18219         (musermode): Convert to Var(TARGET_USERMODE).
18220         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
18221         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
18222         * config/sh/sh.c (sh_option_override): Use
18223         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
18224         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
18225         condition.
18226         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
18227         TARGET_SH4.
18228         (udivsi3_i4_single, divsi3_i4_single): Use
18229         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
18231 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
18233         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
18234         default setting.
18236 2013-03-13  Richard Biener  <rguenther@suse.de>
18238         PR tree-optimization/56608
18239         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
18240         calls when vectorizing basic-blocks.
18242 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
18244         PR plugins/45078
18245         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
18246         tm_file.
18248 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
18250         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
18252 2013-03-11  Jan Hubicka  <jh@suse.cz>
18254         PR lto/56557
18255         * lto-streamer-out.c (output_symbol_p): Skip references from
18256         constructors of external variables.
18258 2013-03-11  Jan Hubicka  <jh@suse.cz>
18260         PR middle-end/56571
18261         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
18262         from pseudos.
18263         * emit-rtl.c (verify_rtx_sharing): Likewise.
18264         (copy_insn_1): Likewise.
18265         * rtl.c (copy_rtx): Likewise.
18267 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
18269         PR target/56591
18270         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
18271         output_operand_lossage message.
18273 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
18275         PR target/56470
18276         * arm.c (shift_op): Validate RTL pattern on the fly.
18277         (arm_print_operand, case 'S'): Don't use shift_operator to validate
18278         the RTL.
18280 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18282         PR target/56347
18283         * config/pa/pa.md (call_value): Check for calls to powf and direct to
18284         new call patterns that clobber %fr12.
18285         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
18286         split and postreload patterns.
18287         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
18288         registers %fr12 and %fr12R as call used.
18290 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
18292         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
18293         (canon_address, record_store, replace_read, check_mem_read_rtx,
18294         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
18295         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
18296         rest_of_handle_dse): Likewise.
18298 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
18300         PR middle-end/56524
18301         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
18302         Add base_optabs.
18303         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
18304         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
18305         (save_optabs_if_changed): Replace with...
18306         (init_tree_optimization_optabs): ...this.
18307         * optabs.c (save_optabs_if_changed): Rename to...
18308         (init_tree_optimization_optabs): ...this.  Take the optimization node
18309         as argument.  Do nothing if the base optabs are already correct.
18310         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
18311         to recompute optabs.
18312         * function.h (function): Remove optabs field.
18313         * function.c (invoke_set_current_function_hook): Call
18314         init_tree_optimization_optabs.  Use the result to initialize
18315         this_fn_optabs.
18317 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
18319         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
18320         if GTMA_HAS_NO_INSTRUMENTATION.
18321         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
18322         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
18323         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
18324         * gimple-pretty-print.c (dump_gimple_transaction): Handle
18325         GTMA_HAS_NO_INSTRUMENTATION.
18327 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
18329         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
18330         libasan_preinit.o.
18332 2013-03-08  Marek Polacek  <polacek@redhat.com>
18333             Jakub Jelinek  <jakub@redhat.com>
18335         PR tree-optimization/56478
18336         * predict.c (is_comparison_with_loop_invariant_p): Change the
18337         type of loop_step to tree.
18338         (predict_loops): Adjust.
18339         (predict_iv_comparison): Perform the computations on double_ints.
18341 2013-03-08  Richard Biener  <rguenther@suse.de>
18343         PR tree-optimization/56570
18344         * tree-cfg.c (verify_expr_location_1): Verify locations for
18345         DECL_DEBUG_EXPR.
18346         * tree-sra.c (create_access_replacement): Strip locations
18347         from DECL_DEBUG_EXPRs.
18349 2013-03-08  Richard Biener  <rguenther@suse.de>
18351         * tree-inline.c (expand_call_inline): Do not associate
18352         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
18353         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
18355 2013-03-08  Richard Biener  <rguenther@suse.de>
18357         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
18358         or block changes with -Og.  Fix for location / block encoding
18359         changes and PHI arguments with locations.
18361 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
18363         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
18364         for all counters.
18365         (struct output_info): Likewise.
18366         (register_overhead): Remove bad gcc_assert.
18367         (bitmap_find_bit): If there is only a single bitmap element, do not
18368         count a miss as a search.
18369         (print_statistics): Update for counter type changes.
18370         (dump_bitmap_statistics): Likewise.  Print headers such that they
18371         are properly lined up with the printed counters.
18373 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
18375         PR tree-optimization/56559
18376         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
18377         check that it has only a single use.
18379 2013-03-07  Richard Biener  <rguenther@suse.de>
18381         * doc/invoke.texi (fwhole-program): Discourage use in combination
18382         with -flto.
18384 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
18386         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
18388         PR tree-optimization/56539
18389         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
18390         instead of GSI_CONTINUE_LINKING as last argument to
18391         force_gimple_operand_gsi.  Adjust function comment.
18393         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
18394         aarch64-cores.def.
18396         PR middle-end/56548
18397         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
18398         promoted mode, convert the result back to the original mode.
18400 2013-03-06  Richard Biener  <rguenther@suse.de>
18402         PR middle-end/56294
18403         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
18404         (insert_updated_phi_nodes_compare_uids): New function.
18405         (update_ssa): Sort symbols_to_rename after UID before
18406         traversing it to insert PHI nodes.
18408 2013-03-06  Richard Biener  <rguenther@suse.de>
18410         PR middle-end/50494
18411         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
18412         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
18414         Revert
18415         2013-02-13  Richard Biener  <rguenther@suse.de>
18417         PR lto/50494
18418         * varasm.c (output_constant_def_1): Get the decl representing
18419         the constant as argument.
18420         (output_constant_def): Wrap output_constant_def_1.
18421         (make_decl_rtl): Use output_constant_def_1 with the decl
18422         representing the constant.
18423         (build_constant_desc): Optionally re-use a decl already
18424         representing the constant.
18425         (tree_output_constant_def): Adjust.
18427 2013-03-06  Joey Ye  <joey.ye@arm.com>
18429         PR lto/50293
18430         * gcc.c (convert_white_space): New function.
18431         (main): Handles white space in function name.
18433 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
18435         PR target/56529
18436         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
18437         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
18438         to SH_DIV_CALL_TABLE for TARGET_SH2.
18439         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
18440         list.
18441         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
18442         call-table options.
18444 2013-03-05  Sterling Augustine  <saugustine@google.com>
18445             Cary Coutant  <ccoutant@google.com>
18447         PR debug/55364
18448         * dwarf2out.c (resolve_addr): Don't call
18449         remove_loc_list_addr_table_entries a second time for the same
18450         expression.
18452 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18454         PR debug/56510
18455         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
18456         (avoid_complex_debug_insns): New function.
18457         (expand_debug_locations): Call it.
18459         PR rtl-optimization/56484
18460         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
18461         lifetimes of hard registers on small register class machines.
18463 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
18465         * config/microblaze/microblaze-protos.h: Rename
18466         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
18467         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
18468         fast_interrupt.
18469         (microblaze_fast_interrupt_function_p): New function.
18470         (microblaze_is_interrupt_handler): Rename to
18471         microblaze_is_interrupt_variant and add fast_interrupt check.
18472         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
18473         (save_restore_insns): Likewise.
18474         (compute_frame_size): Likewise.
18475         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
18476         (microblaze_globalize_label): Likewise.
18477         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
18478         * config/microblaze/microblaze.md: Use wrapper
18479         microblaze_is_interrupt_variant.
18481 2013-03-05  Kai Tietz  <ktietz@redhat.com>
18483         * sdbout.c (sdbout_one_type): Switch to current function's section
18484         supporting cold/hot.
18486 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
18488         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
18489         -mxl-reorder.
18491 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18493         PR middle-end/56461
18494         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
18495         if VALGRIND_GET_VBITS is defined, temporarily make object
18496         memory all defined, and restore previous valgrind addressability
18497         and definability afterwards.  Free this_object at the end.
18499         PR middle-end/56461
18500         * lra.c (lra): Call lra_clear_live_ranges if live_p,
18501         right before calling lra_create_live_ranges, also call it
18502         when clearing live_p.  Only call lra_clear_live_ranges
18503         at the end if live_p.
18505         PR middle-end/56461
18506         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
18508 2013-03-05  Richard Biener  <rguenther@suse.de>
18510         PR tree-optimization/56521
18511         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
18512         value-id.
18514 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
18516         PR c++/55135
18517         * except.h (remove_unreachable_eh_regions): New prototype.
18518         * except.c (remove_eh_handler_splicer): New function, split out
18519         of remove_eh_handler.
18520         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
18521         warning about running it on many EH regions one at a time.
18522         (remove_unreachable_eh_regions_worker): New function, walk the
18523         EH tree in depth-first order and remove non-marked regions.
18524         (remove_unreachable_eh_regions): New function.
18525         * tree-eh.c (mark_reachable_handlers): New function, split out
18526         from remove_unreachable_handlers.
18527         (remove_unreachable_handlers): Use mark_reachable_handlers and
18528         remove_unreachable_eh_regions.
18529         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
18530         and remove_unreachable_eh_regions.
18532 2013-03-05  Richard Biener  <rguenther@suse.de>
18534         PR middle-end/56525
18535         * loop-init.c (fix_loop_structure): Remove loops in two stages,
18536         not freeing them until the end.
18538 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
18540         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
18542 2013-03-05  Richard Biener  <rguenther@suse.de>
18544         PR tree-optimization/56270
18545         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
18546         of loads after scheduling an SLP instance.
18548 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18550         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
18551         tic6x.exp.
18552         (check_gcc_parallelize): Run guality.exp as a separate job from
18553         vect.exp with unsorted.exp and $(dg_target_exps) separately from
18554         struct-layout-1.exp with stackalign.exp.
18556         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
18558         PR middle-end/56461
18559         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
18560         load_index sbitmap even if some bit in it isn't set.
18562         PR middle-end/56461
18563         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
18564         (discover_iteration_bound_by_body_walk): Change queues to
18565         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
18566         spelling in comment.  Call safe_push on queues[bound_index] directly.
18567         Release queues[queue_index] in every iteration unconditionally.
18568         Release bounds vector.
18570         PR middle-end/56461
18571         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
18572         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
18573         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
18574         inner_phis vector.
18576 2013-03-05  Richard Biener  <rguenther@suse.de>
18578         PR lto/56515
18579         * tree-inline.c (remap_blocks_to_null): New function.
18580         (expand_call_inline): When expanding a call stmt without
18581         an associated block inline remap all callee blocks to NULL.
18583 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18585         PR rtl-optimization/56494
18586         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
18587         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
18588         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
18590         PR middle-end/56461
18591         * sel-sched-ir.c (free_sched_pools): Release
18592         succs_info_pool.stack[succs_info_pool.max_top] vectors too
18593         if succs_info_pool.max_top isn't -1.
18595         PR bootstrap/56509
18596         * opts.c (opts_obstack, opts_concat): Moved to...
18597         * opts-common.c (opts_obstack, opts_concat): ... here.
18599 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
18601         PR middle-end/56461
18602         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
18604 2013-03-04  Martin Jambor  <mjambor@suse.cz>
18606         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
18607         all appropriate places.
18609 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
18611         PR tree-optimization/56424
18612         * ipa-split.c (split_function): Do not set the RSO flag if result is
18613         not by reference and its type is a register type.
18615 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
18617         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
18618         (microblaze_legitimate_pic_operand): Likewise
18619         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
18620         new function microblaze_legitimate_pic_operand
18621         * config/microblaze/microblaze-protos.h
18622         (microblaze_legitimate_pic_operand): Declare.
18624 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18626         * config/microblaze/predicates.md (call_insn_simple_operand):
18627         New predicate for supported rtx code types.
18628         * config/microblaze/microblaze.md (call_internal1): Use
18629         call_insn_simple_operand predicate.
18631 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
18633         PR middle-end/56461
18634         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
18635         partitions.ordered_remove.
18637         PR middle-end/56461
18638         * tree-vect-stmts.c (vectorizable_conversion): Don't call
18639         vec_oprnds0.create (1) for modifier == NONE.
18641         PR middle-end/56461
18642         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
18643         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
18644         vec_oprnds1 right before pushing anything to it for
18645         scalar_shift_arg.
18647         PR middle-end/56461
18648         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
18649         set nbbs to 0 instead of having separate code path.
18650         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
18651         instead of false as last argument if returning NULL.
18653 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
18655         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
18656         the attribute is now called "target" instead of "option".
18657         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
18658         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
18659         attribute/pragma name for TARGET_OPTION_VALID_P and
18660         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
18661         * doc/tm.texi: Regenerated.
18663 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
18665         * config/microblaze/microblaze.c:
18666         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
18667         * config/microblaze/microblaze.h: Add -mxl-reorder to
18668         DRIVER_SELF_SPECS.
18669         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
18670         instructions emitted if TARGET_REORDER.
18671         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
18672         or 0 for -m/-mno case, but initialises as 2 to detect default use case
18673         separately.
18675 2013-03-01  Xinliang David Li  <davidxl@google.com>
18677         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
18678         walk length.
18680 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
18682         PR middle-end/56461
18683         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
18684         vector even when returning true.  Fix up function comment formatting.
18686         PR middle-end/56461
18687         * ira-build.c (ira_loop_nodes_count): New variable.
18688         (create_loop_tree_nodes): Initialize it.
18689         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
18691         PR middle-end/56461
18692         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
18693         method on dr_chain and result_chain.
18694         * tree-vect-stmts.c (vectorizable_store): Only call
18695         result_chain.create if j == 0.
18697         PR middle-end/56461
18698         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
18699         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
18700         before overwriting it.
18702 2013-03-01  Tobias Burnus  <burnus@net-b.de>
18704         * doc/extended.texi (C Extensions): Change order in @menu
18705         to match @node.
18706         (Other MIPS Built-in Functions): Move last MIPS entry before
18707         "picoChip Built-in Functions".
18708         (SH Built-in Functions): Move after RX Built-in Functions.
18709         * doc/gcc.texi (Introduction): Change order in @menu
18710         to match @node.
18711         * doc/md.texi (Constraints): Ditto.
18712         * gty.texi (Type Information): Ditto.
18713         (User-provided marking routines for template types): Make
18714         subsection.
18715         * doc/invoke.texi (AArch64 Options): Move before
18716         "Adapteva Epiphany Options".
18718 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
18719             Jakub Jelinek  <jakub@redhat.com>
18721         PR sanitizer/56454
18722         * asan.c (gate_asan): Lookup no_sanitize_address instead of
18723         no_address_safety_analysis attribute.
18724         * doc/extend.texi (no_address_safety_attribute): Rename to
18725         no_sanitize_address attribute, mention no_address_safety_analysis
18726         attribute as deprecated alias.
18728 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
18730         PR middle-end/56461
18731         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
18732         type to vec<vec<tree> > *.
18733         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
18734         to be vec<tree> instead of vec<tree> *, set vec_defs
18735         to vNULL and call vec_defs.create (number_of_vects), adjust other
18736         uses of vec_defs.
18737         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
18738         vectorizable_condition): Adjust vect_get_slp_defs callers.
18740 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
18742         * config/aarch64/aarch64.c
18743         (aarch64_float_const_representable): Remove unused variable.
18745 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
18747         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
18749 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
18751         * config/aarch64/aarch64-builtins.c
18752         (aarch64_init_simd_builtins): Make static.
18754 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
18756         * config/aarch64/aarch64.c
18757         (aarch64_simd_make_constant): Make static.
18759 2013-02-28  Martin Jambor  <mjambor@suse.cz>
18761         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
18762         with no initialization to the RHS of debug statements.
18764 2013-02-28  Martin Jambor  <mjambor@suse.cz>
18766         PR tree-optimization/56294
18767         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
18768         Adjust dumping.
18769         (get_access_replacement): Do not call create_access_replacement.
18770         Assert a replacement exists.
18771         (get_repl_default_def_ssa_name): Create the replacement declaration
18772         itself.
18774 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
18776         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
18777         final_end_function.
18779 2013-02-28  Marek Polacek  <polacek@redhat.com>
18781         PR rtl-optimization/56466
18782         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
18783         if we're changing a loop.
18784         (peel_loops_completely): Likewise.
18786 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
18788         PR c++/55813
18789         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
18791 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
18793         PR target/56445
18794         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
18795         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
18796         INTX_FTYPE_FX, FX_FTYPE_INTX.
18797         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
18799 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
18801         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
18802         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
18803         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
18804         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
18805         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
18806         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
18807         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
18808         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
18809         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
18810         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
18811         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
18812         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
18813         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
18814         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
18815         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
18816         (avrxmega6): Increase max flash segments from 5 to 6.
18817         * config/avr/t-multilib: Regenerate.
18818         * config/avr/avr-tables.opt: Regenerate.
18819         * doc/avr-mmcu.texi: Regenerate.
18821 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
18823         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
18824         (avr_device_to_arch): Rename to avr_device_to_ld.
18825         (avr_device_to_as): New prototype.
18826         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
18827         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
18828         * config/avr/driver-avr.c (avr_device_to_as): New.
18829         (avr_device_to_arch): Rename to avr_device_to_ld.
18831 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
18833         PR middle-end/56461
18834         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
18835         method on dr_chain and result_chain.
18837         PR middle-end/56461
18838         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
18839         pointer_set_destroy on not_executed_last_iteration.
18841         PR middle-end/56461
18842         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
18844         PR middle-end/56461
18845         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
18846         FOR_EACH_DEFINED_FUNCTION when freeing state.
18848         PR middle-end/56461
18849         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
18850         pool_free.
18851         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
18852         overwriting it.
18854         PR middle-end/56461
18855         * ipa-cp.c (decide_whether_version_node): Call vec_free on
18856         known_aggs[i].items and release known_aggs vector.
18858         PR middle-end/56461
18859         * ipa-reference.c (propagate): Free node_info even for alias nodes.
18861 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18863         * config/microblaze/microblaze.c (microblaze_emit_compare):
18864         Use xor for EQ/NE comparisions.
18865         * config/microblaze/microblaze.md (cstoresf4): Add constraints
18866         (cbranchsf4): Adjust operator to comparison_operator.
18868 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
18870         PR middle-end/56461
18871         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
18872         vector.
18873         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
18874         vec_safe_push, always update *slot.
18875         (redirect_edge_var_map_clear): Use vec_free.
18876         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
18877         (free_var_map_entry): Use vec_free.
18878         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
18879         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
18881 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
18883         PR middle-end/45472
18884         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
18885         when the may_trap_p bit of the exprs being merged differs.
18886         Reorder tests for speculativeness in the logical and operator.
18888 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
18890         * incpath.c (add_standard_paths): Use reconcat instead of concat
18891         where appropriate and avoid leaking memory.
18893         * opts.h: Include obstack.h.
18894         (opts_concat): New prototype.
18895         (opts_obstack): New declaration.
18896         * opts.c (opts_concat): New function.
18897         (opts_obstack): New variable.
18898         (init_options_struct): Call gcc_init_obstack on opts_obstack.
18899         (finish_options): Use opts_concat instead of concat
18900         and XOBNEWVEC instead of XNEWVEC.
18901         * opts-common.c (generate_canonical_option, decode_cmdline_option,
18902         generate_option): Likewise.
18903         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
18904         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
18906         PR target/56455
18907         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
18908         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
18910 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
18912         PR middle-end/56461
18913         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
18915 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
18917         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
18918         (arm_block_move_unaligned_straight): Likewise.
18919         (arm_adjust_block_mem): Likewise.
18921 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
18923         PR target/48901
18924         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
18925         temp, cond and label.
18926         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
18928         PR target/52500
18929         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
18930         * config/c6x/c6x.h (dbx_register_map): Update declaration.
18932         PR target/52501
18933         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
18934         of prologue/epilogue functions.
18936         PR target/52550
18937         * config/tilegx/tilegx.c (tilegx_expand_prologue):
18938         Remove unused variable cfa_offset.
18939         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
18941         PR target/54639
18942         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
18943         type promotion to unsigned.
18945         PR target/54640
18946         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
18947         for HOST_WIDE_INT of 32 bit / same size as int.
18948         (arm_block_move_unaligned_straight): Likewise.
18949         (arm_adjust_block_mem): Likewise.
18951         PR target/54662
18952         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
18953         ALL_CFLAGS.
18955 2013-02-26  Marek Polacek  <polacek@redhat.com>
18957         PR tree-optimization/56426
18958         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
18960 2013-02-26  Richard Biener  <rguenther@suse.de>
18962         PR target/56444
18963         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
18964         unused variable loops.
18966 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
18968         PR tree-optimization/56448
18969         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
18970         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
18971         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
18972         later operands of the references, or even first operand for
18973         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
18975         PR tree-optimization/56443
18976         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
18977         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
18978         to type_for_mode langhook.
18980 2013-02-25  Matt Turner  <mattst88@gmail.com>
18982         * doc/invoke.texi: Document r4700.
18984 2013-02-25  Richard Biener  <rguenther@suse.de>
18986         PR tree-optimization/56175
18987         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
18988         split out from ...
18989         (simplify_bitwise_binary): ... here.  Also guard the conversion
18990         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
18992 2013-02-25  Catherine Moore  <clm@codesourcery.com>
18994         Revert:
18995         2013-02-24  Catherine Moore  <clm@codesourcery.com>
18996             Maciej W. Rozycki  <macro@codesourcery.com>
18997             Tom de Vries  <tom@codesourcery.com>
18998             Nathan Sidwell  <nathan@codesourcery.com>
18999             Iain Sandoe  <iain@codesourcery.com>
19000             Nathan Froyd  <froydnj@codesourcery.com>
19001             Chao-ying Fu  <fu@mips.com>
19003         * doc/extend.texi: (micromips, nomicromips, nocompression):
19004         Document new function attributes.
19005         * doc/invoke.texi (minterlink-compressed, mmicromips,
19006         m14k, m14ke, m14kec): Document new options.
19007         (minterlink-mips16): Update documentation.
19008         * doc/md.texi (ZC, ZD): Document new constraints.
19009         * configure.ac (gcc_cv_as_micromips): Check if linker
19010         supports the .set micromips directive.
19011         * configure: Regenerate.
19012         * config.in: Regenerate.
19013         * config/mips/mips-tables.opt: Regenerate.
19014         * config/mips/micromips.md: New file.
19015         * constraints.md (ZC, AD): New constraints.
19016         * config/mips/predicates.md (movep_src_register): New predicate.
19017         (movep_src_operand): New predicate.
19018         (non_volatile_mem_operand): New predicate.
19019         * config/mips/mips.md (multimem): New type.
19020         (length): Differentiate between 17-bit and 18-bit branch offsets.
19021         (MOVEP1, MOVEP2): New mode iterator.
19022         (mov_<load>l): Use ZC constraint.
19023         (mov_<load>r): Likewise.
19024         (mov_<store>l): Likewise.
19025         (mov_<store>r): Likewise.
19026         (*branch_equality<mode>_inverted): Add microMIPS support.
19027         (*branch_equality<mode>): Likewise.
19028         (*jump_absolute): Likewise.
19029         (indirect_jump_<mode>): Likewise.
19030         (tablejump_<mode>): Likewise.
19031         (<optab>_internal): Likewise.
19032         (sibcall_internal): Likewise.
19033         (sibcall_value_internal): Likewise.
19034         (prefetch): Use constraint ZD.
19035         * config/mips/mips.opt (minterlink-compressed): New option.
19036         (minterlink-mips16): Now an alias for minterlink-compressed.
19037         (mmicromips): New option.
19038         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
19039         (compare_and_swap_12): Likewise.
19040         (sync_add<mode>): Likewise.
19041         (sync_<optab>_12): Likewise.
19042         (sync_old_<optab>_12): Likewise.
19043         (sync_new_<optab>_12): Likewise.
19044         (sync_nand_12): Likewise.
19045         (sync_old_nand_12): Likewise.
19046         (sync_new_nand_12): Likewise.
19047         (sync_sub<mode>): Likewise.
19048         (sync_old_add<mode>): Likewise.
19049         (sync_old_sub<mode>): Likewise.
19050         (sync_new_add<mode>): Likewise.
19051         (sync_new_sub<mode>): Likewise.
19052         (sync_<optab><mode>): Likewise.
19053         (sync_old_<optab><mode>): Likewise.
19054         (sync_new_<optab><mode>): Likewise.
19055         (sync_nand<mode>): Likewise.
19056         (sync_old_nand<mode>): Likewise.
19057         (sync_new_nand<mode>): Likewise.
19058         (sync_lock_test_and_set<mode>): Likewise.
19059         (test_and_set_12): Likewise.
19060         (atomic_compare_and_swap<mode>): Likewise.
19061         (atomic_exchange<mode>_llsc): Likewise.
19062         (atomic_fetch_add<mode>_llsc): Likewise.
19063         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
19064         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
19065         (umips_save_restore_pattern_p): Likewise.
19066         (umips_load_store_pair_p): Likewise.
19067         (umips_output_load_store_pair): Likewise.
19068         (umips_movep_target_p): Likewise.
19069         (umips_12bit_offset_address_p): Likewise.
19070         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
19071         (mips_base_mips16): Rename this...
19072         (mips_base_compression_flags): ...to this. Update all uses.
19073         (mips_attribute_table): Add micromips, nomicromips and nocompression.
19074         (mips_mips16_decl_p): Delete.
19075         (mips_nomips16_decl_p): Delete.
19076         (mips_get_compress_on_flags): New function.
19077         (mips_get_compress_off_flags): New function.
19078         (mips_get_compress_mode): New function.
19079         (mips_get_compress_on_name): New function.
19080         (mips_get_compress_off_name): New function.
19081         (mips_insert_attributes): Support multiple compression types.
19082         (mips_merge_decl_attributes): Likewise.
19083         (umips_12bit_offset_address_p): New function.
19084         (mips_start_function_definition): Emit .set micromips directive.
19085         (mips_call_may_need_jalx_p): New function.
19086         (mips_function_ok_for_sibcall): Add microMIPS support.
19087         (mips_print_operand_punctuation): Support short delay slots and
19088         compact jumps.
19089         (umips_swm_mask, umips_swm_encoding): New.
19090         (umips_build_save_restore): New function.
19091         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
19092         (was_mips16_p): Remove.
19093         (old_compression_mode): New.
19094         (mips_set_compression_mode): New function.
19095         (mips_set_current_function): Add microMIPS support.
19096         (mips_option_override): Likewise.
19097         (umips_save_restore_pattern_p): New function.
19098         (umips_output_save_restore): New function.
19099         (umips_load_store_pair_p_1): New function.
19100         (umips_load_store_pair_p): New function.
19101         (umips_output_load_store_pair_1): New function.
19102         (umips_output_load_store_pair): New function.
19103         (umips_movep_target_p) New function.
19104         (mips_prepare_pch_save): Add microMIPS support.
19105         * config/mips/mips.h (TARGET_COMPRESSION): New.
19106         (TARGET_CPU_CPP_BUILTINS): Update macro
19107         to use new compression flags and to support microMIPS.
19108         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
19109         (MIPS_ARCH_FLOAT_SPEC): Likewise.
19110         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
19111         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
19112         (ASM_SPEC): Support mmicromips and mno-micromips.
19113         (M16STORE_REG_P): New macro.
19114         (MIPS_CALL): Support TARGET_MICROMIPS.
19115         (MICROMIPS_J): New macro.
19116         (mips_base_mips16): Rename this...
19117         (mips_base_compression_flags): ...to this.
19118         (UMIPS_12BIT_OFFSET_P): New macro.
19119         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
19120         (MULTILIB_DIRNAMES): Likewise.
19122 2013-02-25  Tom de Vries  <tom@codesourcery.com>
19124         PR rtl-optimization/56131
19125         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
19126         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
19127         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
19129 2013-02-25  Tobias Burnus  <burnus@net-b.de>
19131         * doc/invoke.texi (-fsanitize=): Move from optimization
19132         to debugging options.
19134 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
19136         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
19138 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
19139             Alexander Monakov  <amonakov@ispras.ru>
19141         PR middle-end/56077
19142         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
19143         flush pending lists also on non-jumps.  Adjust comment.
19145 2013-02-24  Catherine Moore  <clm@codesourcery.com>
19146             Maciej W. Rozycki  <macro@codesourcery.com>
19147             Tom de Vries  <tom@codesourcery.com>
19148             Nathan Sidwell  <nathan@codesourcery.com>
19149             Iain Sandoe  <iain@codesourcery.com>
19150             Nathan Froyd  <froydnj@codesourcery.com>
19151             Chao-ying Fu  <fu@mips.com>
19153         * doc/extend.texi: (micromips, nomicromips, nocompression):
19154         Document new function attributes.
19155         * doc/invoke.texi (minterlink-compressed, mmicromips,
19156         m14k, m14ke, m14kec): Document new options.
19157         (minterlink-mips16): Update documentation.
19158         * doc/md.texi (ZC, ZD): Document new constraints.
19159         * configure.ac (gcc_cv_as_micromips): Check if linker
19160         supports the .set micromips directive.
19161         * configure: Regenerate.
19162         * config.in: Regenerate.
19163         * config/mips/mips-tables.opt: Regenerate.
19164         * config/mips/micromips.md: New file.
19165         * constraints.md (ZC, AD): New constraints.
19166         * config/mips/predicates.md (movep_src_register): New predicate.
19167         (movep_src_operand): New predicate.
19168         (non_volatile_mem_operand): New predicate.
19169         * config/mips/mips.md (multimem): New type.
19170         (length): Differentiate between 17-bit and 18-bit branch offsets.
19171         (MOVEP1, MOVEP2): New mode iterator.
19172         (mov_<load>l): Use ZC constraint.
19173         (mov_<load>r): Likewise.
19174         (mov_<store>l): Likewise.
19175         (mov_<store>r): Likewise.
19176         (*branch_equality<mode>_inverted): Add microMIPS support.
19177         (*branch_equality<mode>): Likewise.
19178         (*jump_absolute): Likewise.
19179         (indirect_jump_<mode>): Likewise.
19180         (tablejump_<mode>): Likewise.
19181         (<optab>_internal): Likewise.
19182         (sibcall_internal): Likewise.
19183         (sibcall_value_internal): Likewise.
19184         (prefetch): Use constraint ZD.
19185         * config/mips/mips.opt (minterlink-compressed): New option.
19186         (minterlink-mips16): Now an alias for minterlink-compressed.
19187         (mmicromips): New option.
19188         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
19189         (compare_and_swap_12): Likewise.
19190         (sync_add<mode>): Likewise.
19191         (sync_<optab>_12): Likewise.
19192         (sync_old_<optab>_12): Likewise.
19193         (sync_new_<optab>_12): Likewise.
19194         (sync_nand_12): Likewise.
19195         (sync_old_nand_12): Likewise.
19196         (sync_new_nand_12): Likewise.
19197         (sync_sub<mode>): Likewise.
19198         (sync_old_add<mode>): Likewise.
19199         (sync_old_sub<mode>): Likewise.
19200         (sync_new_add<mode>): Likewise.
19201         (sync_new_sub<mode>): Likewise.
19202         (sync_<optab><mode>): Likewise.
19203         (sync_old_<optab><mode>): Likewise.
19204         (sync_new_<optab><mode>): Likewise.
19205         (sync_nand<mode>): Likewise.
19206         (sync_old_nand<mode>): Likewise.
19207         (sync_new_nand<mode>): Likewise.
19208         (sync_lock_test_and_set<mode>): Likewise.
19209         (test_and_set_12): Likewise.
19210         (atomic_compare_and_swap<mode>): Likewise.
19211         (atomic_exchange<mode>_llsc): Likewise.
19212         (atomic_fetch_add<mode>_llsc): Likewise.
19213         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
19214         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
19215         (umips_save_restore_pattern_p): Likewise.
19216         (umips_load_store_pair_p): Likewise.
19217         (umips_output_load_store_pair): Likewise.
19218         (umips_movep_target_p): Likewise.
19219         (umips_12bit_offset_address_p): Likewise.
19220         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
19221         (mips_base_mips16): Rename this...
19222         (mips_base_compression_flags): ...to this. Update all uses.
19223         (mips_attribute_table): Add micromips, nomicromips and nocompression.
19224         (mips_mips16_decl_p): Delete.
19225         (mips_nomips16_decl_p): Delete.
19226         (mips_get_compress_on_flags): New function.
19227         (mips_get_compress_off_flags): New function.
19228         (mips_get_compress_mode): New function.
19229         (mips_get_compress_on_name): New function.
19230         (mips_get_compress_off_name): New function.
19231         (mips_insert_attributes): Support multiple compression types.
19232         (mips_merge_decl_attributes): Likewise.
19233         (umips_12bit_offset_address_p): New function.
19234         (mips_start_function_definition): Emit .set micromips directive.
19235         (mips_call_may_need_jalx_p): New function.
19236         (mips_function_ok_for_sibcall): Add microMIPS support.
19237         (mips_print_operand_punctuation): Support short delay slots and
19238         compact jumps.
19239         (umips_swm_mask, umips_swm_encoding): New.
19240         (umips_build_save_restore): New function.
19241         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
19242         (was_mips16_p): Remove.
19243         (old_compression_mode): New.
19244         (mips_set_compression_mode): New function.
19245         (mips_set_current_function): Add microMIPS support.
19246         (mips_option_override): Likewise.
19247         (umips_save_restore_pattern_p): New function.
19248         (umips_output_save_restore): New function.
19249         (umips_load_store_pair_p_1): New function.
19250         (umips_load_store_pair_p): New function.
19251         (umips_output_load_store_pair_1): New function.
19252         (umips_output_load_store_pair): New function.
19253         (umips_movep_target_p) New function.
19254         (mips_prepare_pch_save): Add microMIPS support.
19255         * config/mips/mips.h (TARGET_COMPRESSION): New.
19256         (TARGET_CPU_CPP_BUILTINS): Update macro
19257         to use new compression flags and to support microMIPS.
19258         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
19259         (MIPS_ARCH_FLOAT_SPEC): Likewise.
19260         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
19261         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
19262         (ASM_SPEC): Support mmicromips and mno-micromips.
19263         (M16STORE_REG_P): New macro.
19264         (MIPS_CALL): Support TARGET_MICROMIPS.
19265         (MICROMIPS_J): New macro.
19266         (mips_base_mips16): Rename this...
19267         (mips_base_compression_flags): ...to this.
19268         (UMIPS_12BIT_OFFSET_P): New macro.
19269         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
19270         (MULTILIB_DIRNAMES): Likewise.
19272 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
19274         PR target/52555
19275         * target-globals.c (save_target_globals): For init_reg_sets and
19276         target_reinit remporarily set this_fn_optabs to this_target_optabs.
19278 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
19280         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
19281         * config/aarch64/t-aarch64
19282         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
19284 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
19286         PR inline-asm/56148
19287         * lra-constraints.c (process_alt_operands): Reload operand
19288         conflicting with earlier clobber only if no more other conflicting
19289         operands.
19291 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
19293         PR sanitizer/56393
19294         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
19295         if not linking a shared library.
19297 2013-02-22  Seth LaForge  <sethml@google.com>
19299         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
19301 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
19303         * config/arm/arm.md (split for extendsidi): Update condition.
19304         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
19305         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
19306         (qhs_zextenddi_cstr): Likewise.
19308 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
19310         PR middle-end/56420
19311         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
19312         avoid signed wrapping.
19313         (expand_mult): Handle properly multiplication by
19314         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
19315         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
19316         in the compiler if coeff is HOST_WIDE_INT_MIN.
19317         (expand_divmod): Don't make ext_op1 static, change it's type to
19318         uhwi.  Avoid undefined behavior in -INTVAL (op1).
19320         PR rtl-optimization/50339
19321         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
19322         field.
19323         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
19324         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
19325         into splitting_ashiftrt field.
19326         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
19327         ASHIFTRT.
19328         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
19329         choices.
19331 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
19333         PR middle-end/56108
19334         * trans-mem.c (execute_tm_mark): Do not expand transactions that
19335         are sure to go irrevocable.
19337 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
19339         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
19340         scalars are valid operands.
19342 2013-02-21  Martin Jambor  <mjambor@suse.cz>
19344         PR tree-optimization/56310
19345         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
19346         only matching indices and non-negative final offsets.
19347         (intersect_aggregates_with_edge): Pass src_idx to
19348         agg_replacements_to_vector.  Pass src_idx insstead of index to
19349         intersect_with_agg_replacements.
19351 2013-02-21  Martin Jambor  <mjambor@suse.cz>
19353         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
19354         instead of hard-wired defaults.
19356 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
19358         * doc/invoke.texi (MIPS Options): Update documentation of the
19359         floating-point multiply-accumulate instruction restrictions.
19361 2013-02-21  Kostya Serebryany  <kcc@google.com>
19363         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
19364         asan_shadow_offset on x86_64 linux.
19366 2013-02-21  Richard Biener  <rguenther@suse.de>
19368         PR tree-optimization/56415
19369         Revert
19370         2013-02-11  Richard Biener  <rguenther@suse.de>
19372         PR tree-optimization/56273
19373         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
19374         first VRP run.
19376 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
19378         PR bootstrap/56258
19379         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
19380         instead of @itemx.
19382         PR inline-asm/56405
19383         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
19384         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
19386 2013-02-20  Jan Hubicka  <jh@suse.cz>
19388         PR tree-optimization/56265
19389         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
19390         when target is referenced for first time.
19392 2013-02-20  Richard Biener  <rguenther@suse.de>
19394         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
19395         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
19396         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
19397         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
19398         not return anything.
19399         (rename_ssa_copies): Do not remove unused locals.
19400         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
19401         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
19402         * passes.c (execute_function_todo): Do not schedule unused locals
19403         removal if cleanup_tree_cfg did something.
19404         * tree-ssa-live.c (remove_unused_locals): Dump statistics
19405         about the number of removed locals.
19407 2013-02-20  Richard Biener  <rguenther@suse.de>
19409         PR tree-optimization/56398
19410         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
19412 2013-02-20  Martin Jambor  <mjambor@suse.cz>
19414         PR tree-optimization/55334
19415         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
19416         restricted pointers to arrays.
19418 2013-02-20  Richard Biener  <rguenther@suse.de>
19419             Jakub Jelinek  <jakub@redhat.com>
19421         PR tree-optimization/56396
19422         * tree-ssa-ccp.c (n_const_val): New static variable.
19423         (get_value): Return NULL for SSA names we don't have a lattice
19424         entry for.
19425         (ccp_initialize): Initialize n_const_val.
19426         * tree-ssa-copy.c (n_copy_of): New static variable.
19427         (init_copy_prop): Initialize n_copy_of.
19428         (get_value): Return NULL_TREE for SSA names we don't have a
19429         lattice entry for.
19431 2013-02-20  Martin Jambor  <mjambor@suse.cz>
19433         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
19435 2013-02-20  Richard Biener  <rguenther@suse.de>
19437         * genpreds.c (write_lookup_constraint): Do not compare first
19438         letter of the constraint again.
19440 2013-02-20  Richard Biener  <rguenther@suse.de>
19442         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
19443         and ceil_log2.
19444         (get_use_iv_cost): Terminate hashtable walk when coming across
19445         an empty entry.
19447 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
19449         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
19450         reassociation for avx2 targets.
19452 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19454         * config/microblaze/microblaze.c: microblaze_has_clz = 0
19455         Add version check for v8.10.a to enable microblaze_has_clz
19456         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
19457         version and TARGET_PATTERN_COMPARE check
19458         * config/microblaze/microblaze.md: New clzsi2 instruction
19460 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19462         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
19463         function before branching.
19465 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
19467         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
19468         DUMP_INSN_RTX_UID.
19469         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
19471 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
19473         PR middle-end/55889
19474         * sel-sched.c: Include ira.h.
19475         (implicit_clobber_conflict_p): New function.
19476         (moveup_expr): Use it.
19477         * Makefile.in (sel-sched.o): Depend on ira.h.
19479 2013-02-19  Richard Biener  <rguenther@suse.de>
19481         PR tree-optimization/56384
19482         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
19483         (vn_hash_type): Split out from ...
19484         (vn_hash_constant_with_type): ... here.
19485         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
19486         (vn_phi_eq): Compare types from vn_phi_s structure.
19487         (vn_phi_lookup): Populate vn_phi_s type.
19488         (vn_phi_insert): Likewise.
19490 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
19492         PR tree-optimization/56350
19493         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
19494         if haven't found reduction or nested cycle operand, rather than
19495         asserting we must find it.
19497         PR tree-optimization/56381
19498         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
19499         to fold_build3.
19501 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
19502             Jakub Jelinek  <jakub@redhat.com>
19504         PR target/52555
19505         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
19506         (swap_optab_enable): Same.
19507         (init_all_optabs): Use argument instead of global.
19508         * tree.h (struct tree_optimization_option): New field target_optabs.
19509         * expr.h (init_all_optabs): Add argument to prototype.
19510         (TREE_OPTIMIZATION_OPTABS): New.
19511         (save_optabs_if_changed): Protoize.
19512         * optabs.h: Declare this_fn_optabs.
19513         * optabs.c (save_optabs_if_changed): New.
19514         Declare this_fn_optabs.
19515         (init_optabs): Add argument to init_all_optabs() call.
19516         * function.c (invoke_set_current_function_hook): Handle per
19517         function optabs.
19518         * function.h (struct function): New field optabs.
19519         * config/mips/mips.c (mips_set_mips16_mode): Handle when
19520         optimization_current_node has changed.
19521         * target-globals.h (save_target_globals_default_opts): Protoize.
19522         * target-globals.c (save_target_globals_default_opts): New.
19524 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19526         PR target/56347
19527         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
19528         registers %fr12 and %fr12R as call used.
19530         PR target/56214
19531         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
19532         and HImode, require all displacements to be an integer multiple of
19533         their mode size.
19534         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
19535         only allow QImode and HImode when reload is in progress and strict is
19536         true.  Likewise for symbolic addresses.  Use base14_operand to check
19537         displacements in REG+BASE addresses.
19539 2013-02-18  Richard Biener  <rguenther@suse.de>
19541         PR tree-optimization/56366
19542         * tree-vect-loop.c (get_initial_def_for_induction): Properly
19543         handle sign-conversion of outer-loop initial induction value.
19545 2013-02-18  Richard Biener  <rguenther@suse.de>
19547         PR middle-end/56349
19548         * cfghooks.c (merge_blocks): If we merge a latch into another
19549         block adjust references to it.
19550         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
19551         (verify_loop_structure): Verify that a recorded latch is in fact
19552         a latch.
19554 2013-02-18  Richard Biener  <rguenther@suse.de>
19556         PR tree-optimization/56321
19557         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
19558         order SSA name release and virtual operand unlinking.
19560 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19562         * config/microblaze/microblaze.md (save_stack_block): Define.
19563         (restore_stack_block): Likewise.
19565 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19567         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
19568         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
19569         * config/microblaze/microblaze.c (microblaze_option_override):
19570         Bail out early for PIC modes when target does not support PIC.
19572 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19574         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
19575         Replace with a microblaze version.
19576         (microblaze_trampoline_init): Adapt for microblaze.
19577         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
19578         microblaze.
19580 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
19581             Dodji Seketeli  <dodji@redhat.com>
19583         PR asan/56330
19584         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
19585         (instrument_mem_region_access): Do not forget to always put
19586         instrumentation of the of 'base' and 'base + len' in a "if (len !=
19587         0) statement, even for cases where either 'base' or 'base + len'
19588         are not instrumented -- because they have been previously
19589         instrumented.  Simplify the logic by putting all the statements
19590         instrument 'base + len' inside a sequence, and then insert that
19591         sequence right before the current insertion point.  Then, to
19592         instrument 'base + len', just get an iterator on that statement.
19593         And do not forget to update the pointer to iterator the function
19594         received as argument.
19596 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
19598         PR rtl-optimization/56348
19599         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
19601 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
19603         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
19604         (clean_graph_dump_file): Pass base to start_graph_dump.
19606 2013-02-14  Richard Henderson  <rth@redhat.com>
19608         PR target/55941
19609         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
19611 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
19613         * collect2-aix.h: Define F_LOADONLY.
19615 2013-02-14  Richard Biener  <rguenther@suse.de>
19617         PR lto/50494
19618         * varasm.c (output_constant_def_1): Get the decl representing
19619         the constant as argument.
19620         (output_constant_def): Wrap output_constant_def_1.
19621         (make_decl_rtl): Use output_constant_def_1 with the decl
19622         representing the constant.
19623         (build_constant_desc): Optionally re-use a decl already
19624         representing the constant.
19625         (tree_output_constant_def): Adjust.
19627 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
19629         Fix an asan crash
19630         * asan.c (instrument_builtin_call):  Really put the length of the
19631         second source argument into src1_len.
19633 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
19635         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
19636         argument.  If it is false, don't create edge from then_bb to
19637         fallthru_bb.
19638         (insert_if_then_before_iter): Pass true to it.
19639         (build_check_stmt): Pass false to it.
19640         (transform_statements): Flush hash table only on extended basic
19641         block boundaries, rather than at the beginning of every bb.
19642         Don't flush hash table on nonfreeing_call_p calls.
19643         * tree-flow.h (nonfreeing_call_p): New prototype.
19644         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
19646 2013-02-13  David S. Miller  <davem@davemloft.net>
19648         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
19650 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
19652         PR target/56184
19653         * ira.c (max_regno_before_ira): Move from ...
19654         (ira): ... here.
19655         (fix_reg_equiv_init): Use max_regno_before_ira instead of
19656         vec_safe_length.
19658 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
19660         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
19662 2013-02-13  Richard Biener  <rguenther@suse.de>
19664         PR lto/56295
19665         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
19666         globals in MEM_REFs.
19668 2013-02-13  Richard Biener  <rguenther@suse.de>
19670         * loop-init.c (loop_optimizer_init): Clear loop state when
19671         re-initializing preserved loops.
19672         * loop-unswitch.c (unswitch_single_loop): Return whether
19673         we unswitched the loop.  Do not verify loop state here.
19674         (unswitch_loops): When we unswitched a loop discover new loops.
19676 2013-02-13  Kostya Serebryany  <kcc@google.com>
19678         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
19679         on x86_64 linux.
19680         * sanitizer.def: Rename __asan_init to __asan_init_v1.
19682 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
19684         Avoid instrumenting duplicated memory access in the same basic block
19685         * Makefile.in (asan.o): Add new dependency on hash-table.h
19686         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
19687         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
19688         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
19689         (free_mem_ref_resources, has_mem_ref_been_instrumented)
19690         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
19691         (get_mem_ref_of_assignment): New functions.
19692         (get_mem_refs_of_builtin_call): Extract from
19693         instrument_builtin_call and tweak a little bit to make it fit with
19694         the new signature.
19695         (instrument_builtin_call): Use the new
19696         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
19697         of is_gimple_builtin_call.
19698         (instrument_derefs, instrument_mem_region_access): Insert the
19699         instrumented memory reference into the hash table.
19700         (maybe_instrument_assignment): Renamed instrument_assignment into
19701         this, and change it to advance the iterator when instrumentation
19702         actually happened and return true in that case.  This makes it
19703         homogeneous with maybe_instrument_assignment, and thus give a
19704         chance to callers to be more 'regular'.
19705         (transform_statements): Clear the memory reference hash table
19706         whenever we enter a new BB, when we cross a function call, or when
19707         we are done transforming statements.  Use
19708         maybe_instrument_assignment instead of instrumentation.  No more
19709         need to special case maybe_instrument_assignment and advance the
19710         iterator after calling it; it's now handled just like
19711         maybe_instrument_call.  Update comment.
19713 2013-02-13  Richard Biener  <rguenther@suse.de>
19715         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
19716         Fix loop discovery code.
19718 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
19720         PR inline-asm/56148
19721         * lra-constraints.c (process_alt_operands): Match early clobber
19722         operand with itself.  Check conflicts with earlyclobber only if
19723         the operand is not reloaded.  Prefer to reload conflicting operand
19724         if earlyclobber and matching operands are the same.
19726 2013-02-12  Richard Biener  <rguenther@suse.de>
19728         PR lto/56297
19729         * lto-streamer-out.c (write_symbol): Do not output symbols
19730         for hard register variables.
19732 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
19734         PR target/54222
19735         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
19736         (umulsidi3_insn, mulsidi3_insn): New insns.
19738 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
19740         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
19741         (struct tune_params): Add vec_costs field.
19742         * config/arm/arm.c (arm_builtin_vectorization_cost)
19743         (arm_add_stmt_cost): New functions.
19744         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
19745         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
19746         (arm_default_vec_cost): New struct of type cpu_vec_costs.
19747         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
19748         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
19749         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
19750         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
19752 2013-02-12  Richard Biener  <rguenther@suse.de>
19754         PR lto/56295
19755         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
19756         decls again if possible.
19758 2013-02-12  Richard Biener  <rguenther@suse.de>
19760         PR middle-end/56288
19761         * tree-ssa.c (verify_ssa_name): Fix check, move
19762         SSA_NAME_IN_FREE_LIST check up.
19764 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
19765             Steven Bosscher   <steven@gcc.gnu.org>
19767         PR rtl-optimization/56151
19768         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
19769         equal to op0 or op1, and last_insn pattern is CODE operation
19770         with MEM dest and one of the operands matches that MEM.
19772 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
19774         * doc/extend.texi: Document Function Multiversioning and "default"
19775         parameter string to target attribute.
19776         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
19777         target attribute parameter is "default".
19778         (ix86_compare_version_priority): Remove checks for target attribute.
19779         (ix86_mangle_function_version_assembler_name): Change error to sorry.
19780         Remove check for target attribute equal to NULL. Add assert.
19781         (ix86_generate_version_dispatcher_body): Change error to sorry.
19783 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
19784             Jack Howarth  <howarth@bromo.med.uc.edu>
19785             Patrick Marlier  <patrick.marlier@gmail.com>
19787         PR libitm/55693
19788         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
19789         define ENDFILE_SPEC as TM_DESTRUCTOR.
19790         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
19792 2013-02-11  Alexander Potapenko  <glider@google.com>
19793             Jack Howarth  <howarth@bromo.med.uc.edu>
19794             Jakub Jelinek  <jakub@redhat.com>
19796         PR sanitizer/55617
19797         * config/darwin.c (cdtor_record): Rename ctor_record.
19798         (sort_cdtor_records): Rename sort_ctor_records.
19799         (finalize_dtors): New routine to sort destructors by
19800         priority before use in assemble_integer.
19801         (machopic_asm_out_destructor): Use finalize_dtors if needed.
19803 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
19805         PR rtl-optimization/56275
19806         * simplify-rtx.c (avoid_constant_pool_reference): Check that
19807         offset is non-negative and less than cmode size before
19808         calling simplify_subreg.
19810 2013-02-11  Richard Biener  <rguenther@suse.de>
19812         PR tree-optimization/56264
19813         * cfgloop.h (fix_loop_structure): Adjust prototype.
19814         * loop-init.c (fix_loop_structure): Return the number of
19815         newly discovered loops.
19816         * tree-cfgcleanup.c (repair_loop_structures): When new loops
19817         are discovered, do a full loop-closed SSA rewrite.
19819 2013-02-11  Richard Biener  <rguenther@suse.de>
19821         PR tree-optimization/56273
19822         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
19823         first VRP run.
19824         (check_array_ref): Fix missing newline in dumps.
19825         (search_for_addr_array): Likewise.
19827 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
19829         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
19831 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
19833         PR target/56256
19834         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
19836 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
19838         PR rtl-optimization/56246
19839         * lra-constraints.c (simplify_operand_subreg): Try to reuse
19840         reload pseudo.
19841         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
19842         constraints are satisfied.
19844 2013-02-08  Jeff Law  <law@redhat.com>
19846         PR debug/53948
19847         * emit-rtl.c (reg_is_parm_p): New function.
19848         * regs.h (reg_is_parm_p): New prototype.
19849         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
19850         callee-clobbered registers.
19852 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
19854         PR target/56043
19855         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
19856         If there is no implicit builtin declaration, just return NULL.
19858 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
19860         * config/i386/sse.md (FMAMODEM): New mode iterator.
19861         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
19862         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
19864 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
19866         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
19867         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
19868         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
19870 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19872         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
19873         (microblaze*-*-elf): Likewise.
19874         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
19875         LINK_SPEC.
19876         * config/microblaze/microblaze-c.c: Add builtin defines for
19877         _LITTLE_ENDIAN and _BIG_ENDIAN.
19878         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
19879         add to TARGET_DEFAULT flags.
19880         Expand ASM_SPEC and LINK_SPEC.
19881         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
19882         * config/microblaze/microblaze.md: Update extendsidi2 and
19883         movdi_internal instructions to use low-order / high-order reg
19884         print_operands.
19885         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
19886         options and inversemask / mask of LITTLE_ENDIAN.
19887         * config/microblaze/t-microblaze: Expand multilib options to
19888         include mlittle-endian (le) and update exceptions patterns.
19890 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
19892         PR rtl-optimization/56195
19893         * lra-constraints.c (get_reload_reg): Don't reuse regs
19894         if they have smaller mode than requested, if they have
19895         wider mode than requested, try to return a SUBREG.
19897         PR tree-optimization/56250
19898         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
19899         if type is unsigned and code isn't MULT_EXPR.
19901 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
19903         PR tree-optimization/56064
19904         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
19905         bits according to mode.
19906         * fixed-value.h (fixed_from_double_int)
19907         (const_fixed_from_double_int): Adjust comments.
19909 2013-02-08  Richard Biener  <rguenther@suse.de>
19911         PR lto/56231
19912         * lto-streamer.h (struct data_in): Remove current_file, current_line
19913         and current_col members.
19914         * lto-streamer-out.c (lto_output_location): Stream changed bits
19915         en-block for efficiency.
19916         * lto-streamer-in.c (clear_line_info): Remove.
19917         (lto_input_location): Cache current file, line and column
19918         globally via local statics.  Read changed bits en-block.
19919         (input_function): Do not call clear_line_info.
19920         (lto_read_body): Likewise.
19921         (lto_input_toplevel_asms): Likewise.
19923 2013-02-08  Michael Matz  <matz@suse.de>
19925         PR tree-optimization/52448
19926         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
19927         (nt_call_phase): New static.
19928         (add_or_mark_expr): Only mark accesses with newer phase than any
19929         call seen.
19930         (nonfreeing_call_p): New.
19931         (nt_init_block): Update nt_call_phase, mark blocks as visited.
19932         (nt_fini_block): Keep blocks marked as visited.
19933         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
19935 2013-02-08  Richard Biener  <rguenther@suse.de>
19937         * ira.c (ira): Free broken dominator information.
19939 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
19941         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
19943 2013-02-08  Marek Polacek  <polacek@redhat.com>
19945         * cfgloop.c (verify_loop_structure): Add more checking of headers.
19947 2013-02-08  Richard Biener  <rguenther@suse.de>
19949         PR middle-end/56181
19950         * cfgloop.h (flow_loops_find): Adjust.
19951         (bb_loop_header_p): Declare.
19952         * cfgloop.c (bb_loop_header_p): New function split out from ...
19953         (flow_loops_find): ... here.  Adjust function signature,
19954         support incremental loop structure update.
19955         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
19956         * cfgloopmanip.c (fix_loop_structure): Move ...
19957         * loop-init.c (fix_loop_structure): ... here.
19958         (apply_loop_flags): Split out from ...
19959         (loop_optimizer_init): ... here.
19960         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
19961         in incremental mode, only remove dead loops here.
19963 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
19965         PR target/54222
19966         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
19967         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
19968         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
19969         (*round<mode>3.libgcc): New insns for fixed-modes.
19970         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
19971         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
19972         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
19973         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
19974         implementations.  Define to __builtin_avr_absFX,
19975         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
19976         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
19977         __builtin_avr_countlsFX, respectively.
19978         * config/avr/avr-c.c (target.h): Include it.
19979         (enum avr_builtin_id): New enum.
19980         (avr_resolve_overloaded_builtin): New static function.
19981         (avr_register_target_pragmas): Use it to set
19982         targetm.resolve_overloaded_builtin.
19983         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
19984         tree nodes used by DEF_BUILTIN.
19985         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
19986         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
19987         <AVR_BUILTIN_xxBITS>: Same.
19989 2013-02-08  Richard Biener  <rguenther@suse.de>
19991         * cfgloop.c (verify_loop_structure): Properly handle
19992         a loop exiting to another loop header.
19993         * ira-int.h (ira_loops): Remove.
19994         * ira.c (ira_loops): Remove.
19995         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
19996         (do_reload): Use loop_optimizer_finalize.
19997         * ira-build.c (create_loop_tree_nodes): Use get_loops and
19998         number_of_loops to access the loop tree.
19999         (more_one_region_p): Likewise.
20000         (finish_loop_tree_nodes): Likewise.
20001         (rebuild_regno_allocno_maps): Likewise.
20002         (mark_loops_for_removal): Likewise.
20003         (mark_all_loops_for_removal): Likewise.
20004         (remove_unnecessary_regions): Likewise.
20005         (ira_build): Likewise.
20006         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
20008 2013-02-08  Richard Biener  <rguenther@suse.de>
20010         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
20011         * ipa-pure-const.c (analyze_function): Avoid calling
20012         mark_irreducible_loops twice.
20013         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
20015 2013-02-07  David S. Miller  <davem@davemloft.net>
20017         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
20018         on 'reg'.
20019         * var-tracking.c (vt_add_function_parameter): Test the presence of
20020         HAVE_window_save properly and do not remap argument registers when
20021         we have a leaf function.
20023 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
20025         PR bootstrap/56227
20026         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
20027         instead of "ll".
20028         * config/i386/i386.c (ix86_print_operand): Ditto.
20030 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
20032         * lra-constraints.c (process_alt_operands): Fix recently added comment.
20034 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
20036         PR rtl-optimization/56225
20037         * lra-constraints.c (process_alt_operands): Check that reload hard
20038         reg can hold value for strict_low_part.
20040 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
20042         PR debug/56154
20043         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
20044         dwarf2out_end_function.
20045         (in_first_function_p, maybe_at_text_label_p,
20046         first_loclabel_num_not_at_text_label): New variables.
20047         (dwarf2out_var_location): In the first function find out
20048         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
20049         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
20050         functions.
20052 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
20054         PR rtl-optimization/56178
20055         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
20056         SUBREG of a register.  Tidy up related block of code.
20057         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
20058         note if the source is a register or a SUBREG of a register.
20060 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
20062         PR target/56228
20063         * config/rs6000/rs6000.md (ptrm): New mode attr.
20064         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
20065         call_value_indirect_aix<pttrsize>,
20066         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
20067         m in constraints.
20069 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
20071         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
20072         if -bnortl. Convert to strcmp and strncmp.
20074 2013-02-07  Alan Modra  <amodra@gmail.com>
20076         PR target/54009
20077         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
20078         addresses won't wrap when offsetting.
20079         (rs6000_secondary_reload): Provide secondary reloads needed for
20080         wrapping LO_SUM addresses.
20082 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
20084         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
20085         MACH, just __MACH__.
20087 2013-02-06  Richard Biener  <rguenther@suse.de>
20089         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
20090         instead of calling fix_loop_structure.
20092 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
20094         PR middle-end/56217
20095         * omp-low.c (use_pointer_for_field): Return false if
20096         lower_send_shared_vars doesn't generate any copy-out code.
20098 2013-02-06  Tom de Vries  <tom@codesourcery.com>
20100         PR rtl-optimization/56131
20101         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
20102         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
20103         of the label is NULL.  Add comment.
20105 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
20107         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
20109         PR sanitizer/55374
20110         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
20111         (STATIC_LIBTSAN_LIBS): Likewise.
20112         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
20113         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
20114         is defined, don't add anything else beyond that.
20115         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
20116         (LINK_COMMAND_SPEC): Use them.
20118         PR tree-optimization/56205
20119         * tree-stdarg.c (check_all_va_list_escapes): Return true if
20120         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
20121         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
20123 2013-02-05  Richard Biener  <rguenther@suse.de>
20125         PR tree-optimization/53342
20126         PR tree-optimization/53185
20127         * tree-vectorizer.h (vect_check_strided_load): Remove.
20128         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
20129         not disallow peeling for vectorized strided loads.
20130         (vect_check_strided_load): Make static and simplify.
20131         (vect_analyze_data_refs): Adjust.
20132         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
20133         correctly when vectorizing strided loads.
20135 2013-02-05  Richard Biener  <rguenther@suse.de>
20137         * doc/install.texi: Refer to ISL, not PPL.
20139 2013-02-05  Jan Hubicka  <jh@suse.cz>
20141         PR tree-optimization/55789
20142         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
20144 2013-02-05  Jan Hubicka  <jh@suse.cz>
20146         PR tree-optimization/55789
20147         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
20148         the dead call anyway.
20150 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
20152         PR sanitizer/55374
20153         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
20155 2013-02-04  Alexander Potapenko  <glider@google.com>
20156             Jack Howarth  <howarth@bromo.med.uc.edu>
20157             Jakub Jelinek  <jakub@redhat.com>
20159         PR sanitizer/55617
20160         * config/darwin.c (sort_ctor_records): Stabilized qsort
20161         on constructor priority by using original position.
20162         (finalize_ctors): New routine to sort constructors by
20163         priority before use in assemble_integer.
20164         (machopic_asm_out_constructor): Use finalize_ctors if needed.
20166 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
20168         PR libstdc++/54314
20169         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
20170         about visibility on artificial decls.
20171         * config/sol2.c (solaris_assemble_visibility): Likewise.
20173 2013-02-04  Kai Tietz  <ktietz@redhat.com>
20175         PR target/56186
20176         * config/i386/i386.c (function_value_ms_64): Add additional valtype
20177         argument and improve checking of return-argument types for 16-byte
20178         modes.
20179         (ix86_function_value_1): Add additional valtype argument on call
20180         of function_value_64.
20181         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
20182         handling infunction_value_64 function.
20184 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
20186         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
20188 2013-02-04  Richard Biener  <rguenther@suse.de>
20190         PR tree-optimization/56188
20191         * tree-ssa-structalias.c (label_visit): Consider case with
20192         initially non-empty points-to set.
20193         (perform_var_substitution): Dump node mapping and clean up.
20195 2013-02-04  Richard Guenther  <rguenther@suse.de>
20197         PR lto/56168
20198         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
20199         node prevail as last resort.
20200         (lto_symtab_merge_decls): Remove guard on LTRANS here.
20201         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
20203 2013-02-04  Richard Biener  <rguenther@suse.de>
20205         PR tree-optimization/56113
20206         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
20207         Merge into ...
20208         (equiv_class_lookup_or_add): ... this.
20209         (label_visit): Adjust and fix error in previous patch.
20210         (perform_var_substitution): Adjust.
20212 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
20214         * config/sh/divtab.c: Fix formatting and comments throughout the file.
20215         * config/sh/sh4-300.md: Likewise.
20216         * config/sh/sh4a.md: Likewise.
20217         * config/sh/constraints.md: Likewise.
20218         * config/sh/sh.md: Likewise.
20219         * config/sh/netbsd-elf.h: Likewise.
20220         * config/sh/predicates.md: Likewise.
20221         * config/sh/sh-protos.h: Likewise.
20222         * config/sh/ushmedia.h: Likewise.
20223         * config/sh/linux.h: Likewise.
20224         * config/sh/sh.c: Likewise.
20225         * config/sh/superh.h: Likewise.
20226         * config/sh/elf.h: Likewise.
20227         * config/sh/sh4.md: Likewise.
20228         * config/sh/sh.h: Likewise.
20230 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20232         * config/pa/constraints.md: Adjust unused letters.  Change "T"
20233         constraint to match_test floating_point_store_memory_operand().
20234         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
20235         (base14_operand): New.
20236         (floating_point_store_memory_operand): New.
20237         (integer_store_memory_operand): Revise to use base14_operand and
20238         reg_plus_base_memory_operand.
20239         (move_dest_operand): Allow symbolic_memory_operands.
20240         (symbolic_memory_operand): Check for LO_SOM.
20241         (symbolic_operand): Change default case to break.
20242         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
20243         CONST_DOUBLE values to be reloaded by putting them into memory when
20244         the destination is a floating point register.
20245         (movdf): Remove code to handle CONST_DOUBLE.
20246         (movsf): Likewise.
20247         (reload_indf_r1): New.
20248         (reload_insf_r1): New.
20249         Consistently use "Q" and "T" constraints with integer and floating
20250         point move instructions, respectively.
20251         (movdi): Remove FAIL.
20252         Change predicate for source operand unamed DImode move from
20253         general_operand to move_src_operand.
20254         (umulsidi3): Change predicate for destination operand to
20255         register_operand.
20256         Likewise for similar unamed patterns.
20257         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
20258         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
20259         (hppa_legitimize_address): Simplify mask calculation.
20260         (pa_emit_move_sequence): Revised handling of secondary reloads from
20261         REG+D addresses for floating point loads and stores.  Directly handle
20262         loading CONST0_RTX (mode) to a floating point register.
20263         (pa_secondary_reload): Handle reloading DF and SFmode constant values
20264         to floating point registers.  Don't restrict secondary reloads to
20265         floating point registers to integer modes.  Revise some comments and
20266         cleanup some code.
20267         (TARGET_LEGITIMATE_ADDRESS_P): Define.
20268         (pa_legitimate_address_p): New.
20269         (pa_legitimize_reload_address): New.
20270         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
20271         (STRICT_REG_OK_FOR_BASE_P): New.
20272         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
20273         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
20275 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
20276             Andrew Dixie  <andrewd@gentrack.com>
20278         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
20279         flag set.
20281 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
20283         * expmed.c (extract_bit_field_1): Pass the full width of the
20284         structure to get_best_reg_extraction_insn.
20286 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
20288         PR target/54601
20289         * configure.ac (use_cxa_atexit): Add AIX.
20290         * configure: Regenerate.
20292         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
20294 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
20296         PR debug/54793
20297         * final.c (need_profile_function): New variable.
20298         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
20299         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
20300         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
20301         notes, targetm.asm_out.function_prologue doesn't emit anything,
20302         HAVE_prologue and profiler should be emitted before prologue,
20303         set need_profile_function instead of emitting it.
20304         (final_scan_insn): If need_profile_function, emit
20305         profile_function on the first NOTE_INSN_BASIC_BLOCK or
20306         NOTE_INSN_FUNCTION_BEG note.
20308 2013-02-01  Richard Henderson  <rth@redhat.com>
20310         * config/rs6000/rs6000.md (smulditi3): New.
20311         (umulditi3): New.
20313         * config/alpha/alpha.md (umulditi3): New.
20315 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
20317         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
20318         (ASM_OUTPUT_ALIGNED_LOCAL): New.
20320 2013-02-01  Richard Biener  <rguenther@suse.de>
20322         PR tree-optimization/56113
20323         * tree-ssa-structalias.c (label_visit): Reduce work for
20324         single-predecessor nodes.
20326 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
20328         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
20329         range isn't testing for zero.
20331 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
20333         PR middle-end/56113
20334         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
20336 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
20337             Nick Clifton  <nickc@redhat.com>
20339         * config/v850/constraints.md (Q): Define as a memory constraint.
20340         * config/v850/predicates.md (label_ref_operand): New predicate.
20341         (e3v5_shift_operand): New predicate.
20342         (ior_operator): New predicate.
20343         * config/v850/t-v850: Add e3v5 multilib.
20344         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
20345         (v850_gen_movdi): Prototype.
20346         * config/v850/v850.c: Add support for e3v5 architecture.
20347         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
20348         TARGET_V850E_UP.
20349         (construct_save_jarl): Add e3v5 long JARL support.
20350         (v850_adjust_insn_length): New function.  Adjust length of call
20351         insns when using e3v5 instructions.
20352         (v850_gen_movdi): New function: Generate instructions to move a
20353         DImode value.
20354         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
20355         (CPP_SPEC): Define __v850e3v5__ as appropriate.
20356         (TARGET_USE_FPU): Enable for e3v5.
20357         (CONST_OK_FOR_W): New macro.
20358         (ADJUST_INSN_LENGTH): Define.
20359         * config/v850/v850.md (UNSPEC_LOOP): Define.
20360         (attr cpu): Add v850e3v5.
20361         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
20362         (movdi): New pattern.
20363         (movdi_internal): New pattern.
20364         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
20365         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
20366         (cstoresf4): Likewise.
20367         (cstoredf4): Likewise.
20368         (insv): New pattern.
20369         (rotlso3_a): New pattern.
20370         (rotlsi3_b): New pattern
20371         (rotlsi3_v850e3v5): New pattern.
20372         (doloop_begin): New pattern.
20373         (fix_loop_counter): New pattern.
20374         (doloop_end): New pattern.
20375         (branch_normal): Add e3v5 long branch support.
20376         (branch_invert): Likewise.
20377         (branch_z_normal): Likewise.
20378         (branch_z_invert): Likewise.
20379         (branch_nz_normal): Likewise.
20380         (branch_nz_invert): Likewise.
20381         (call_internal_short): Add e3v5 register-indirect JARL support.
20382         (call_internal_long): Likewise.
20383         (call_value_internal_short): Likewise.
20384         (call_value_internal_long): Likewise.
20385         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
20386         (mloop): New option.
20387         * config.gcc: Add support for configuring v840e3v5 target.
20388         * doc/invoke.texi: Document new v850 specific command line options.
20390 2013-01-31  Paul Koning  <ni1d@arrl.net>
20392         PR debug/55059
20393         PR debug/54508
20394         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
20395         children if parent is a class.
20396         (prune_unused_types_prune): Don't add DW_AT_declaration.
20398 2013-01-31  Richard Biener  <rguenther@suse.de>
20400         PR tree-optimization/56157
20401         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
20402         match up operand with SLP child.
20404 2013-01-31  Jason Merrill  <jason@redhat.com>
20406         PR debug/54410
20407         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
20408         parameters the first time.
20409         (gen_scheduled_generic_parms_dies): Check completeness here.
20411 2013-01-31  Richard Biener  <rguenther@suse.de>
20413         PR middle-end/53073
20414         * common.opt (faggressive-loop-optimizations): New flag,
20415         enabled by default.
20416         * doc/invoke.texi (faggressive-loop-optimizations): Document.
20417         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
20418         infer_loop_bounds_from_undefined by it.
20420 2013-01-31  Richard Biener  <rguenther@suse.de>
20422         PR tree-optimization/56150
20423         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
20424         visit virtual operands.
20425         (find_uses_to_rename_bb): Likewise.
20427 2013-01-31  Richard Biener  <rguenther@suse.de>
20429         PR tree-optimization/56150
20430         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
20431         mixed store non-store stmts.
20433 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
20435         PR sanitizer/55374
20436         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
20437         LIBASAN_EARLY_SPEC is defined.
20438         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
20439         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
20440         before %o.
20441         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
20443         PR c++/55742
20444         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
20445         invalid args instead of ICEing on it.
20446         (ix86_valid_target_attribute_tree): Return error_mark_node if
20447         ix86_valid_target_attribute_inner_p failed.
20448         (ix86_valid_target_attribute_p): Return false only if
20449         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
20450         target("default") attribute.
20451         (sorted_attr_string): Change argument from const char * to tree,
20452         merge in all target attribute arguments rather than just one.
20453         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
20454         instead of free.  Avoid using strcat.
20455         (ix86_mangle_function_version_assembler_name): Mangle
20456         target("default") as if no target attribute is present.  Adjust
20457         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
20458         instead of xmalloc and XDELETEVEC instead of free.
20459         (ix86_function_versions): Don't return true if one of the decls
20460         doesn't have target attribute.  If they don't and one of the decls
20461         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
20462         sorted_attr_string caller.  Use XDELETEVEC instead of free.
20463         (ix86_supports_function_versions): Remove.
20464         (make_name): Fix up formatting.
20465         (make_dispatcher_decl): Remove resolver_name and its initialization.
20466         Avoid leaking memory.
20467         (is_function_default_version): Return true if there is
20468         target("default") attribute rather than no target attribute at all.
20469         (make_resolver_func): Avoid leaking memory.
20470         (ix86_generate_version_dispatcher_body): Likewise.
20471         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
20472         * target.def (supports_function_versions): Remove.
20473         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
20474         * doc/tm.texi: Regenerated.
20476 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
20478         PR rtl-optimization/56144
20479         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
20480         for values with side effects.
20482 2013-01-30  Richard Biener  <rguenther@suse.de>
20484         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
20485         (sparseset_pop): Likewise.
20486         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
20487         to be able to use quick_push in the worker loop.
20489 2013-01-30  Marek Polacek  <polacek@redhat.com>
20491         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
20493 2013-01-30  Richard Biener  <rguenther@suse.de>
20495         PR lto/56147
20496         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
20498 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
20500         PR tree-optimization/56064
20501         * fixed-value.c (fixed_from_double_int): New function.
20502         * fixed-value.h (fixed_from_double_int): New prototype.
20503         (const_fixed_from_double_int): New static inline function.
20504         * fold-const.c (native_interpret_fixed): New static function.
20505         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
20506         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
20507         (native_encode_fixed): New static function.
20508         (native_encode_expr) <FIXED_CST>: Use it.
20509         (native_interpret_int): Move double_int worker code to...
20510         * double-int.c (double_int::from_buffer): ...this new static method.
20511         * double-int.h (double_int::from_buffer): Prototype it.
20513 2013-01-30  Richard Biener  <rguenther@suse.de>
20515         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
20516         New pointer-map and obstack.
20517         (init_alias_vars): Allocate pointer-map and obstack.
20518         (delete_points_to_sets): Free them.
20519         (find_what_var_points_to): Cache result.
20520         (find_what_p_points_to): Adjust for changed interface of
20521         find_what_var_points_to.
20522         (compute_points_to_sets): Likewise.
20523         (ipa_pta_execute): Likewise.
20525 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20527         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
20528         * configure: Regenerate.
20529         * config.in: Regenerate.
20530         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
20531         #nobits/#progbits if supported.
20533 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
20535         PR target/56121
20536         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
20537         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
20538         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
20540 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20542         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
20543         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
20545 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20547         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
20548         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
20550 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20552         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
20553         declaration.
20554         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
20555         * config/arm/cortex-a7.md: New bypasses using
20556         arm_mac_accumulator_is_result.
20558 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20560         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
20561         (cortex_a7_neon_mla): Likewise.
20562         (cortex_a7_fpfmad): New reservation.
20563         (cortex_a7_fpmacs): Use ffmas and update required units.
20564         (cortex_a7_fpmuld): Update required units and latency.
20565         (cortex_a7_fpmacd): Likewise.
20566         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
20567         (cortex_a7_neon). Likewise.
20568         (bypass) Update participating units.
20570 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20572         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
20573         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
20574         from fmac to ffma.
20575         * config/arm/vfp11.md (vfp_farith): Use ffmas.
20576         (vfp_fmul): Use ffmad.
20577         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
20578         (cortex_r4_fmacd): Use ffmad.
20579         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
20580         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
20581         (cortex_a9_fmacd): Use ffmad.
20582         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
20583         (cortex_a8_vfp_macd): Use ffmad.
20584         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
20585         (cortex_a5_fpmacd): Use ffmad.
20586         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
20587         (cortex_a15_vfp_macd): Use ffmad.
20588         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
20590 2013-01-29  Jason Merrill  <jason@redhat.com>
20592         PR libstdc++/54314
20593         * varasm.c (default_assemble_visibility): Don't warn about
20594         visibility on artificial decls.
20596 2013-01-29  Richard Biener  <rguenther@suse.de>
20598         PR tree-optimization/56113
20599         * tree-ssa-structalias.c (equiv_class_lookup): Also return
20600         the bitmap leader.
20601         (label_visit): Free duplicate bitmaps and record the leader instead.
20602         (perform_var_substitution): Adjust.
20604 2013-01-29  Richard Biener  <rguenther@suse.de>
20606         PR tree-optimization/55270
20607         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
20608         the CFG, schedule loops for fixup.
20610 2013-01-29  Nick Clifton  <nickc@redhat.com>
20612         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
20613         SP_REG.
20615 2013-01-28  Leif Ekblad  <leif@rdos.net>
20617         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
20618         * config/i386/i386.h (TARGET_RDOS): New macro.
20619         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
20620         * config/i386/i386.c (ix86_option_override_internal): For 64bit
20621         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
20622         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
20623         DEFAULT_LARGE_SECTION_THRESHOLD.
20624         * config/i386/i386.md (R14_REG, R15_REG): New constants.
20625         * config/i386/rdos.h: New file.
20626         * config/i386/rdos64.h: New file.
20628 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
20630         PR other/54814
20631         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
20632         TEST_HARD_REG_BIT.
20634 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
20636         PR rtl-optimization/56117
20637         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
20638         call cselib_lookup_from_insn on the MEM before calling
20639         add_insn_mem_dependence.
20641 2013-01-28  Richard Biener  <rguenther@suse.de>
20643         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
20644         to a stmt that didn't have one.
20645         (copy_phis_for_bb): Likewise for PHI arguments.
20646         (copy_debug_stmt): Likewise for debug stmts.
20648 2013-01-28  Richard Biener  <rguenther@suse.de>
20650         PR tree-optimization/56034
20651         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
20652         (partition_builtin_p): Adjust.
20653         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
20654         it is the last partition.
20655         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
20656         up the vertex for the definition.
20657         (classify_partition): Classify whether a partition is a
20658         PKIND_REDUCTION, thus has uses outside of the loop.
20659         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
20660         Merge all PKIND_REDUCTION partitions into the last partition.
20661         (tree_loop_distribution): Seed partitions from reductions as well.
20663 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
20665         PR tree-optimization/56125
20666         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
20667         pow(x,c) into sqrt(x) * powi(x, n/2) or
20668         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
20669         optimizing for size.
20670         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
20671         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
20672         integer.
20674         PR tree-optimization/56094
20675         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
20676         to UNKNOWN_LOCATION while gimplifying expr.
20678 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
20680         PR target/56114
20681         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
20682         operand 0 in movabs insn template for -masm=intel asm alternative.
20683         (*movabs<mode>_2): Ditto for operand 1.
20685 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
20687         PR target/54663
20688         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
20689         of microblaze-c.o
20691 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20693         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
20694         tm_file.
20696 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20698         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
20699         Undef to avoid warning.
20701 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
20703         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
20704         * configure: Regenerate.
20706 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
20708         PR tree-optimization/56098
20709         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
20710         for stmts with volatile ops.
20711         (cond_store_replacement): Don't optimize if assign has volatile ops.
20712         (cond_if_else_store_replacement_1): Don't optimize if either
20713         then_assign or else_assign have volatile ops.
20714         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
20715         volatile ops.
20717 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
20719         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
20721 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
20723         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
20724         missing ':' in asm example.
20726 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
20728         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
20729         entries into lane and laneq entries.
20730         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
20731         Remove AdvSIMD scalar modes.
20732         (aarch64_sq<r>dmulh_laneq<mode>): New.
20733         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
20734         modes.
20735         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
20736         builtin implementations to relfect changes in RTL in aarch64-simd.md.
20737         * config/aarch64/iterators.md (VCOND): New.
20738         (VCONQ): New.
20740 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
20742         PR target/54222
20743         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
20744         Add NULL LIBNAME argument to existing definitions.
20745         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
20746         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
20747         * config/avr/avr.c (DEF_BUILTIN): Same.
20748         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
20749         (avr_expand_builtin): Expand to a vanilla call if a libgcc
20750         implementation is available (DECL_ASSEMBLER_NAME is set).
20751         (avr_fold_absfx): New static function.
20752         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
20753         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
20754         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
20755         AVR_BUILTIN_ABSLLK.
20756         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
20757         (abshk, absk, abslk, absllk): Provide as static inline functions.
20759 2013-01-25  Marek Polacek  <polacek@redhat.com>
20761         PR tree-optimization/56035
20762         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
20764 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
20766         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
20767         (*movtf_internal_rex64): Add (!o,C) alternative
20768         (*movxf_internal_rex64): Ditto.
20769         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
20771 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
20773         * doc/invoke.texi: fix typo.
20774         * doc/objc.texi: fix typo.
20776 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
20778         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
20779         for the first two alternatives.
20781 2013-01-24  Diego Novillo  <dnovillo@google.com>
20783         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
20784         (ggc-zone.o): Remove.
20785         * configure.ac: Remove option --with-gc.
20786         * configure: Re-generate.
20787         * doc/install.texi: Remove documentation for --with-gc.
20788         * gengtype.c (write_enum_defn): Remove.  Update all users.
20789         (write_Types_process_field): Remove generation of gt_e_* argument.
20790         (output_type_enum): Remove.  Update all users.
20791         (write_enum_defn): Remove.  Update all users.
20792         (enum alloc_zone): Remove.  Update all users.
20793         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
20794         * ggc-common.c (ggc_splay_alloc): Remove first argument.
20795         Update all callers.
20796         (struct ptr_data): Remove field TYPE.  Update all users.
20797         (gt_pch_note_object): Remove argument TYPE.  Update all users.
20798         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
20799         Update all users.
20800         * ggc-none.c (ggc_alloc_typed_stat): Remove.
20801         (struct alloc_zone): Remove.
20802         (ggc_internal_alloc_zone_stat): Remove.
20803         (ggc_internal_cleared_alloc_zone_stat): Remove.
20804         * ggc-page.c (ggc_alloc_typed_stat): Remove.
20805         (ggc_pch_count_object): Remove last argument.  Update all users.
20806         (ggc_pch_alloc_object): Remove last argument.  Update all users.
20807         (struct alloc_zone): Remove.
20808         * ggc-zone.c: Remove.
20809         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
20810         (struct alloc_zone): Remove.
20811         (ggc_alloc_typed_stat): Remove.
20812         (ggc_alloc_typed): Remove.
20813         (ggc_splay_alloc): Remove first argument.
20814         (rtl_zone): Remove.  Update all users.
20815         (tree_zone): Remove.  Update all users.
20816         (tree_id_zone): Remove.  Update all users.
20817         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
20818         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
20819         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
20820         * tree-ssanames.c: Remove references to zone allocator in comments.
20822 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
20824         * config/avr/avr.c (avr_out_fract): Make register numbers that
20825         might be outside of source operand signed.
20827 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
20829         * config/i386/constraints.md (Yf): New constraint.
20830         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
20831         of f constraint to conditionaly disable x87 register preferences.
20832         (*movdf_internal): Ditto.
20833         (*movsf_internal): Ditto.
20835 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
20837         PR inline-asm/55934
20838         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
20839         that have operands with impossible constraints.
20840         Add a FIXME for a speed-up opportunity.
20841         * lra-constraints.c (process_alt_operands): Verify that a class
20842         selected from constraints on asms is valid for the operand mode.
20843         (curr_insn_transform): Remove incorrect comment.
20845 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
20847         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
20848         TOC operand is a valid symbol ref in the constant pool.
20850 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20852         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
20854 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
20856         PR target/54222
20857         * config/avr/stdfix.h: New file.
20858         * t-avr (stdfix-gcc.h): New rule to build it.
20859         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
20861 2013-01-23  Kostya Serebryany  <kcc@google.com>
20863         * config/darwin.h: remove dependency on
20864         CoreFoundation (asan on Mac OS).
20866 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
20868         PR target/49069
20869         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
20870         instead of cmpdi_operand for first comparison operand.
20871         Don't assert that comparison operands aren't both constants.
20873 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
20875         * doc/install.texi (Downloading the Source): Update references to
20876         downloading separate components.
20878 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
20880         * doc/extend.texi (__int128): Improve grammar.
20882 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
20884         PR target/56028
20885         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
20886         alternative to (o,r).
20887         (*movdi_internal_rex64): Remove (!o,n) alternative.
20888         (DImode immediate->memory splitter): Remove.
20889         (DImode immediate->memory peephole2): Remove.
20890         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
20891         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
20892         alternative to (!o,*r).
20893         (*movtf_internal_sse): New pattern.
20894         (*movxf_internal_rex64): New pattern.
20895         (*movxf_internal): Disable for TARGET_64BIT.
20896         (*movdf_internal_rex64): Remove (!o,F) alternative.
20898 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
20900         PR middle-end/56074
20901         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
20902         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
20903         * tree-vect-loop-manip.c (find_loop_location): Also ignore
20904         stmt locations where LOCATION_LOCUS of the stmt location is
20905         UNKNOWN_LOCATION or BUILTINS_LOCATION.
20907         PR target/55686
20908         * config/i386/i386.md (UNSPEC_STOS): New.
20909         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
20910         *strsetqi_1): Add UNSPEC_STOS.
20912 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
20914         PR c++/56067
20915         * doc/invoke.texi: Remove left over -Wsynth example.
20917 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
20919         PR tree-optimization/56051
20920         * fold-const.c (fold_binary_loc): Don't fold
20921         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
20922         a narrowing conversion, or widening conversion from signed
20923         to unsigned.
20925 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
20927         PR rtl-optimization/56023
20928         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
20929         dependent on debug instruction.
20931 2013-01-21  Martin Jambor  <mjambor@suse.cz>
20933         PR middle-end/56022
20934         * function.c (allocate_struct_function): Call
20935         invoke_set_current_function_hook earlier.
20937 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
20939         * reload1.c (init_reload): Only initialize reload_obstack
20940         during the first call.
20942 2013-01-21  Marek Polacek  <polacek@redhat.com>
20944         * cfgloop.c (verify_loop_structure): Fix up grammar.
20946 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
20948         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
20949         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
20951 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
20953         PR target/56058
20954         * config/arm/marvell-pj4.md: Update copyright year.
20955         Fix up use of alu to alu_reg and simple_alu_imm.
20957 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
20959         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
20961 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
20963         PR target/55433
20964         * lra-constraints.c (curr_insn_transform): Don't reuse original
20965         insn for secondary memory move when memory mode should be different.
20967 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20969         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
20970         atomic_storedi_1): New patterns.
20972 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
20974         btver2 pipeline descriptions.
20975         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
20976         descriptions.
20977         * config/i386/i386.md (btver2_decode): New type attributes.
20978         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
20979         type attributes.
20980         * config/i386/btver2.md: New file describing btver2 pipelines.
20982 2013-01-19  Andrew Pinski  <apinski@cavium.com>
20984         PR tree-optimization/52631
20985         * tree-ssa-sccvn (visit_use): Before looking up the original
20986         statement, try looking up the simplified expression.
20988 2013-01-19  Anthony Green  <green@moxielogic.com>
20990         * config/moxie/moxie.c (moxie_expand_prologue): Set
20991         current_function_static_stack_size.
20993 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
20995         PR tree-optimization/56029
20996         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
20997         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
20999 2013-01-18  Sharad Singhai  <singhai@google.com>
21001         PR tree-optimization/55995
21002         * dumpfile.c (dump_loc): Print location only if available.
21003         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
21005 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
21007         PR target/55433
21008         * lra-constraints.c (curr_insn_transform): Reuse original insn for
21009         secondary memory move.
21010         (inherit_reload_reg): Use rclass instead of cl for
21011         check_secondary_memory_needed_p.
21013 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
21015         PR middle-end/56015
21016         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
21017         the case where writing real complex part of target modifies op1.
21019 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
21021         * config/aarch64/aarch64-simd.md
21022         (aarch64_vcond_internal<mode>): Handle unordered cases.
21023         * config/aarch64/iterators.md (v_cmp_result): New.
21025 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
21026             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
21028         * config/arm/marvell-pj4.md: New file.
21029         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
21030         * config/arm/arm.md (generic_sched): Add marvell_pj4.
21031         (generic_vfp): Likewise.
21032         * config/arm/arm-cores.def: Add marvell-pj4.
21033         * config/arm/arm-tune.md: Regenerate.
21034         * config/arm/arm-tables.opt: Regenerate.
21035         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
21036         * doc/invoke.texi: Document marvell-pj4.
21038 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
21040         * config/aarch64/arm_neon.h: Map scalar types to standard types.
21042 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
21044         PR debug/54114
21045         PR debug/54402
21046         PR debug/49888
21047         * var-tracking.c (negative_power_of_two_p): New.
21048         (global_get_addr_cache, local_get_addr_cache): New.
21049         (get_addr_from_global_cache, get_addr_from_local_cache): New.
21050         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
21051         heading comment.
21052         (vt_stack_offset_p): Remove.
21053         (vt_canon_true_dep): Always canonicalize loc's address.
21054         (clobber_overlapping_mems): Make sure we have a MEM.
21055         (local_get_addr_clear_given_value): New.
21056         (val_reset): Clear local cached entries.
21057         (compute_bb_dataflow): Create and release the local cache.
21058         Disable duplicate MEMs clobbering.
21059         (emit_notes_in_bb): Clobber MEMs likewise.
21060         (vt_emit_notes): Create and release the local cache.
21061         (vt_initialize, vt_finalize): Create and release the global
21062         cache, respectively.
21063         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
21065 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
21067         PR libmudflap/53359
21068         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
21069         not found in the symtab.
21071 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
21073         PR debug/56006
21074         PR rtl-optimization/55547
21075         PR rtl-optimization/53827
21076         PR debug/53671
21077         PR debug/49888
21078         * alias.c (offset_overlap_p): New, factored out of...
21079         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
21080         the conservative special case for symbolic constants.  Don't
21081         adjust zero sizes on alignment.
21083 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
21085         PR rtl-optimization/52573
21086         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
21087         REG_UNUSED for the same register.
21089 2013-01-17  Richard Biener  <rguenther@suse.de>
21090             Marek Polacek  <polacek@redhat.com>
21092         PR rtl-optimization/55833
21093         * loop-unswitch.c (unswitch_loops): Move loop verification...
21094         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
21095         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
21096         Set it to true when we're removing a loop from hierarchy tree in
21097         an irreducible region.
21098         (fix_bb_placements): Adjust caller.
21099         (fix_loop_placements): Likewise.
21101 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
21103         * config/avr/builtins.def (DEF_BUILTIN): Factor out
21104         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
21105         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
21106         Remove ID.  Adjust comments.
21107         * config/avr/avr-c.c (avr_builtin_name): Remove.
21108         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
21109         * config/avr/avr.c (avr_tolower): New static function.
21110         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
21111         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
21112         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
21113         default expansion.
21115 2013-01-17  Jan Hubicka  <jh@suse.cz>
21117         PR tree-optimization/55273
21118         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
21120 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
21122         PR target/55981
21123         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
21124         store through atomic_store<mode>_1.
21125         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
21127 2013-01-17  Martin Jambor  <mjambor@suse.cz>
21129         PR tree-optimizations/55264
21130         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
21131         for virtual methods.
21132         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
21133         virtual methods before inlining is over.
21134         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
21135         virtual functions.
21136         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
21137         non-virtual.
21139 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
21141         PR rtl-optimization/56005
21142         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
21143         pending reads for prefetch.
21145 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
21147         * config/aarch64/aarch64.md
21148         (*cstoresi_neg_uxtw): New pattern.
21149         (*cmovsi_insn_uxtw): New pattern.
21150         (*<optab>si3_uxtw): New pattern.
21151         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
21152         (*<optab>si3_insn_uxtw): New pattern.
21153         (*bswapsi2_uxtw): New pattern.
21155 2013-01-16  Richard Biener  <rguenther@suse.de>
21157         * tree-inline.c (tree_function_versioning): Remove set but
21158         never used variable.
21160 2013-01-16  Richard Biener  <rguenther@suse.de>
21162         PR tree-optimization/55964
21163         * tree-flow.h (rename_variables_in_loop): Remove.
21164         (rename_variables_in_bb): Likewise.
21165         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
21166         (copy_loop_before): Adjust and delete update-ssa status.
21167         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
21168         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
21169         (rename_variables_in_loop): Remove.
21170         (slpeel_update_phis_for_duplicate_loop): Likewise.
21171         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
21172         use available cfg machinery instead of duplicating it.
21173         Update PHI nodes and perform poor-mans SSA update here.
21174         (slpeel_tree_peel_loop_to_edge): Adjust.
21176 2013-01-16  Richard Biener  <rguenther@suse.de>
21178         PR tree-optimization/54767
21179         PR tree-optimization/53465
21180         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
21181         (vrp_visit_phi_node): For PHI arguments coming via backedges
21182         drop all symbolical range information.
21183         (execute_vrp): Compute backedges.
21185 2013-01-16  Richard Biener  <rguenther@suse.de>
21187         * doc/install.texi: Update CLooG and ISL requirements to
21188         0.18.0 and 0.11.1.
21190 2013-01-16  Christian Bruel  <christian.bruel@st.com>
21192         PR target/55301
21193         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
21194         (broken_move): Handle UNSPECV_SP_SWITCH_B.
21195         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
21197 2013-01-16  DJ Delorie  <dj@redhat.com>
21199         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
21200         (UNSPECV_SP_SWITCH_E): New.
21201         (sp_switch_1): Change to an unspec.
21202         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
21203         replace $r15.
21205 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
21207         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
21208         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
21209         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
21210         (expand_mem_thread_fence): Ditto.
21211         (expand_mem_signal_fence): Ditto.
21212         (expand_atomic_load): Ditto.
21213         (expand_atomic_store): Ditto.
21215 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
21217         PR rtl-optimization/55547
21218         PR rtl-optimization/53827
21219         PR debug/53671
21220         PR debug/49888
21221         * alias.c (memrefs_conflict_p): Set sizes to negative after
21222         AND adjustments.
21224 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
21226         PR target/55940
21227         * function.c (thread_prologue_and_epilogue_insns): Always
21228         add crtl->drap_reg to set_up_by_prologue.set, even if
21229         stack_realign_drap is false.
21231 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
21233         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
21234         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
21235         *call): Fix indention.
21237 2013-01-15  Tom de Vries  <tom@codesourcery.com>
21239         PR target/55876
21240         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
21241         Update comment.
21243 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
21245         PR rtl-optimization/55153
21246         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
21248 2013-01-15  Martin Jambor  <mjambor@suse.cz>
21250         PR tree-optimization/55920
21251         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
21252         accesses as grp_to_be_debug_replaced.
21254 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
21256         PR tree-optimization/55920
21257         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
21258         there is non-useless type conversion needed from debug rhs to lhs,
21259         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
21261 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
21262             Mikael Pettersson  <mikpe@it.uu.se>
21264         PR target/43961
21265         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
21266         Thumb.
21267         (ASM_OUTPUT_CASE_LABEL): Remove.
21268         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
21269         * final.c (shorten_branches): Update alignment of labels before
21270         jump tables if CASE_VECTOR_SHORTEN_MODE.
21272 2013-01-15  Richard Biener  <rguenther@suse.de>
21274         PR bootstrap/55961
21275         * system.h: Do not include gmp.h for building host tools.
21277 2013-01-15  Richard Biener  <rguenther@suse.de>
21279         PR middle-end/55882
21280         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
21281         account for bitpos when computing alignment.
21283 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
21285         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
21286         (ix86_target_macros_internal): Likewise.
21288         * config/i386/i386.c (m_CORE2I7): Removed.
21289         (m_CORE_HASWELL): New macro.
21290         (m_CORE_ALL): Likewise.
21291         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
21292         (initial_ix86_arch_features): Likewise.
21293         (processor_target_table): Initializations for Core avx2.
21294         (cpu_names): New names "core-avx2".
21295         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
21296         PROCESSOR_CORE_HASWELL.
21297         (ix86_issue_rate): New case.
21298         (ia32_multipass_dfa_lookahead): Likewise.
21299         (ix86_sched_init_global): Likewise.
21301         * config/i386/i386.h (TARGET_HASWELL): New macro.
21302         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
21303         (processor_type): New PROCESSOR_HASWELL.
21305 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
21307         PR tree-optimization/55955
21308         * tree-vect-loop.c (vectorizable_reduction): Give up early on
21309         *SHIFT_EXPR and *ROTATE_EXPR codes.
21311         PR tree-optimization/48766
21312         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
21313         -ftrapv disable -fwrapv.
21315 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
21317         PR target/55974
21318         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
21319         etc. to 1 and not to __flash.
21320         Use LL suffix for __INT24_MAX__ with -mint8.
21321         Use ULL suffix for __UINT24_MAX__ with -mint8.
21323 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
21325         * config/avr/avr-arch.h
21326         (struct base_arch_s): Use typedef avr_arch_t instead.
21327         (struct arch_info_s): Use typedef avr_arch_info_t instead.
21328         (struct mcu_type_s): Use typedef avr_mcu_t instead.
21329         * config/avr/avr.c: Same.
21330         * config/avr/avr-devices.c: Same.
21331         * config/avr/driver-avr.c: Same.
21332         * config/avr/gen-avr-mmcu-texi.c: Same.
21333         * config/avr/avr-mcus.def: Adjust comment.
21335 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
21337         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
21338         * config/aarch64/iterators.md (VALLDI): New.
21340 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
21341             Andi Kleen  <ak@linux.intel.com>
21343         PR target/55948
21344         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
21345         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
21346         memmodel flag.
21348 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
21350         * config/avr/avr-stdint.h: Remove trailing blanks.
21351         * config/avr/avr-log.h: Same.
21352         * config/avr/avr-arch.h: Same.
21353         * config/avr/avr-devices.c: Same.
21354         * config/avr/avr-dimode.md: Same.
21355         * config/avr/predicates.md: Same.
21356         * config/avr/avr-c.c: Same.  And fix typo.
21358         * config/avr/avr-protos.h: Same.  And:
21359         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
21360         (init_cumulative_args): Rename to avr_init_cumulative_args.
21361         (expand_prologue): Rename to avr_expand_prologue.
21362         (expand_epilogue): Rename to avr_expand_epilogue.
21363         (adjust_insn_length): Rename to avr_adjust_insn_length.
21364         (notice_update_cc): Rename to avr_notice_update_cc.
21365         (final_prescan_insn): Rename to avr_final_prescan_insn.
21366         * config/avr/avr.c: Same.
21367         * config/avr/avr.h: Same.
21368         * config/avr/avr.md: Remove trailing blanks.
21369         (prologue): Use avr_expand_prologue.
21370         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
21372 2013-01-14  Richard Biener  <rguenther@suse.de>
21374         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
21375         verify_location, collect_subblocks): New functions.
21376         (verify_gimple_in_cfg): Verify that locations only reference
21377         BLOCKs in the functions BLOCK tree.
21379 2013-01-14  Richard Biener  <rguenther@suse.de>
21381         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
21382         PHI argument.
21383         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
21384         unshare reference.
21385         (insert_out_of_ssa_copy_on_edge): Likewise.
21386         (rewrite_close_phi_out_of_ssa): Likewise.
21387         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
21388         debug expressions.
21389         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
21390         propagated constants.
21391         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
21392         can not be shared.
21394 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
21396         * config/avr/avr-modes.def: Add GPL copyright notice.
21398 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
21400         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
21401         MEMMODEL_MASK to determine memory model.
21402         (atomic_store<mode>): Ditto from operands[2].
21403         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
21405 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
21407         PR fortran/55935
21408         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
21409         (fold_gimple_assign): Don't call unshare_expr here.
21410         (fold_ctor_reference): Call unshare_expr.
21412 2013-01-13  Terry Guo  <terry.guo@arm.com>
21414         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
21415         * doc/fragments.texi: Document MULTILIB_REUSE.
21416         * gcc.c (multilib_reuse): New internal spec.
21417         (set_multilib_dir): Also search multilib from multilib_reuse.
21418         * genmultilib (tmpmultilib3): Refactor code.
21419         (tmpmultilib4): Ditto.
21420         (multilib_reuse): New multilib argument.
21422 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
21424         * Makefile.in: Update copyright.
21426 2013-01-12  Tom de Vries  <tom@codesourcery.com>
21428         PR middle-end/55890
21429         * calls.c (expand_call): Check if arg_nr is valid.
21431 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
21433         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
21434         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
21435         documentation.  Add missing '__' in front of
21436         __builtin_ia32_packssdw256.
21438 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21440         PR target/55719
21441         * config/s390/s390.c (s390_preferred_reload_class): Do not return
21442         NO_REGS for larl operands.
21443         (s390_reload_larl_operand): Use s390_load_address instead of
21444         emit_move_insn.
21446 2013-01-11  Richard Biener  <rguenther@suse.de>
21448         * tree-cfg.c (verify_node_sharing_1): Split out from ...
21449         (verify_node_sharing): ... here.
21450         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
21452 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
21454         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
21455         Substitute TREECHECKING.
21456         * configure: Regenerate.
21457         * Makefile.in (TREECHECKING): New.
21459 2013-01-11  Richard Guenther  <rguenther@suse.de>
21461         PR tree-optimization/44061
21462         * tree-vrp.c (extract_range_basic): Compute zero as
21463         value-range for __builtin_constant_p of function parameters.
21465 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
21467         Update copyright years.
21469 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
21471         PR rtl-optimization/55672
21472         * lra-eliminations.c (mark_not_eliminable): Permit addition with
21473         const to be eliminable.
21475 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
21477         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
21478         * configure: Regenerate.
21480 2013-01-10  Richard Biener  <rguenther@suse.de>
21482         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
21484 2013-01-10  Richard Biener  <rguenther@suse.de>
21486         PR bootstrap/55792
21487         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
21488         locations for virtual PHI arguments.
21489         (rewrite_update_phi_arguments): Likewise.
21491 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
21493         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
21494         on to assembler.
21496 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
21498         PR tree-optimization/55921
21499         * tree-complex.c (expand_complex_asm): New function.
21500         (expand_complex_operations_1): Call it for GIMPLE_ASM.
21502 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21504         PR target/55718
21505         * config/s390/s390.c (s390_symref_operand_p)
21506         (s390_loadrelative_operand_p): Merge the two functions.
21507         (s390_check_qrst_address, print_operand_address): Add parameters
21508         to s390_loadrelative_operand_p invokation.
21509         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
21510         (s390_reload_larl_operand, s390_secondary_reload): Use
21511         s390_loadrelative_operand_p instead of s390_symref_operand_p.
21512         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
21514 2013-01-09  Mike Stump  <mikestump@comcast.net>
21516         * dse.c (record_store): Remove unnecessary assert.
21518 2013-01-09  Jan Hubicka  <jh@suse.cz>
21520         PR tree-optimization/55569
21521         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
21522         * cfgloop.h (scale_loop_profile): Likewise.
21524 2013-01-09  Jan Hubicka  <jh@suse.cz>
21526         PR lto/45375
21527         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
21528         functions.
21529         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
21531 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
21533         PR middle-end/55114
21534         * expr.h (maybe_emit_group_store): Declare.
21535         * expr.c (maybe_emit_group_store): New function.
21536         * builtins.c (expand_builtin_int_roundingfn): Call it.
21537         (expand_builtin_int_roundingfn_2): Likewise.
21539 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
21541         PR rtl-optimization/55829
21542         * lra-constraints.c (match_reload): Add code for absent output.
21543         (curr_insn_transform): Add code for reloads of matched inputs
21544         without output.
21546 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
21548         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
21549         attribute of movddup insn to DF.
21550         (*vec_interleave_lowv2df): Ditto.
21551         (vec_dupv2df): Ditto.
21553 2013-01-09  Jan Hubicka  <jh@suse.cz>
21555         PR tree-optimiation/55875
21556         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
21557         EVERY_ITERATION parameter.
21558         (number_of_iterations_exit): Check if exit is executed every iteration.
21559         (idx_infer_loop_bounds): Similarly here.
21560         (n_of_executions_at_most): Simplify
21561         to only test for cases where statement is dominated by the
21562         particular bound; handle correctly the "postdominance" test.
21563         (scev_probably_wraps_p): Use max loop iterations info
21564         as a global bound first.
21566 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
21567             Nick Clifton  <nickc@redhat.com>
21569         * config/v850/v850.md (cbranchsf4): New pattern.
21570         (cstoresf4): New pattern.
21571         (cbranchdf4): New pattern.
21572         (cstoredf4): New pattern.
21573         (movsicc): Disallow floating point comparisons.
21574         (cmpsf_le_insn): Fix order of operators.
21575         (cmpsf_lt_insn): Likewise.
21576         (cmpsf_eq_insn): Likewise.
21577         (cmpdf_le_insn): Likewise.
21578         (cmpdf_lt_insn): Likewise.
21579         (cmpdf_eq_insn): Likewise.
21580         (cmpsf_ge_insn): Use LE comparison.
21581         (cmpdf_ge_insn): Likewise.
21582         (cmpsf_gt_insn): Use LT comparison.
21583         (cmpdf_gt_insn): Likewise.
21584         (cmpsf_ne_insn): Delete pattern.
21585         (cmpdf_ne_insn): Delete pattern.
21586         * config/v850/v850.c (v850_gen_float_compare): Use
21587         gen_cmpdf_eq_insn for NE comparison.
21588         (v850_float_z_comparison_operator)
21589         (v850_float_nz_comparison_operator): Move from here ...
21590         * config/v850/predicates.md: ... to here.  Move GT and GE
21591         comparisons into v850_float_z_comparison_operator.
21592         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
21593         Delete prototype.
21594         (v850_float_nz_comparison_operator): Likewise.
21596 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21598         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
21599         with calls to gen_insvsi/gen_insvdi.
21601 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
21603         * config/i386/i386.c (initial_ix86_tune_features): Set up
21604         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
21606 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
21607             Jakub Jelinek  <jakub@redhat.com>
21609         PR tree-optimization/48189
21610         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
21611         If nitercst is 0, don't predict the exit edge.
21613 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21615         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
21616         in asm_fprintf with reg_names.
21617         (aarch64_print_operand_address): Likewise.
21618         (aarch64_return_addr): Likewise.
21619         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
21621 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21623         * config/pa/pa.h (VAL_U6_BITS_P): Define.
21624         (INT_U6_BITS): Likewise.
21625         * config/pa/predicates.md (uint6_operand): New predicate.
21626         (shift5_operand, shift6_operand): Likewise.
21627         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
21628         arith32_operand.
21629         (lshrdi3): Use shift6_operand.
21630         (shrpsi4, shrpdi4): New insn patterns.
21631         (extzv): Delete expander.
21632         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
21633         predicates in unamed zero extract patterns.  Tighten common constraint.
21634         (extv): Delete expander.
21635         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
21636         predicates in unamed sign extract patterns.  Tighten common constraint.
21637         (insv): Delete expander.
21638         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
21639         predicates in unamed insert patterns.  Tighten common constraint.
21640         Change uint32_operand predicate to uint6_operand predicate in unamed
21641         DImode pattern to insert constant values of type 1...1xxxx.
21643 2013-01-04  Jan Hubicka  <jh@suse.cz>
21645         PR tree-optimization/55823
21646         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
21647         issue.
21649 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
21650             Uros Bizjak  <ubizjak@gmail.com>
21652         PR rtl-optimization/55845
21653         * df-problems.c (can_move_insns_across): Stop scanning at
21654         volatile_insn_p source instruction or give up if
21655         across_from .. across_to range contains any volatile_insn_p
21656         instructions.
21658 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
21660         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
21661         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
21662         Declare.
21663         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
21664         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
21666 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
21668         PR fortran/55341
21669         * asan.c (asan_clear_shadow): New function.
21670         (asan_emit_stack_protection): Use it.
21672 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
21674         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
21675         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
21676         with tab instead of space.
21678 2013-01-08  Nick Clifton  <nickc@redhat.com>
21680         * config/rl78/rl78.c (rl78_expand_prologue): Always select
21681         register bank 0 at the start of an interrupt handler.
21682         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
21683         MDBH registers.
21685 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
21687         * config/aarch64/aarch64-simd.md
21688         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
21689         (aarch64_simd_bsl): Likewise.
21690         (aarch64_vcond_internal<mode>): Likewise.
21691         (vcond<mode><mode>): Likewise.
21692         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
21693         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
21695 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
21697         * config/aarch64/aarch64-builtins.c
21698         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
21700 2013-01-08  Martin Jambor  <mjambor@suse.cz>
21702         PR debug/55579
21703         * tree-sra.c (analyze_access_subtree): Return true also after
21704         potentially creating a debug-only replacement.
21706 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
21708         PR middle-end/55890
21709         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
21711         PR tree-optimization/54120
21712         * tree-vrp.c (range_fits_type_p): Don't allow
21713         src_precision < precision from signed vr to unsigned_p
21714         if vr->min or vr->max is negative.
21715         (simplify_float_conversion_using_ranges): Test can_float_p
21716         against CODE_FOR_nothing.
21718 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
21719             Richard Biener  <rguenther@suse.de>
21721         PR middle-end/55851
21722         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
21723         types instead of just INTEGER_TYPE types.
21725 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
21727         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
21728         TF_SIZE): Define.
21730 2013-01-07  Steve Ellcey  <sellcey@mips.com>
21732         PR target/42661
21733         * config/mips/mips.opt: Change mad to mmad to match documentation.
21735 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21737         PR target/55897
21738         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
21739         .progmemx.data now.
21741 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21743         PR target/55897
21744         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
21745         (avr_addrspace_t): Add .section_name field.
21746         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
21747         array size.
21748         (avr_addrspace): Same.  Initialize .section_name.  Remove last
21749         NULL entry.  Put __memx into .progmemx.data.
21750         (progmem_section_prefix): Remove.
21751         (avr_asm_init_sections): No need to initialize progmem_section.
21752         (avr_asm_named_section): Use avr_addrspace[].section_name to get
21753         section name prefix.
21754         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
21755         retrieve the progmem section.
21756         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
21757         boundary to run over avr_addrspace[].
21758         (avr_register_target_pragmas): Ditto.
21760 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
21762         * varasm.c (output_constant_def_contents): For asan_protect_global
21763         protected strings, adjust DECL_ALIGN if needed, before testing for
21764         anchored symbols.
21765         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
21766         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
21767         normal decls.
21768         (output_object_block): For asan protected decls, emit asan padding
21769         after their contents.
21770         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
21771         (asan_finish_file): Test it here instead.
21773 2013-01-07  Nick Clifton  <nickc@redhat.com>
21774             Matthias Klose  <doko@debian.org>
21775             Doug Kwan  <dougkwan@google.com>
21776             H.J. Lu  <hongjiu.lu@intel.com>
21778         PR driver/55470
21779         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
21781         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
21783         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
21785         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
21787         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
21789 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21791         PR target/54461
21792         * doc/install.texi (Cross-Compiler-Specific Options): Document
21793         --with-avrlibc.
21795 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
21797         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
21798         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
21799         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
21800         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
21801         vqmovun_high_s64): Fix source operand number and update copyright.
21803 2013-01-07  Richard Biener  <rguenther@suse.de>
21805         PR middle-end/55890
21806         * gimple.h (gimple_call_builtin_p): New overload.
21807         * gimple.c (validate_call): New function.
21808         (gimple_call_builtin_p): Likewise.
21809         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
21810         Use gimple_call_builtin_p.
21811         (find_func_clobbers): Likewise.
21812         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
21813         (strlen_optimize_stmt): Likewise.
21815 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
21817         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
21818         (vld1q_dup_*): Likewise.
21819         (vld1_*): Likewise.
21820         (vld1q_*): Likewise.
21821         (vld1_lane_*): Likewise.
21822         (vld1q_lane_*): Likewise.
21824 2013-01-07  Richard Biener  <rguenther@suse.de>
21826         * lto-streamer.h (LTO_minor_version): Bump to 2.
21828 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
21830         * config/aarch64/aarch64-protos.h
21831         (aarch64_const_double_zero_rtx_p): Rename to...
21832         (aarch64_float_const_zero_rtx_p): ...this.
21833         (aarch64_float_const_representable_p): New.
21834         (aarch64_output_simd_mov_immediate): Likewise.
21835         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
21836         move immediate case.
21837         * config/aarch64/aarch64.c
21838         (aarch64_const_double_zero_rtx_p): Rename to...
21839         (aarch64_float_const_zero_rtx_p): ...this.
21840         (aarch64_print_operand): Allow printing of new constants.
21841         (aarch64_valid_floating_const): New.
21842         (aarch64_legitimate_constant_p): Check for valid floating-point
21843         constants.
21844         (aarch64_simd_valid_immediate): Likewise.
21845         (aarch64_vect_float_const_representable_p): New.
21846         (aarch64_float_const_representable_p): Likewise.
21847         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
21848         (aarch64_output_simd_mov_immediate): New.
21849         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
21850         (*movdf_aarch64): Likewise.
21851         * config/aarch64/constraints.md (Ufc): New.
21852         (Y): call aarch64_float_const_zero_rtx.
21853         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
21855 2013-01-07  Richard Biener  <rguenther@suse.de>
21857         PR tree-optimization/55888
21858         PR tree-optimization/55862
21859         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
21860         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
21861         not if it is contained therein.
21863 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21865         * config/avr/t-avr: Typo.
21867 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
21869         PR55243
21870         * config/avr/t-avr: Don't automatically rebuild
21871         $(srcdir)/config/avr/t-multilib
21872         $(srcdir)/config/avr/avr-tables.opt
21873         $(srcdir)/doc/avr-mmcu.texi
21874         (avr-mcus): New phony target to build them on request.
21875         (s-avr-mlib, s-avr-mmcu-texi): Remove.
21876         * avr/avr-mcus.def: Adjust comments.
21878 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
21880         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
21882 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
21884         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
21886 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
21888         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
21890 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
21892         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
21893         to generate profiling.
21894         * config/rs6000/aix64.h (LIB_SPEC): Same.
21896 2013-01-04  Andrew Pinski  <apinski@cavium.com>
21898         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
21899         New function.
21900         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
21902 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
21904         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
21905         unconditionally.
21906         (ix86_expand_move): Ditto.
21907         (ix86_zero_extend_to_Pmode): Ditto.
21908         (ix86_expand_call): Ditto.
21909         (ix86_expand_special_args_builtin): Ditto.
21910         (ix86_expand_builtin): Ditto.
21912 2013-01-04  Richard Biener  <rguenther@suse.de>
21914         PR tree-optimization/55862
21915         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
21916         translating them through PHI nodes.
21918 2013-01-04  Martin Jambor  <mjambor@suse.cz>
21920         PR tree-optimization/55755
21921         * tree-sra.c (sra_modify_assign): Do not check that an access has no
21922         children when trying to avoid producing a VIEW_CONVERT_EXPR.
21924 2013-01-04  Marek Polacek  <polacek@redhat.com>
21926         PR middle-end/55859
21927         * opts.c (default_options_optimization): Clarify error message.
21929 2013-01-04  Richard Biener  <rguenther@suse.de>
21931         PR middle-end/55863
21932         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
21933         reassociation.
21935 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21937         PR target/53789
21938         * config/pa/pa.md (movsi): Revert previous change.
21939         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
21940         references.
21942 2013-01-03  Richard Henderson  <rth@redhat.com>
21944         * config/i386/i386.c (ix86_expand_move): Always assign to op1
21945         after eliminating TLS symbols.
21947 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
21949         PR bootstrap/50167
21950         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
21951         * graphite-poly.c (debug_gmp_value): Likewise.
21953 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
21955         PR target/55712
21956         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
21957         selected code model, define __code_mode_small__, __code_model_medium__,
21958         __code_model_large__, __code_model_32__ or __code_model_kernel__.
21959         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
21960         xchg temporary register with %k.  Declare temporary register as
21961         early clobbered.
21962         [__x86_64__]: For medium and large code models, preserve %rbx register.
21964 2013-01-03  Richard Biener  <rguenther@suse.de>
21966         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
21967         (dump_subscript): Adjust.
21968         (finalize_ddr_dependent): Do not dump redundant info.
21969         (analyze_siv_subscript): Adjust.
21970         (subscript_dependence_tester): Likewise.
21971         (compute_affine_dependence): Likewise.
21973 2013-01-03  Richard Biener  <rguenther@suse.de>
21975         Revert
21976         2013-01-03  Richard Biener  <rguenther@suse.de>
21978         PR tree-optimization/55857
21979         * tree-vect-stmts.c (vectorizable_load): Do not setup
21980         re-alignment for invariant loads.
21982         2013-01-02  Richard Biener  <rguenther@suse.de>
21984         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
21985         invariant load do not generate a vector load from the scalar location.
21987 2013-01-03  Richard Biener  <rguenther@suse.de>
21989         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
21990         for not vectorizing.
21991         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
21992         not build INDIRECT_REFs, call get_name once only.
21993         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
21994         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
21996 2013-01-03  Richard Biener  <rguenther@suse.de>
21998         PR tree-optimization/55857
21999         * tree-vect-stmts.c (vectorizable_load): Do not setup
22000         re-alignment for invariant loads.
22002 2013-01-03  Richard Biener  <rguenther@suse.de>
22004         PR lto/55848
22005         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
22006         prefer a built-in decl.
22008 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
22010         * gcc.c (process_command): Update copyright notice dates.
22011         * gcov.c (print_version): Likewise.
22012         * gcov-dump.c (print_version): Likewise.
22014         PR rtl-optimization/55838
22015         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
22016         iv0.step, iv1.step and step.
22018 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
22019             Marc Glisse  <marc.glisse@inria.fr>
22021         PR tree-optimization/55832
22022         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
22023         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
22024         integer_{one,zero}_node.
22026 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
22028         PR debug/54402
22029         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
22030         * var-tracking.c (reverse_op): Don't add reverse ops to
22031         VALUEs that have already
22032         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
22034 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
22036         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
22038 2013-01-02  Teresa Johnson  <tejohnson@google.com>
22040         * dumpfile.c (dump_loc): Print filename with location.
22041         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
22042         new location_t parameter to emit complete unroll message with
22043         new dump framework.
22044         (canonicalize_loop_induction_variables): Compute loops location
22045         and pass to try_unroll_loop_completely.
22046         * loop-unroll.c (report_unroll_peel): New function.
22047         (peel_loops_completely): Use new dump format with location
22048         for main dumpfile message, and invoke report_unroll_peel on success.
22049         (decide_unrolling_and_peeling): Ditto.
22050         (decide_peel_once_rolling): Remove old dumpfile message subsumed
22051         by report_unroll_peel.
22052         (decide_peel_completely): Ditto.
22053         (decide_unroll_constant_iterations): Ditto.
22054         (decide_unroll_runtime_iterations): Ditto.
22055         (decide_peel_simple): Ditto.
22056         (decide_unroll_stupid): Ditto.
22057         * cfgloop.c (get_loop_location): New function.
22058         * cfgloop.h (get_loop_location): Declare.
22060 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
22062         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
22063         NULL.
22065 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
22067         PR middle-end/55198
22068         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
22069         BLKmode objects when EXPAND_MEMORY is specified.
22071 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
22073         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
22074         in loop predicate.
22075         (fold_builtin_cpu): Do not share cpu model decls across statements.
22077 2013-01-02  Jason Merrill  <jason@redhat.com>
22079         PR c++/55804
22080         * tree.c (build_array_type_1): Revert earlier change.
22082 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
22084         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
22085         "cortex-a57".
22086         * config/aarch64/aarch64-tune.md: Re-generate.
22088 2013-01-02  Richard Biener  <rguenther@suse.de>
22090         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
22091         invariant load do not generate a vector load from the scalar location.
22093 2013-01-02  Richard Biener  <rguenther@suse.de>
22095         PR bootstrap/55784
22096         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
22097         * configure: Regenerate.
22099 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
22101         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
22102         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
22103         (expand_builtin_int_roundingfn_2): Keep the original target around
22104         for the fallback case.
22106 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
22108         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
22109         to be clear for sign changes.
22111 2013-01-01  Jan Hubicka  <jh@suse.cz>
22113         * ipa-inline-analysis.c: Fix formatting.
22115 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
22117         PR tree-optimization/55831
22118         * tree-vect-loop.c (get_initial_def_for_induction): Use
22119         gsi_after_labels instead of gsi_start_bb.
22121 Copyright (C) 2013 Free Software Foundation, Inc.
22123 Copying and distribution of this file, with or without modification,
22124 are permitted in any medium without royalty provided the copyright
22125 notice and this notice are preserved.