2013-09-18 Marc Glisse <marc.glisse@inria.fr>
[official-gcc.git] / gcc / ChangeLog
blobd222ee09d2b3138bdcb45c69d4ad9f4dea4b9240
1 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
3         PR rtl-optimization/58438
4         * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
5         * lra-constraints.c (undo_optional_reloads): Keep optional reloads
6         from previous subpasses.
8 2013-09-18  Richard Earnshaw  <rearnsha@arm.com>
10         * arm.c (arm_get_frame_offsets): Validate architecture supports
11         LDRD/STRD before accepting the tuning preference.
12         (arm_expand_prologue): Likewise.
13         (arm_expand_epilogue): Likewise.
15 2013-09-18  Richard Biener  <rguenther@suse.de>
17         PR tree-optimization/58417
18         * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
19         have chrecs with symbols defined in the loop as operands.
20         (chrec_fold_multiply): Likewise.
21         * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
22         parameters before folding binary operations.
23         (struct instantiate_cache_entry_hasher): Remove.
24         (struct instantiate_cache_type): Use a pointer-map.
25         (instantiate_cache_type::instantiate_cache_type): New function.
26         (instantiate_cache_type::get): Likewise.
27         (instantiate_cache_type::set): Likewise.
28         (instantiate_cache_type::~instantiate_cache_type): Adjust.
29         (get_instantiated_value_entry): Likewise.
30         (global_cache): New global.
31         (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
32         instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
33         instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
34         (instantiate_scev_name): Adjust.
35         (instantiate_scev): Construct global instead of local cache.
36         (resolve_mixers): Likewise.
38 2013-09-18  Daniel Morris  <danielm@ecoscentric.com>
39             Paolo Carlini  <paolo.carlini@oracle.com>
41         PR c++/58458
42         * doc/implement-cxx.texi: Fix references to the C++ standards.
44 2013-09-18  Jakub Jelinek  <jakub@redhat.com>
46         * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
47         * tree-vect-data-refs.c (vect_analyze_data_refs): For
48         simd_lane_access drs, update also DR_ALIGNED_TO.
50 2013-09-18  Marek Polacek  <polacek@redhat.com>
52         PR sanitizer/58411
53         * doc/extend.texi: Document no_sanitize_undefined attribute.
54         * builtins.c (fold_builtin_0): Don't sanitize function if it has the
55         no_sanitize_undefined attribute.
57 2013-09-18  Nick Clifton  <nickc@redhat.com>
59         * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
60         (ASM_DECLARE_FUNCTION_NAME): Define.
62 2013-09-17  Trevor Saunders  <tsaunders@mozilla.com>
64         * compare-elim.c (find_comparison_dom_walker): New class
65         (find_comparisons_in_bb): Rename to
66         find_comparison_dom_walker::before_dom_children
67         (find_comparisons): Adjust
68         * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
69         adjust.
70         (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
71         * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
72         (init_walk_dominator_tree): Remove declaration.
73         (fini_walk_dominator_tree): Remove declaration.
74         * fwprop.c (single_def_use_dom_walker): New class
75         (single_def_use_enter_block): Convert to
76         single_def_use_dom_walker::before_dom_children.
77         (single_def_use_leave_block): Convert to
78         single_def_use_dom_walker::after_dom_children.
79         (build_single_def_use_links): Adjust.
80         * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
81         class.
82         (find_candidates_in_block): Convert to
83         find_candidates_dom_walker::before_dom_children.
84         (execute_strength_reduction): Adjust.
85         * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
86         (sese_dom_walker): New class.
87         (sese_dom_walker::sese_dom_walker): New constructor.
88         (sese_dom_walker::~sese_dom_walker): New destructor.
89         (build_sese_conditions_before): Convert to
90         sese_dom_walker::before_dom_children.
91         (build_sese_conditions_after): Convert to
92         sese_dom_walker::after_dom_children.
93         (build_poly_scop): Adjust
94         * tree-into-ssa.c (rewrite_dom_walker): New class
95         (rewrite_enter_block): Convert to
96         rewrite_dom_walker::before_dom_children.
97         (rewrite_leave_block): Convert to
98         rewrite_dom_walker::after_dom_children.
99         (rewrite_update_dom_walker): New class.
100         (rewrite_update_enter_block): Convert to
101         rewrite_update_dom_walker::before_dom_children.
102         (rewrite_update_leave_block): Convert to
103         rewrite_update_dom_walker::after_dom_children.
104         (rewrite_blocks, rewrite_into_ssa): Adjust.
105         (mark_def_dom_walker): New class.
106         (mark_def_dom_walker::mark_def_dom_walker): New constructor.
107         (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
108         (mark_def_sites_blocks): Convert to
109         mark_def_dom_walker::before_dom_children.
110         (mark_def_site_blocks): Remove.
111         * tree-ssa-dom.c (dom_opt_dom_walker): New class.
112         (tree_ssa_dominator_optimize): Adjust.
113         (dom_thread_across_edge): Convert to method
114         dom_opt_dom_walker::thread_across_edge.
115         (dom_opt_enter_block): Convert to member function
116         dom_opt_dom_walker::before_dom_children.
117         (dom_opt_leave_block): Convert to member function
118         dom_opt_dom_walker::after_dom_children.
119         * tree-ssa-dse.c (dse_dom_walker): New class.
120         (dse_enter_block): Convert to member function
121         dse_dom_walker::before_dom_children.
122         (tree_ssa_dse): Adjust.
123         * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
124         (determine_invariantness_stmt): Convert to method
125         invariantness_dom_walker::before_dom_children.
126         (determine_invariantness): Remove
127         (move_computations_dom_walker): New class.
128         (move_computations_stmt): Convert to method
129         move_computations_dom_walker::before_dom_children.
130         (move_computations, tree_ssa_lim): Adjust.
131         * tree-ssa-phiopt.c (nontrapping_dom_walker): new class
132         (nt_init_block): Make method
133         notrappping_dom_walker::before_dom_children.
134         (nt_fini_block): Make
135         method nontrapping_dom_walker::after_dom_children.
136         (get_non_trapping): Adjust.
137         * tree-ssa-pre.c (eliminate_dom_walker): New class.
138         (eliminate_bb): Make method eliminate_dom_walker::before_dom_children.
139         (eliminate_leave_block): Make method.
140         eliminate_dom_walker::after_dom_children.
141         (eliminate): Adjust
142         * tree-ssa-strlen.c (strlen_dom_walker): New class.
143         (strlen_enter_block): Make method
144         strlen_dom_walker::before_dom_children.
145         (strlen_leave_block): Make
146         method strlen_dom_walker::after_dom_children.
147         (tree_ssa_strlen): Adjust.
148         * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
149         (tree_ssa_uncprop): Adjust.
150         (uncprop_leave_block): Make method
151         uncprop_dom_walker::after_dom_children.
152         (uncprop_leave_block): Make method
153         uncprop_dom_walker::before_dom_children.
155 2013-09-18  Bin Cheng  <bin.cheng@arm.com>
157         * config/arm/arm.c (thumb1_reorg): Search for flag setting insn
158         before branch in same basic block.  Check both src and dest of
159         the move insn.
161 2013-09-17  Nick Clifton  <nickc@redhat.com>
163         * config/rl78/rl78-real.md (bf): New pattern.
164         (bt): New pattern.
165         * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
166         (rl78_print_operand): Do not put a # before a %B.
167         * config/rl78/rl78.opt: Tweak doc strings.
169 2013-09-17  DJ Delorie  <dj@redhat.com>
171         * config/rl78/constraints.md (Wcv): Allow up to $r31.
172         * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
173         (rl78_option_override): Likewise, if -mallregs.
174         (is_virtual_register): Likewise.
175         * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
176         (REGNO_OK_FOR_BASE_P): Likewise.
177         * config/rl78/rl78.opt (-mallregs): New.
179 2013-09-17  Nick Clifton  <nickc@redhat.com>
181         * config/rl78/rl78.c (need_to_save): Change return type to bool.
182         For interrupt functions: save all call clobbered registers if the
183         interrupt handler is not a leaf function.
184         (rl78_expand_prologue): Always recompute the frame information.
185         For interrupt functions: only select bank 0 if one of the bank 0
186         registers is going to be psuhed.
188 2013-09-17  DJ Delorie  <dj@redhat.com>
190         * config/rl78/constraints.md: For each W* constraint, rename to C*
191         and create a W* constraint that checks for an optional ES: prefix
192         pattern also.
193         * config/rl78/rl78.md (UNS_ES_ADDR): New.
194         (es_addr): New.  Used to wrap far addresses.
195         * config/rl78/rl78-protos.h (rl78_es_addr): New.
196         (rl78_es_base): New.
197         * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
198         wrapped far addresses.
199         (rl78_print_operand_1): Unwrap far addresses before processing.
200         (rl78_lo16): Wrap far addresses in unspecs.
201         (rl78_es_addr): New.
202         (rl78_es_base): New.
203         (insn_ok_now): Check for not-yet-wrapped far addresses.
204         (transcode_memory_rtx): Properly re-wrap far addresses.
206 2013-09-17  Sebastian Huber  <sebastian.huber@embedded-brains.de>
208         * config/sparc/t-rtems: Add leon3 multilibs.
210 2013-09-17  Cong Hou  <congh@google.com>
212         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
213         when checking the dot production pattern. The type of rhs operand
214         of multiply is now checked correctly.
216 2013-09-17  Jeff Law  <law@redhat.com>
218         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
219         edge implied equivalences into successor phis.
220         * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
221         here from tree-ssa-threadedge.c.
222         (mark_threaded_blocks): When threading through a joiner, if both
223         successors of the joiner's clone reach the same block, verify the
224         PHI arguments are equal.  If not, cancel the jump threading request.
225         * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
226         tree-ssa-threadupdate.c
227         (thread_across_edge): Don't check PHI argument equality when
228         threading through joiner block here.
229         
230 2013-09-17  Andrew MacLeod <amacleod@redhat.com>
232         * tree-flow.h (ssa_undefined_value_p): Remove prototype.
233         * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
234         (warn_uninit, warn_uninitialized_vars, execute_early_warn_uninitialized,
235         make_pass_early_warn_uninitialized): Move to tree-ssa-uninit.c.
236         * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c
237         (has_undefined_value_p): New.  Pass dependant parts of 
238         ssa_undefined_value_p.
239         (uninit_undefined_value_p): Use has_undefined_value_p.
240         (warn_uninit, warn_uninitialized_vars, execute_early_warn_uninitialized,
241         make_pass_early_warn_uninitialized): Move from tree-ssa.c
242         * tree-ssa.h: Adjust prototypes
244 2013-09-17  Jan Hubicka  <jh@suse.cz>
246         PR middle-end/58332
247         * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
248         * ipa-inline.c (can_inline_edge_p): Do not downgrade
249         FUNCTION_NOT_OPTIMIZED.
250         * ipa-inline-analysis.c (compute_inline_parameters): Function
251         not optimized is not inlinable unless it is alwaysinline.
252         (inline_analyze_function): Force calls in not optimized
253         function not inlinable.
255 2013-09-17  Jan Hubicka  <jh@suse.cz>
257         PR middle-end/58329
258         * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
259         to return NULL.
260         * ipa.c (function_and_variable_visibility): Likewise.
261         * ipa-profile.c (ipa_profile): Likewise.
263 2013-09-17  Bernd Edlinger  <bernd.edlinger@hotmail.de>
265         PR ipa/58398
266         * cgraph.c (cgraph_function_body_availability): Check for ifunc
267         attribute, and don't inline the resolver in this case.
269 2013-09-17  Teresa Johnson  <tejohnson@google.com>
271         * coverage.c (get_coverage_counts): Add missing newline.
273 2013-09-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
275         PR tree-optimization/58088
276         * fold-const.c (mask_with_trailing_zeros): New function.
277         (fold_binary_loc): Make sure we don't recurse infinitely
278         when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
279         Use mask_with_trailing_zeros where appropriate.
281 2013-09-17  Yuri Rumyantsev  <ysrumyan@gmail.com>
283         * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
284         of 'prev' var to get better distance estimation.
286 2013-09-17  Eric Botcazou  <ebotcazou@adacore.com>
288         * tree-inline.h (struct copy_body_data): Add transform_parameter.
289         * tree-inline.c (is_parameter_of): New predicate.
290         (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
291         a parameter has been remapped.
292         (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
293         (optimize_inline_calls): Initialize transform_parameter.
294         (copy_gimple_seq_and_replace_locals): Likewise.
295         (tree_function_versioning): Likewise.
296         (maybe_inline_call_in_expr): Likewise.
298 2013-09-17  Nick Clifton  <nickc@redhat.com>
300         * config/msp430/msp430-protos.h: Add prototypes for new functions.
301         * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
302         interrupt handlers.
303         (is_attr_func): New function.
304         (msp430_is_interrupt_func): New function.
305         (is_naked_func): New function.
306         (is_reentrant_func): New function.
307         (is_critical_func): New function.
308         (msp430_start_function): Add annotations for function attributes.
309         (msp430_attr): New function.
310         (msp430_attribute_table): New.
311         (msp430_function_section): New function.
312         (TARGET_ASM_FUNCTION_SECTION): Define.
313         (msp430_builtin): New enum.
314         (msp430_init_builtins): New function.
315         (msp430_builtin_devl): New function.
316         (msp430_expand_builtin): New function.
317         (TARGET_INIT_BUILTINS): Define.
318         (TARGET_EXPAND_BUILTINS): Define.
319         (TARGET_BUILTIN_DECL): Define.
320         (msp430_expand_prologue): Add support for naked, interrupt,
321         critical and reentrant functions.
322         (msp430_expand_epilogue): Likewise.
323         (msp430_print_operand): Handle 'O' character.
324         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
325         NO_TRAMPOLINES.
326         * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
327         UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
328         (pushm): Use a 'n' rather than an 'i' constraint.
329         (msp_return): Add generation of the interrupt return instruction.
330         (disable_interrupts): New pattern.
331         (enable_interrupts): New pattern.
332         (push_intr_state): New pattern.
333         (pop_intr_state): New pattern.
334         (bic_SR): New pattern.
335         (bis_SR): New pattern.
336         * doc/extend.texi: Document MSP430 function attributes and builtin
337         functions.
339 2013-09-17  Richard Biener  <rguenther@suse.de>
341         PR tree-optimization/58432
342         * tree-loop-distribution.c (tree_loop_distribution): Also
343         scan PHIs for outside loop uses and seed a partition from them.
345 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
347         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
348         (restructure_reference): Call backtrace_base_for_ref.
350 2013-09-17  Alan Modra  <amodra@gmail.com>
352         PR target/57589
353         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11 patch.
355 2013-09-16  DJ Delorie  <dj@redhat.com>
357         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
358         vregs location for RL78/G10.
359         (rl78_expand_prologue): Avoid SEL on G10.
360         (rl78_expand_epilogue): Likewise.
361         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
362         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
363         __RL78_G10__ when appropriate.
364         (ASM_SPEC): Pass -mg10 along to the assembler.
365         * config/rl78/rl78.md (sel_rb): Disable for G10.
366         * config/rl78/rl78.opt: Add -mg10 option.
367         * config/rl78/t-rl78: Add -mg10 multilib.
369 2013-09-16  Xinliang David Li  <davidxl@google.com>
371         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
372         * omp-low.c (omp_max_vf): Ditto.
373         (expand_omp_simd): Ditto.
374         * tree-vectorizer.c (vectorize_loops): Ditto.
375         (gate_vect_slp): Ditto.
376         (gate_increase_alignment): Ditto.
377         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
378         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
379         (gate_tree_vectorize): Name change.
380         (tree_vectorize): Ditto.
381         (pass_vectorize::gate): Call new function.
382         (pass_vectorize::execute): Ditto.
383         opts.c: O3 default setting change.
384         (finish_options): Check new flag.
385         * doc/invoke.texi: Document new flags.
386         * common.opt: New flags.
388 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
390         * doc/tm.texi.in (Cond Exec Macros): Remove node.
391         (Condition Code): Don't reference it.
392         * doc/tm.texi: Regenerate.
394 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
396         PR middle-end/58418
397         * lra-constraints.c (undo_optional_reloads): Consider all optional
398         reload even if it did not get a hard reg.
400 2013-09-16  Teresa Johnson  <tejohnson@google.com>
402         * dumpfile.c (dump_loc): Remove newline emission.
403         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
404         emission to dump_printf_loc calls where missing.
405         (vect_mark_for_runtime_alias_test): Ditto.
406         (vect_analyze_data_ref_dependence): Ditto.
407         (vect_analyze_data_ref_dependences): Ditto.
408         (vect_slp_analyze_data_ref_dependence): Ditto.
409         (vect_slp_analyze_data_ref_dependences): Ditto.
410         (vect_compute_data_ref_alignment): Ditto.
411         (vect_update_misalignment_for_peel): Ditto.
412         (vect_verify_datarefs_alignment): Ditto.
413         (vector_alignment_reachable_p): Ditto.
414         (vect_get_data_access_cost): Ditto.
415         (vect_enhance_data_refs_alignment): Ditto.
416         (vect_find_same_alignment_drs): Ditto.
417         (vect_analyze_data_refs_alignment): Ditto.
418         (vect_analyze_group_access): Ditto.
419         (vect_analyze_data_ref_access): Ditto.
420         (vect_analyze_data_ref_accesses): Ditto.
421         (vect_prune_runtime_alias_test_list): Ditto.
422         (vect_analyze_data_refs): Ditto.
423         (vect_create_addr_base_for_vector_ref): Ditto.
424         (vect_create_data_ref_ptr): Ditto.
425         (vect_grouped_store_supported): Ditto.
426         (vect_grouped_load_supported): Ditto.
427         * value-prof.c (check_counter): Ditto.
428         (check_ic_target): Ditto.
429         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
430         (vect_recog_widen_mult_pattern): Ditto.
431         (vect_recog_widen_sum_pattern): Ditto.
432         (vect_recog_over_widening_pattern): Ditto.
433         (vect_recog_widen_shift_pattern): Ditto.
434         (vect_recog_rotate_pattern): Ditto.
435         (vect_recog_vector_vector_shift_pattern): Ditto.
436         (vect_recog_divmod_pattern): Ditto.
437         (vect_recog_mixed_size_cond_pattern): Ditto.
438         (vect_recog_bool_pattern): Ditto.
439         (vect_pattern_recog_1): Ditto.
440         (vect_pattern_recog): Ditto.
441         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
442         (vect_is_simple_iv_evolution): Ditto.
443         (vect_analyze_scalar_cycles_1): Ditto.
444         (vect_get_loop_niters): Ditto.
445         (vect_analyze_loop_1): Ditto.
446         (vect_analyze_loop_form): Ditto.
447         (vect_analyze_loop_operations): Ditto.
448         (vect_analyze_loop_2): Ditto.
449         (vect_analyze_loop): Ditto.
450         (report_vect_op): Ditto.
451         (vect_is_slp_reduction): Ditto.
452         (vect_is_simple_reduction_1): Ditto.
453         (vect_get_known_peeling_cost): Ditto.
454         (vect_estimate_min_profitable_iters): Ditto.
455         (vect_model_reduction_cost): Ditto.
456         (vect_model_induction_cost): Ditto.
457         (get_initial_def_for_induction): Ditto.
458         (vect_create_epilog_for_reduction): Ditto.
459         (vectorizable_reduction): Ditto.
460         (vectorizable_induction): Ditto.
461         (vectorizable_live_operation): Ditto.
462         (vect_loop_kill_debug_uses): Ditto.
463         (vect_transform_loop): Ditto.
464         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
465         (vect_stmt_relevant_p): Ditto.
466         (process_use): Ditto.
467         (vect_mark_stmts_to_be_vectorized): Ditto.
468         (vect_model_simple_cost): Ditto.
469         (vect_model_promotion_demotion_cost): Ditto.
470         (vect_model_store_cost): Ditto.
471         (vect_get_store_cost): Ditto.
472         (vect_model_load_cost): Ditto.
473         (vect_get_load_cost): Ditto.
474         (vect_init_vector_1): Ditto.
475         (vect_get_vec_def_for_operand): Ditto.
476         (vect_finish_stmt_generation): Ditto.
477         (vectorizable_call): Ditto.
478         (vectorizable_conversion): Ditto.
479         (vectorizable_assignment): Ditto.
480         (vectorizable_shift): Ditto.
481         (vectorizable_operation): Ditto.
482         (vectorizable_store): Ditto.
483         (vectorizable_load): Ditto.
484         (vectorizable_condition): Ditto.
485         (vect_analyze_stmt): Ditto.
486         (vect_transform_stmt): Ditto.
487         (vect_is_simple_use): Ditto.
488         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
489         (vect_can_advance_ivs_p): Ditto.
490         (vect_update_ivs_after_vectorizer): Ditto.
491         (vect_do_peeling_for_loop_bound): Ditto.
492         (vect_gen_niters_for_prolog_loop): Ditto.
493         (vect_update_inits_of_drs): Ditto.
494         (vect_create_cond_for_alias_checks): Ditto.
495         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
496         (vect_build_slp_tree_1): Ditto.
497         (vect_supported_load_permutation_p): Ditto.
498         (vect_analyze_slp_instance): Ditto.
499         (vect_analyze_slp): Ditto.
500         (vect_make_slp_decision): Ditto.
501         (vect_detect_hybrid_slp): Ditto.
502         (vect_bb_vectorization_profitable_p): Ditto.
503         (vect_slp_analyze_bb_1): Ditto.
504         (vect_update_slp_costs_according_to_vf): Ditto.
505         (vect_get_mask_element): Ditto.
506         (vect_transform_slp_perm_load): Ditto.
507         (vect_schedule_slp_instance): Ditto.
508         (vect_schedule_slp): Ditto.
509         (vect_slp_transform_bb): Ditto.
510         * profile.c (read_profile_edge_counts): Ditto.
511         (compute_branch_probabilities): Ditto.
512         * coverage.c (get_coverage_counts): Ditto.
514 2013-09-16  Diego Novillo  <dnovillo@google.com>
516         * tree-core.h: Add missing comment lines from refactoring
517         of tree.h.
519 2013-09-16  Jan Hubicka  <jh@suse.cz>
521         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
522         abstract functions; for static functions check the presence
523         of body.
525 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
527         * config/aarch64/aarch64-simd-builtins.def (fma): New.
528         * config/aarch64/aarch64-simd.md
529         (aarch64_mla_elt<mode>): New.
530         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
531         (aarch64_mls_elt<mode>): Likewise.
532         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
533         (aarch64_fma4_elt<mode>): Likewise.
534         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
535         (aarch64_fma4_elt_to_128v2df): Likewise.
536         (aarch64_fma4_elt_to_64df): Likewise.
537         (fnma<mode>4): Likewise.
538         (aarch64_fnma4_elt<mode>): Likewise.
539         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
540         (aarch64_fnma4_elt_to_128v2df): Likewise.
541         (aarch64_fnma4_elt_to_64df): Likewise.
542         * config/aarch64/iterators.md (VDQSF): New.
543         * config/aarch64/arm_neon.h
544         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
545         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
547 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
549         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
550         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
551         (aarch64_mul3_elt_to_128df): Likewise.
552         (aarch64_mul3_elt_to_64v2df): Likewise.
553         * config/aarch64/iterators.md (VEL): Also handle DFmode.
554         (VMUL): New.
555         (VMUL_CHANGE_NLANES) Likewise.
556         (h_con): Likewise.
557         (f): Likewise.
558         * config/aarch64/arm_neon.h
559         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
561 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
563         * config/aarch64/arm_neon.h
564         (vcvtx_high_f32_f64): Fix parameters.
566 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
567             Uros Bizjak  <ubizjak@gmail.com>
569         * config/alpha.c: Include tree-ssanames.h.
571 2013-09-16  Richard Biener  <rguenther@suse.de>
573         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
574         (dot_rdg_1): Handle control_dd.
575         (create_edge_for_control_dependence): New function.
576         (create_rdg_edges): Add control dependences if asked for.
577         (build_rdg): Likewise.
578         (generate_loops_for_partition): If there are not necessary
579         control stmts remove all their dependencies.
580         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
581         (distribute_loop): Pass on control dependences.
582         (tree_loop_distribution): Compute control dependences and remove
583         restriction on number of loop nodes.
585 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
587         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
588         for internal calls.
590 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
592         * cse.c (try_const_anchors): Punt on CC modes.
594 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
596         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
598 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
600         PR target/58382
601         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
602         calls to word_mode.
604 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
606         PR target/48094
607         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
608         seen.
609         (darwin_objc1_section): Likewise.
610         (darwin_file_end): Emit Image Info section when required.
612 2013-09-14  Jan Hubicka  <jh@suse.cz>
614         * tree-into-ssa.c (gate_into_ssa): New.
615         (pass_data_build_ssa): Use it.
616         * cgraph.h (expand_thunk): Update prototype.
617         * cgraphunit.c (analyze_function): Expand thunks early.
618         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
619         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
620         set lowered flag; do not add new function.
621         (assemble_thunks_and_aliases): Update.
622         * tree-ssa.c (gate_init_datastructures): New gate.
623         (pass_data_init_datastructures): Use it.
625 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
627         PR target/58269
628         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
629         xmm register set described in the psABI.
631 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
633         * dwarf2out.c (should_emit_struct_debug): Add check
634         for type_decl variable is not NULL.
636 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
638         * config.gcc: Use new winnt-c.c target hooks
639         * config/t-winnt: New file
640         * config/winnt-c.c: New file
641         * doc/tm.texi.in: Document new hook
642         * doc/tm.texi: Regenerated
644 2013-09-13  Jan Hubicka  <jh@suse.cz>
646         PR middle-end/58094
647         * ipa-inline.c (check_callers): New function.
648         (check_caller_edge): Remove.
649         (want_inline_function_to_all_callers_p): Also permit alises that are
650         called dirrectly.
651         (inline_to_all_callers): Terminate the walk when devirtualization
652         introduce new calls.
654 2013-09-13  Jan Hubicka  <jh@suse.cz>
656         * ipa-inline-analysis.c (struct growth_data): Add node.
657         (do_estimate_growth_1): Fix detection of recursion.
659 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
661         PR tree-optimization/58392
662         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
663         to avoid shadowing of outer loop variable.  If
664         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
665         replace_by_duplicate_decl simduid of loops that have it set and
666         set dest_cfun->has_simduid_loops and/or
667         dest_cfun->has_force_vect_loops.
668         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
669         instead of maybe_lookup_decl.
670         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
671         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
672         call.  Copy over force_vect and copy and remap simduid.  Set
673         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
674         (copy_cfg_body): Remove blocks_to_copy argument.  Use
675         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
676         caller.  Don't set cfun->has_simduid_loops and/or
677         cfun->has_force_vect_loops here.
678         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
679         caller.
680         (expand_call_inline, tree_function_versioning): Adjust copy_body
681         callers.
683 2013-09-13  Martin Jambor  <mjambor@suse.cz>
685         PR bootstrap/58388
686         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
687         the assert if the edge was a speculative one.
689 2013-09-13  Richard Biener  <rguenther@suse.de>
691         * tree-data-ref.h (known_dependences_p): Move here ...
692         * tree-loop-distribution.c (known_dependences_p): ... from here.
693         (dump_rdg_component, debug_rdg_component): Remove.
694         (dump_rdg): Adjust.
695         (generate_loops_for_partition): Use gimple_uid instead of
696         relying on matching stmt visit order.
697         (rdg_build_partitions): Take starting stmt vector.
698         (ldist_gen): Merge into ...
699         (distribute_loop): ... this function.  Do not compute starting
700         vertices vector.
701         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
703 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
705         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
706         Set type attribute correctly. Set predicable_short_it attribute.
707         (cmpsi_shiftsi): Remove %? from output template.
709 2013-09-13  Richard Biener  <rguenther@suse.de>
711         * tree-loop-distribution.c (struct rdg_component,
712         rdg_defs_used_in_other_loops_p, free_rdg_components,
713         rdg_build_components): Remove.
714         (stmts_from_loop): Do not record virtual PHIs.
715         (generate_loops_for_partition): Skip virtual PHIs.
716         (build_rdg_partition_for_component): Rename to ...
717         (build_rdg_partition_for_vertex): ... this and adjust.
718         (rdg_build_partitions): Take a vector of starting vertices
719         instead of components.  Remove unnecessary leftover handling.
720         (ldist_gen): Do not build components or record other stores.
721         (distribute_loop): Do not distribute loops containing stmts
722         with side-effects.
724 2013-09-13  Christian Bruel  <christian.bruel@st.com>
726         PR target/58314
727         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
729 2013-09-13  Kai Tietz  <ktietz@redhat.com>
731         * config.gcc: Separate cases for mingw and cygwin targets,
732         and add 64-bit cygwin target case.
734         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
735         dll-export inline-functions.
736         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
739 2013-09-13  Jeff Law  <law@redhat.com>
741         PR middle-end/58387
742         Revert:
743         2013-09-06  Jeff Law  <law@redhat.com>
745         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
746         edge implied equivalences into successor phis.
748 2013-09-12  DJ Delorie  <dj@redhat.com>
750         * config/rl78/rl78-virt.md: Change from | to \; for asm line
751         separators.
753 2013-09-12  Brooks Moses  <bmoses@google.com>
755         PR driver/42955
756         * Makefile.in: Do not install driver binaries in $(target)/bin.
758 2013-09-12  DJ Delorie  <dj@redhat.com>
760         * config/rl78/rl78.opt (mrelax): New.
761         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
762         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
764         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
765         between SP and FP.
766         (rl78_expand_epilogue): Likewise.
768 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
770         PR middle-end/58335
771         * lra-eliminations.c (remove_reg_equal_offset_note): New.
772         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
773         pointer elimination with using remove_reg_equal_offset_note.
775 2013-09-12  DJ Delorie  <dj@redhat.com>
777         * config/msp430/: New port.
778         * config.gcc (msp430): Added.
779         * doc/invoke.texi: Document MSP430 options.
780         * doc/install.texi: Document msp430-elf
781         * doc/md.texi: Document msp430-elf
782         * doc/contrib.texi: Document msp430-elf
784         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
785         PSImode.
787 2013-09-12  Martin Jambor  <mjambor@suse.cz>
789         PR ipa/58389
790         * ipa-prop.c (remove_described_reference): Give up if the edge in the
791         reference descriptor is NULL.
792         (ipa_edge_removal_hook): If owning a reference descriptor, set its
793         edge to NULL.
795 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
797         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
798         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
799         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
800         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
801         make_temp_ssa_name): move to tree-ssanames.h
802         * tree-ssa-alias.h: Move prototype.
803         * tree-ssa.h: Include tree-ssanames.h.
804         * tree-ssanames.c (FREE_SSANAMES): Move to here.
805         * tree-ssanames.h: New.  Move items from tree-flow*.h
806         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
808 2013-09-12  Richard Biener  <rguenther@suse.de>
810         PR tree-optimization/58404
811         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
812         propagate non-invariant addresses into dereferences wrapped
813         in component references.
815 2013-09-12  Richard Biener  <rguenther@suse.de>
817         PR tree-optimization/58402
818         * passes.def: Move pass_late_warn_uninitialized later.
820 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
822         * tree-ssa.h: New.  Move content from tree-flow.h and
823         tree-flow-inline.h.
824         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
825         Move prototypes belonging to tree-ssa.c.
826         * tree-flow-inline.h (redirect_edge_var_map_def,
827         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
828         tree-ssa.h.
829         * gimple.h: Adjust prototypes.
830         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
831         to...
832         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
833         * tree.h: Move prototype to tree-ssa.h.
834         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
835         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
836         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
837         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
838         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
839         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
840         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
841         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
842         graphite-blocking.c, graphite-clast-to-gimple.c,
843         graphite-dependences.c, graphite-interchange.c,
844         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
845         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
846         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
847         ipa-reference.c, ipa-split.c, ipa-utils.c,
848         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
849         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
850         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
851         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
852         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
853         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
854         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
855         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
856         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
857         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
858         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
859         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
860         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
861         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
862         value-prof.c, var-tracking.c,
863         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
865 2013-09-12  Richard Biener  <rguenther@suse.de>
867         PR tree-optimization/58396
868         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
869         (build_rdg): Take a loop-nest parameter, fix memleaks.
870         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
872 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
874         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
875         for SLM.
877 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
879         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
880         The FMA instruction names should have a 'v' prefix.
882 2013-09-12  Richard Biener  <rguenther@suse.de>
884         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
885         (dot_rdg): Use popen instead of system in optional code.
886         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
887         (already_processed_vertex_p): Adjust.
888         (has_anti_or_output_dependence, predecessor_has_mem_write,
889         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
890         rdg_flag_uses): Remove.
891         (rdg_flag_vertex): Simplify.
892         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
893         remove recursion.
894         (build_rdg_partition_for_component): Process the first vertex
895         of a component only.
896         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
898 2013-09-12  Alan Modra  <amodra@gmail.com>
900         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
902 2013-09-11  DJ Delorie  <dj@redhat.com>
903             Nick Clifton  <nickc@redhat.com>
905         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
906         (rl78_stack_based_mem): New.
907         * config/rl78/constraints.md (Iv08): New.
908         (Iv16): New.
909         (Iv24): New.
910         (Is09): New.
911         (Is17): New.
912         (Is25): New.
913         (ISsi): New.
914         (IShi): New.
915         (ISqi): New.
916         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
917         (movhi): Likewise.
918         (movsi): Change from expand to insn-and-split.
919         (ashrsi3): Clobber AX.
920         (lshrsi3): New.
921         (ashlsi3): New.
922         (cbranchsi4): New.
923         * config/rl78/rl78.md (CC_REG): Fix.
924         (addsi3): Allow memory and immediate operands.
925         (addsi3_internal): Split into...
926         (addsi3_internal_virt): ...new, and ...
927         (addsi3_internal_real): ...new.
928         (subsi): New.
929         (subsi3_internal_virt): New.
930         (subsi3_internal_real): New.
931         (mulsi3): Add memory operand.
932         (mulsi3_rl78): Likewise.
933         (mulsi3_g13): Likewise.
934         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
935         (cbranchqi4_real): Add more constraint options.
936         (cbranchhi4_real): Expand pattern.
937         (cbranchhi4_real_signed): New.
938         (cbranchhi4_real_inverted): New.
939         (cbranchsi4_real_lt): New.
940         (cbranchsi4_real_ge): New.
941         (cbranchsi4_real_signed): New.
942         (cbranchsi4_real): New.
943         (peephole2): New.
944         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
945         constant shifts.
946         (lshrsi3_virt): Likewise.
947         (ashlsi3_virt): Likewise.
948         (cbranchqi4_virt_signed): New.
949         (cbranchhi4_virt_signed): New.
950         (cbranchsi4_virt): New.
951         * config/rl78/rl78.c: Whitespace fixes throughout.
952         (move_elim_pass): New.
953         (pass_data_rl78_move_elim): New.
954         (pass_rl78_move_elim): New.
955         (make_pass_rl78_move_elim): New.
956         (rl78_devirt_info): Run devirt earlier.
957         (rl78_move_elim_info): New.
958         (rl78_asm_file_start): Register it.
959         (rl78_split_movsi): New.
960         (rl78_as_legitimate_address): Allow virtual base registers when
961         appropriate.
962         (rl78_addr_space_convert): Remove spurious debug stuff.
963         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
964         (rl78_print_operand): More cases for not printing '#'.
965         (rl78_expand_compare): Remove most of the logic.
966         (content_memory): New.
967         (clear_content_memory): New.
968         (get_content_index): New.
969         (get_content_name): New.
970         (display_content_memory): New.
971         (update_content): New.
972         (record_content): New.
973         (already_contains): New.
974         (insn_ok_now): Re-recog insns with virtual registers.
975         (add_postponed_content_update): New.
976         (process_postponed_content_update): New.
977         (gen_and_emit_move): New.
978         (transcode_memory_rtx): Record new location content.
979         Use gen_and_emit_move.
980         (force_into_acc): New.
981         (move_to_acc): Use gen_and_emit_move.
982         (move_from_acc): Likewise.
983         (move_acc_to_reg): Likewise.
984         (move_to_x): Likewise.
985         (move_to_hl): Likewise.
986         (move_to_de): Likewise.
987         (rl78_alloc_physical_registers_op1): Record location content.
988         (has_constraint): New.
989         (rl78_alloc_physical_registers_op2): Record location content.
990         Optimize use of HL.
991         (rl78_alloc_physical_registers_ro1): Likewise.
992         (rl78_alloc_physical_registers_cmp): Likewise.
993         (rl78_alloc_physical_registers_umul): Likewise.
994         (rl78_alloc_address_registers_macax): New.
995         (rl78_alloc_physical_registers): Initialize and set location
996         content memory as needed.
997         (rl78_reorg): Make sure split2 is called.
998         (rl78_rtx_costs): New.
1000 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
1002         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
1003         for (not (neg ...)) and (neg (not ...)) cases.
1005 2013-09-11  Richard Biener  <rguenther@suse.de>
1007         PR middle-end/58377
1008         * passes.def: Split critical edges before late uninit warning passes.
1009         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
1011 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
1013         PR tree-optimization/58385
1014         * fold-const.c (build_range_check): If both low and high are NULL,
1015         use omit_one_operand_loc to preserve exp side-effects.
1017 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1019         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
1021 2013-09-11  Richard Biener  <rguenther@suse.de>
1023         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
1024         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
1025         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
1026         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
1027         stmts_from_loop, known_dependences_p, build_empty_rdg,
1028         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
1029         * tree-loop-distribution.c: ... here.
1030         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
1031         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
1032         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
1033         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
1034         * tree-loop-distribution.c: ... here.
1035         * tree-loop-distribution.c: Include gimple-pretty-print.h.
1036         (struct partition_s): Add loops member.
1037         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
1038         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
1039         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
1041 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1042             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1043             Sergey Lega  <sergey.s.lega@intel.com>
1044             Anna Tikhonova  <anna.tikhonova@intel.com>
1045             Ilya Tocar  <ilya.tocar@intel.com>
1046             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1047             Ilya Verbin  <ilya.verbin@intel.com>
1048             Kirill Yukhin  <kirill.yukhin@intel.com>
1049             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1051         * config/i386/constraints.md (k): New.
1052         (Yk): Ditto.
1053         * config/i386/i386.c (const regclass_map): Add new mask registers.
1054         (dbx_register_map): Ditto.
1055         (dbx64_register_map): Ditto.
1056         (svr4_dbx_register_map): Ditto.
1057         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
1058         disabled.
1059         (ix86_preferred_reload_class): Disable constants for mask registers.
1060         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
1061         (ix86_hard_regno_mode_ok): Support new mask registers.
1062         (x86_order_regs_for_local_alloc): Ditto.
1063         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
1064         (FIXED_REGISTERS): Add new mask registers.
1065         (CALL_USED_REGISTERS): Ditto.
1066         (REG_ALLOC_ORDER): Ditto.
1067         (VALID_MASK_REG_MODE): New.
1068         (FIRST_MASK_REG): Ditto.
1069         (LAST_MASK_REG): Ditto.
1070         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
1071         (MAYBE_MASK_CLASS_P): New.
1072         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
1073         (REG_CLASS_CONTENTS): Ditto.
1074         (MASK_REGNO_P): New.
1075         (ANY_MASK_REG_P): Ditto.
1076         (HI_REGISTER_NAMES): Add new mask registers.
1077         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
1078         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
1079         mask registers.
1080         (attribute "type"): Add mskmov, msklog.
1081         (attribute "length_immediate"): Support them.
1082         (attribute "memory"): Ditto.
1083         (attribute "prefix_0f"): Ditto.
1084         (*movhi_internal): Support new mask registers.
1085         (*movqi_internal): Ditto.
1086         (define_split): Split out clobber pattern is a logic
1087         insn on mask registers.
1088         (*k<logic><mode>): New.
1089         (*andhi_1): Extend to support mask regs.
1090         (*andqi_1): Extend to support mask regs.
1091         (kandn<mode>): New.
1092         (define_split): Split and-not to and and not if operands
1093         are not mask regs.
1094         (*<code><mode>_1): Separate HI mode to new pattern...
1095         (*<code>hi_1): This.
1096         (*<code>qi_1): Extend to support mask regs.
1097         (kxnor<mode>): New.
1098         (kortestzhi): Ditto.
1099         (kortestchi): Ditto.
1100         (kunpckhi): Ditto.
1101         (*one_cmpl<mode>2_1): Remove HImode and handle it...
1102         (*one_cmplhi2_1): ...Here, now with mask registers support.
1103         (*one_cmplqi2_1): Support new mask registers.
1104         (HI/QImode arithmetics splitter): Don't split if mask registers
1105         are used.
1106         (HI/QImode not splitter): Ditto.
1107         * config/i386/predicated.md (mask_reg_operand): New.
1108         (general_reg_operand): Ditto.
1110 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1112         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
1113         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
1115 2013-09-10  Jeff Law  <law@redhat.com>
1117         PR tree-optimization/58380
1118         * tree-ssa-threadupdate.c (thread_block): Recognize another case
1119         of threading through a buried loop header.
1121         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
1122         return value for single successor case.
1124 2013-09-10  Jan Hubicka  <jh@suse.cz>
1126         * ipa-devirt.c (ipa_devirt): Enable with LTO.
1128 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
1130         PR target/58361
1131         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
1132         support conditional execution.
1133         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
1135 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
1137         * lra.c (lra): Clear lra_optional_reload_pseudos before every
1138         constraint pass.
1139         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
1140         Check destination too to check move insn.
1141         (undo_optional_reloads): Add check that the original peudo did not
1142         changed its allocation and the optional reload was inherited on last
1143         inheritance pass.  Break loop after deciding to keep optional reload.
1144         (lra_undo_inheritance): Add check that inherited pseudo still in
1145         memory.
1147 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
1149         * config/aarch64/aarch64.md (generic_sched): New.
1150         * config/aarch64/aarch64-generic.md (load): Make conditional
1151         on generic_sched attribute.
1152         (nonload): Likewise.
1154 2013-09-10  Jan Hubicka  <jh@suse.cz>
1156         * lto-cgraph.c: Include ipa-utils.h.
1157         (compute_ltrans_boundary): Also add possible targets into the boundary.
1159 2013-09-10  Jan Hubicka  <jh@suse.cz>
1161         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
1162         VAR_DECL of vtable rather than full expression.
1164 2013-09-10  Jan Hubicka  <jh@suse.cz>
1165             Paolo Carlini  <paolo.carlini@oracle.com>
1167         * cgraphunit.c (analyze_functions): Save input_location, set it
1168         to UNKNOWN_LOCATION and restore it at the end.
1170 2013-09-10  Martin Jambor  <mjambor@suse.cz>
1172         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
1173         represented by a thunk.
1175 2013-09-10  Jeff Law  <law@redhat.com>
1177         PR tree-optimization/58343
1178         * tree-ssa-threadupdate.c (thread_block): Identify and disable
1179         jump threading requests through loop headers buried in the middle
1180         of a jump threading path.
1182         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
1183         in return value/type.
1185 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
1187         PR rtl-optimization/58365
1188         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
1189         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
1190         it differs.
1192 2013-09-10  Richard Biener  <rguenther@suse.de>
1194         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
1195         * tree-data-ref.c (create_rdg_vertices): Collect all data
1196         references, signal failure to the caller, use data-ref API.
1197         (build_rdg): Compute data references only once.  Maintain lifetime
1198         of data references and data dependences from within RDG.
1199         (free_rdg): Free dependence relations.
1200         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
1201         inventing extra dependences.
1202         (distribute_loop): Update for RDG API changes.
1204 2013-09-10  Kai Tietz  <ktietz@redhat.com>
1206         * doc/invoke.texi (fms-extensions): Document changed
1207         behavior for ms-abi targets.
1208         * config/i386/i386.c (ix86_option_override_internal):
1209         Set default value of option -fms-extension for ms-abi targets.
1211 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
1213         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
1215 2013-09-10  Alan Modra  <amodra@gmail.com>
1217         PR target/58330
1218         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
1220 2013-09-10  Alan Modra  <amodra@gmail.com>
1222         * config/rs6000/predicates.md (add_cint_operand): New.
1223         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
1224         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
1225         using add_cint_operand.
1226         (largetoc_high_plus_aix): Likewise.
1228 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
1230         PR tree-optimization/58364
1231         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
1232         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
1233         the current range can't be an unconditional true or false.
1235 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
1237         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
1239 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
1241         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
1243 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
1245         PR c++/43452
1246         * doc/invoke.texi (-Wdelete-incomplete): Document it.
1248 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
1250         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
1251         NO_REGS for immediate that can't be moved directly into FP_REGS.
1253 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1255         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
1256         comparison with negated operand.
1257         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
1258         RTL form.
1260 2013-09-09  Richard Biener  <rguenther@suse.de>
1262         PR middle-end/58326
1263         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
1264         of a subloop record all its block as affecting loop-closed SSA form.
1266 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
1268         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
1269         of an rtx/bitpos pair.
1270         (store_fixed_bit_field): Update accordingly.
1272 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
1274         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
1275         GEN_INT.
1276         * builtins.c (expand_errno_check): Likewise.
1277         * dwarf2cfi.c (init_return_column_size): Likewise.
1278         * except.c (sjlj_mark_call_sites): Likewise.
1279         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
1280         * lra-constraints.c (emit_inc): Likewise.
1281         * ree.c (combine_set_extension): Likewise.
1282         * regmove.c (fixup_match_2): Likewise.
1283         * reload1.c (inc_for_reload): Likewise.
1285 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
1287         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
1288         (force_to_mode, simplify_shift_const_1, simplify_comparison):
1289         Use gen_int_mode with the mode of the associated simplify_* call.
1290         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
1291         * expmed.c (expand_shift_1): Likewise.
1292         * function.c (instantiate_virtual_regs_in_insn): Likewise.
1293         * loop-iv.c (iv_number_of_iterations): Likewise.
1294         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
1295         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
1297 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
1299         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
1300         of the associated expand_* call.
1301         (asan_emit_stack_protection): Likewise.
1302         * builtins.c (round_trampoline_addr): Likewise.
1303         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
1304         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
1305         (emit_store_flag): Likewise.
1306         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
1307         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
1308         Likewise.
1309         * function.c (instantiate_virtual_regs_in_insn): Likewise.
1310         * ifcvt.c (noce_try_store_flag_constants): Likewise.
1311         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
1312         * modulo-sched.c (generate_prolog_epilog): Likewise.
1313         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
1314         (expand_ctz, expand_ffs, expand_unop): Likewise.
1316 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
1318         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
1319         of the associated gen_rtx_* call.
1320         * caller-save.c (init_caller_save): Likewise.
1321         * combine.c (find_split_point, make_extraction): Likewise.
1322         (make_compound_operation): Likewise.
1323         * dwarf2out.c (mem_loc_descriptor): Likewise.
1324         * explow.c (plus_constant, probe_stack_range): Likewise.
1325         * expmed.c (expand_mult_const): Likewise.
1326         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
1327         * reload1.c (init_reload): Likewise.
1328         * valtrack.c (cleanup_auto_inc_dec): Likewise.
1329         * var-tracking.c (adjust_mems): Likewise.
1330         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
1331         rather than gen_rtx_fmt_ee.
1333 2013-09-09  Jan Hubicka  <jh@suse.cz>
1335         PR middle-end/58294
1336         * value-prof.c (gimple_ic): Copy also abnormal edges.
1338 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
1340         * asan.c (asan_shadow_cst): Use gen_int_mode.
1342 2013-09-08  Jan Hubicka  <jh@suse.cz>
1344         * ipa-profile.c: Add toplevel comment.
1345         (ipa_propagate_frequency_1): Be more conservative when profile is read.
1346         (contains_hot_call_p): New function.
1347         (ipa_propagate_frequency): Set frequencies based on counts when
1348         profile is read.
1349         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
1350         profile; do not tamper with profile after inlining if it is read.
1352 2013-09-08  Jan Hubicka  <jh@suse.cz>
1354         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
1355         speculative edges.
1357 2013-09-08  Jan Hubicka  <jh@suse.cz>
1359         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
1360         summary generation.
1362 2013-09-08  Jeff Law  <law@redhat.com>
1364         PR bootstrap/58340
1365         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
1366         of 'found'.
1368 2013-09-08  Andi Kleen  <ak@linux.intel.com>
1370         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
1372 2013-09-08  Jan Hubicka  <jh@suse.cz>
1374         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
1376 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
1378         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
1379         for non-debug insns.
1380         * lra.c (new_insn_reg): Take the containing insn as a parameter.
1381         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
1382         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
1383         accordingly.
1385 2013-09-08  Jan Hubicka  <jh@suse.cz>
1387         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
1388         targets and devirtualize to BUILT_IN_UNREACHABLE.
1389         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
1390         * ipa.c (walk_polymorphic_call_targets): New function.
1391         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
1392         functions; use the new timevar.
1393         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
1394         was removed from the program.
1395         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
1396         not consider it in the walk when its vtable is dead.
1397         (possible_polymorphic_call_targets_1): Pass anonymous flag to
1398         record_binfo.
1399         (devirt_variable_node_removal_hook): New function.
1400         (possible_polymorphic_call_targets): Also register
1401         devirt_variable_node_removal_hook.
1402         (ipa_devirt): Do not do non-speculative devirtualization.
1403         (gate_ipa_devirt): One execute if devirtualizing speculatively.
1405 2013-09-08  Jan Hubicka  <jh@suse.cz>
1407         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
1408         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
1409         varpool_remove_variable_insertion_hook): Declare.
1410         * varpool.c (varpool_node_hook_list): New structure.
1411         (first_varpool_node_removal_hook,
1412         first_varpool_variable_insertion_hook): New variables.
1413         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
1414         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
1415         varpool_remove_variable_insertion_hook,
1416         varpool_call_variable_insertion_hooks): New functions.
1417         (varpool_remove_node): Use it.
1419 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
1421         PR c++/54941
1422         * diagnostic.c (diagnostic_build_prefix): When s.file is
1423         "<built-in>" don't output line and column numbers.
1425 2013-09-06  Jan Hubicka  <jh@suse.cz>
1427         * cgraphunit.c (expand_thunk): Get body before touching arguments.
1428         * lto-streamer-out.c: Stream thunks, too.
1429         * lto-streamer-in.c (input_function): Pop cfun here
1430         (lto_read_body): Instead of here.
1432 2013-09-06  Caroline Tice  <cmtice@google.com>
1434         * doc/install.texi: Add documentation for the --enable-vtable-verify
1435         and the --disable-libvtv configure options.
1437 2013-09-06  Jeff Law  <law@redhat.com>
1439         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
1440         edge implied equivalences into successor phis.
1442 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
1444         * resource.c (mark_referenced_resources): Handle COND_EXEC.
1446 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
1448         * resource.c (mark_target_live_regs): Compute resources taking
1449         into account if a call is predicated or not.
1451 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
1453         * toplev.c (output_stack_usage): Be prepared for suffixes created by
1454         the compiler in the function names.
1456 2013-09-06  Jan Hubicka  <jh@suse.cz>
1458         PR middle-end/58094
1459         * ipa-inline.c (has_caller_p): New function.
1460         (want_inline_function_to_all_callers_p): Use it.
1461         (sum_callers, inline_to_all_callers): Break out from ...
1462         (ipa_inline): ... here.
1464 2013-09-06  Jan Hubicka  <jh@suse.cz>
1466         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
1467         only when AVX is enabled.
1469 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1471         * config/aarch64/aarch64.md
1472         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
1473         is fpsimd_<load/store>2.
1474         (load_pair<mode>): Likewise.
1475         (store_pair<mode>): Likewise.
1477 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1479         * config/arm/types.md (type): Add "mrs" type.
1480         * config/aarch64/aarch64.md
1481         (aarch64_load_tp_hard): Make type "mrs".
1482         * config/arm/arm.md
1483         (load_tp_hard): Make type "mrs".
1484         * config/arm/cortex-a15.md: Update with new attributes.
1485         * config/arm/cortex-a5.md: Update with new attributes.
1486         * config/arm/cortex-a53.md: Update with new attributes.
1487         * config/arm/cortex-a7.md: Update with new attributes.
1488         * config/arm/cortex-a8.md: Update with new attributes.
1489         * config/arm/cortex-a9.md: Update with new attributes.
1490         * config/arm/cortex-m4.md: Update with new attributes.
1491         * config/arm/cortex-r4.md: Update with new attributes.
1492         * config/arm/fa526.md: Update with new attributes.
1493         * config/arm/fa606te.md: Update with new attributes.
1494         * config/arm/fa626te.md: Update with new attributes.
1495         * config/arm/fa726te.md: Update with new attributes.
1497 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1499         * config/aarch64/aarch64.md
1500         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
1501         (*movtf_aarch64): Likewise.
1502         * config/arm/arm.md
1503         (thumb1_movdi_insn): Use "multiple" for type where more than one
1504         instruction is used for a move.
1505         (*arm32_movhf): Likewise.
1506         (*thumb_movdf_insn): Likewise.
1508 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1510         * config/arm/types.md (type): Rename fcpys to fmov.
1511         * config/arm/vfp.md
1512         (*arm_movsi_vfp): Rename type fcpys as fmov.
1513         (*thumb2_movsi_vfp): Likewise
1514         (*movhf_vfp_neon): Likewise
1515         (*movhf_vfp): Likewise
1516         (*movsf_vfp): Likewise
1517         (*thumb2_movsf_vfp): Likewise
1518         (*movsfcc_vfp): Likewise
1519         (*thumb2_movsfcc_vfp): Likewise
1520         * config/aarch64/aarch64-simd.md
1521         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
1522         * config/aarch64/aarch64.md
1523         (*movsi_aarch64): Replace type mov_reg with fmovs.
1524         (*movdi_aarch64): Likewise
1525         (*movsf_aarch64): Likewise
1526         (*movdf_aarch64): Likewise
1527         * config/arm/arm.c
1528         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
1529         * config/arm/iwmmxt.md
1530         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
1531         * config/arm/arm1020e.md: Update with new attributes.
1532         * config/arm/cortex-a15-neon.md: Update with new attributes.
1533         * config/arm/cortex-a5.md: Update with new attributes.
1534         * config/arm/cortex-a53.md: Update with new attributes.
1535         * config/arm/cortex-a7.md: Update with new attributes.
1536         * config/arm/cortex-a8-neon.md: Update with new attributes.
1537         * config/arm/cortex-a9.md: Update with new attributes.
1538         * config/arm/cortex-m4-fpu.md: Update with new attributes.
1539         * config/arm/cortex-r4f.md: Update with new attributes.
1540         * config/arm/marvell-pj4.md: Update with new attributes.
1541         * config/arm/vfp11.md: Update with new attributes.
1543 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1545         * config/aarch64/aarch64.md
1546         (*madd<mode>): Fix type attribute.
1547         (*maddsi_uxtw): Likewise.
1548         (*msub<mode>): Likewise.
1549         (*msubsi_uxtw): Likewise.
1550         (<su_optab>maddsidi4): Likewise.
1551         (<su_optab>msubsidi4): Likewise.
1553 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1555         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
1556         * config/arm/arm.md (core_cycles): Remove fdiv.
1557         * config/arm/vfp.md:
1558         (*sqrtsf2_vfp): Update for attribute changes.
1559         (*sqrtdf2_vfp): Likewise.
1560         * config/aarch64/aarch64.md:
1561         (sqrt<mode>2): Update for attribute changes.
1562         * config/arm/arm1020e.md: Update with new attributes.
1563         * config/arm/cortex-a15-neon.md: Update with new attributes.
1564         * config/arm/cortex-a5.md: Update with new attributes.
1565         * config/arm/cortex-a53.md: Update with new attributes.
1566         * config/arm/cortex-a7.md: Update with new attributes.
1567         * config/arm/cortex-a8-neon.md: Update with new attributes.
1568         * config/arm/cortex-a9.md: Update with new attributes.
1569         * config/arm/cortex-m4-fpu.md: Update with new attributes.
1570         * config/arm/cortex-r4f.md: Update with new attributes.
1571         * config/arm/marvell-pj4.md: Update with new attributes.
1572         * config/arm/vfp11.md: Update with new attributes.
1574 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1576         * config/arm/types.md
1577         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
1578         * config/aarch64/aarch64.md
1579         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
1580         new attributes.
1581         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
1582         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
1583         (float<GPI:mode><GPF:mode>2): Likewise.
1584         * config/arm/vfp.md
1585         (*truncsisf2_vfp): Update with new attributes.
1586         (*truncsidf2_vfp): Likewise.
1587         (fixuns_truncsfsi2): Likewise.
1588         (fixuns_truncdfsi2): Likewise.
1589         (*floatsisf2_vfp): Likewise.
1590         (*floatsidf2_vfp): Likewise.
1591         (floatunssisf2): Likewise.
1592         (floatunssidf2): Likewise.
1593         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
1594         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
1595         * config/arm/arm1020e.md: Update with new attributes.
1596         * config/arm/cortex-a15-neon.md: Update with new attributes.
1597         * config/arm/cortex-a5.md: Update with new attributes.
1598         * config/arm/cortex-a53.md: Update with new attributes.
1599         * config/arm/cortex-a7.md: Update with new attributes.
1600         * config/arm/cortex-a8-neon.md: Update with new attributes.
1601         * config/arm/cortex-a9.md: Update with new attributes.
1602         * config/arm/cortex-m4-fpu.md: Update with new attributes.
1603         * config/arm/cortex-r4f.md: Update with new attributes.
1604         * config/arm/marvell-pj4.md: Update with new attributes.
1605         * config/arm/vfp11.md: Update with new attributes.
1607 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1609         * config/aarch64/arm_neon.h
1610         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
1611         (vqtbx<1,2,3,4><q>_s8): Likewise.
1613 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1615         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
1616         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
1617         (add<mode>3): Add type attribute.
1618         (add<mode>3): Likewise.
1619         (usadd<mode>3): Likewise.
1620         (ssadd<mode>3): Likewise.
1621         (sub<mode>3): Likewise.
1622         (sub<mode>3): Likewise.
1623         (ussub<mode>3): Likewise.
1624         (sssub<mode>3): Likewise.
1625         (ssmulsa3): Likewise.
1626         (usmulusa3): Likewise.
1627         (arm_usatsihi): Likewise.
1628         * config/arm/vfp.md
1629         (*movdi_vfp): Add types for all instructions.
1630         (*movdi_vfp_cortexa8): Likewise.
1631         (*movhf_vfp_neon): Likewise.
1632         (*movhf_vfp): Likewise.
1633         (*movdf_vfp): Likewise.
1634         (*thumb2_movdf_vfp): Likewise.
1635         (*thumb2_movdfcc_vfp): Likewise.
1636         * config/arm/arm.md: Add type attribute to all insn patterns.
1637         (*thumb1_adddi3): Add type attribute.
1638         (*arm_adddi3): Likewise.
1639         (*adddi_sesidi_di): Likewise.
1640         (*adddi_zesidi_di): Likewise.
1641         (*thumb1_addsi3): Likewise.
1642         (addsi3_compare0): Likewise.
1643         (*addsi3_compare0_scratch): Likewise.
1644         (*compare_negsi_si): Likewise.
1645         (cmpsi2_addneg): Likewise.
1646         (*addsi3_carryin_<optab>): Likewise.
1647         (*addsi3_carryin_alt2_<optab>): Likewise.
1648         (*addsi3_carryin_clobercc_<optab>): Likewise.
1649         (*subsi3_carryin): Likewise.
1650         (*subsi3_carryin_const): Likewise.
1651         (*subsi3_carryin_compare): Likewise.
1652         (*subsi3_carryin_compare_const): Likewise.
1653         (*arm_subdi3): Likewise.
1654         (*thumb_subdi3): Likewise.
1655         (*subdi_di_zesidi): Likewise.
1656         (*subdi_di_sesidi): Likewise.
1657         (*subdi_zesidi_di): Likewise.
1658         (*subdi_sesidi_di): Likewise.
1659         (*subdi_zesidi_ze): Likewise.
1660         (thumb1_subsi3_insn): Likewise.
1661         (*arm_subsi3_insn): Likewise.
1662         (*anddi3_insn): Likewise.
1663         (*anddi_zesidi_di): Likewise.
1664         (*anddi_sesdi_di): Likewise.
1665         (*ne_zeroextracts): Likewise.
1666         (*ne_zeroextracts): Likewise.
1667         (*ite_ne_zeroextr): Likewise.
1668         (*ite_ne_zeroextr): Likewise.
1669         (*anddi_notdi_di): Likewise.
1670         (*anddi_notzesidi): Likewise.
1671         (*anddi_notsesidi): Likewise.
1672         (andsi_notsi_si): Likewise.
1673         (thumb1_bicsi3): Likewise.
1674         (*iordi3_insn): Likewise.
1675         (*iordi_zesidi_di): Likewise.
1676         (*iordi_sesidi_di): Likewise.
1677         (*thumb1_iorsi3_insn): Likewise.
1678         (*xordi3_insn): Likewise.
1679         (*xordi_zesidi_di): Likewise.
1680         (*xordi_sesidi_di): Likewise.
1681         (*arm_xorsi3): Likewise.
1682         (*andsi_iorsi3_no): Likewise.
1683         (*smax_0): Likewise.
1684         (*smax_m1): Likewise.
1685         (*arm_smax_insn): Likewise.
1686         (*smin_0): Likewise.
1687         (*arm_smin_insn): Likewise.
1688         (*arm_umaxsi3): Likewise.
1689         (*arm_uminsi3): Likewise.
1690         (*minmax_arithsi): Likewise.
1691         (*minmax_arithsi_): Likewise.
1692         (*satsi_<SAT:code>): Likewise.
1693         (arm_ashldi3_1bit): Likewise.
1694         (arm_ashrdi3_1bit): Likewise.
1695         (arm_lshrdi3_1bit): Likewise.
1696         (*arm_negdi2): Likewise.
1697         (*thumb1_negdi2): Likewise.
1698         (*arm_negsi2): Likewise.
1699         (*thumb1_negsi2): Likewise.
1700         (*negdi_extendsid): Likewise.
1701         (*negdi_zero_extend): Likewise.
1702         (*arm_abssi2): Likewise.
1703         (*thumb1_abssi2): Likewise.
1704         (*arm_neg_abssi2): Likewise.
1705         (*thumb1_neg_abss): Likewise.
1706         (one_cmpldi2): Likewise.
1707         (extend<mode>di2): Likewise.
1708         (*compareqi_eq0): Likewise.
1709         (*arm_extendhisi2addsi): Likewise.
1710         (*arm_movdi): Likewise.
1711         (*thumb1_movdi_insn): Likewise.
1712         (*arm_movt): Likewise.
1713         (*thumb1_movsi_insn): Likewise.
1714         (pic_add_dot_plus_four): Likewise.
1715         (pic_add_dot_plus_eight): Likewise.
1716         (tls_load_dot_plus_eight): Likewise.
1717         (*thumb1_movhi_insn): Likewise.
1718         (*thumb1_movsf_insn): Likewise.
1719         (*movdf_soft_insn): Likewise.
1720         (*thumb_movdf_insn): Likewise.
1721         (cbranchsi4_insn): Likewise.
1722         (cbranchsi4_scratch): Likewise.
1723         (*negated_cbranchsi4): Likewise.
1724         (*tbit_cbranch): Likewise.
1725         (*tlobits_cbranch): Likewise.
1726         (*tstsi3_cbranch): Likewise.
1727         (*cbranchne_decr1): Likewise.
1728         (*addsi3_cbranch): Likewise.
1729         (*addsi3_cbranch_scratch): Likewise.
1730         (*arm_cmpdi_insn): Likewise.
1731         (*arm_cmpdi_unsig): Likewise.
1732         (*arm_cmpdi_zero): Likewise.
1733         (*thumb_cmpdi_zero): Likewise.
1734         (*deleted_compare): Likewise.
1735         (*mov_scc): Likewise.
1736         (*mov_negscc): Likewise.
1737         (*mov_notscc): Likewise.
1738         (*cstoresi_eq0_thumb1_insn): Likewise.
1739         (cstoresi_nltu_thumb1): Likewise.
1740         (cstoresi_ltu_thu): Likewise.
1741         (thumb1_addsi3_addgeu): Likewise.
1742         (*arm_jump): Likewise.
1743         (*thumb_jump): Likewise.
1744         (*check_arch2): Likewise.
1745         (arm_casesi_internal): Likewise.
1746         (thumb1_casesi_dispatch): Likewise.
1747         (*arm_indirect_jump): Likewise.
1748         (*thumb1_indirect_jump): Likewise.
1749         (nop): Likewise.
1750         (*and_scc): Likewise.
1751         (*ior_scc): Likewise.
1752         (*compare_scc): Likewise.
1753         (*cond_move): Likewise.
1754         (*cond_arith): Likewise.
1755         (*cond_sub): Likewise.
1756         (*cmp_ite0): Likewise.
1757         (*cmp_ite1): Likewise.
1758         (*cmp_and): Likewise.
1759         (*cmp_ior): Likewise.
1760         (*ior_scc_scc): Likewise.
1761         (*ior_scc_scc_cmp): Likewise.
1762         (*and_scc_scc): Likewise.
1763         (*and_scc_scc_cmp): Likewise.
1764         (*and_scc_scc_nod): Likewise.
1765         (*negscc): Likewise.
1766         (movcond_addsi): Likewise.
1767         (movcond): Likewise.
1768         (*ifcompare_plus_move): Likewise.
1769         (*if_plus_move): Likewise.
1770         (*ifcompare_move_plus): Likewise.
1771         (*if_move_plus): Likewise.
1772         (*ifcompare_arith_arith): Likewise.
1773         (*if_arith_arith): Likewise.
1774         (*ifcompare_arith_move): Likewise.
1775         (*if_arith_move): Likewise.
1776         (*ifcompare_move_arith): Likewise.
1777         (*if_move_arith): Likewise.
1778         (*ifcompare_move_not): Likewise.
1779         (*if_move_not): Likewise.
1780         (*ifcompare_not_move): Likewise.
1781         (*if_not_move): Likewise.
1782         (*ifcompare_shift_move): Likewise.
1783         (*if_shift_move): Likewise.
1784         (*ifcompare_move_shift): Likewise.
1785         (*if_move_shift): Likewise.
1786         (*ifcompare_shift_shift): Likewise.
1787         (*ifcompare_not_arith): Likewise.
1788         (*ifcompare_arith_not): Likewise.
1789         (*if_arith_not): Likewise.
1790         (*ifcompare_neg_move): Likewise.
1791         (*if_neg_move): Likewise.
1792         (*ifcompare_move_neg): Likewise.
1793         (*if_move_neg): Likewise.
1794         (prologue_thumb1_interwork): Likewise.
1795         (*cond_move_not): Likewise.
1796         (*sign_extract_onebit): Likewise.
1797         (*not_signextract_onebit): Likewise.
1798         (stack_tie): Likewise.
1799         (align_4): Likewise.
1800         (align_8): Likewise.
1801         (consttable_end): Likewise.
1802         (consttable_1): Likewise.
1803         (consttable_2): Likewise.
1804         (consttable_4): Likewise.
1805         (consttable_8): Likewise.
1806         (consttable_16): Likewise.
1807         (*thumb1_tablejump): Likewise.
1808         (prefetch): Likewise.
1809         (force_register_use): Likewise.
1810         (thumb_eh_return): Likewise.
1811         (load_tp_hard): Likewise.
1812         (load_tp_soft): Likewise.
1813         (tlscall): Likewise.
1814         (*arm_movtas_ze): Likewise.
1815         (*arm_rev): Likewise.
1816         (*arm_revsh): Likewise.
1817         (*arm_rev16): Likewise.
1818         * config/arm/thumb2.md
1819         (*thumb2_smaxsi3): Likewise.
1820         (*thumb2_sminsi3): Likewise.
1821         (*thumb32_umaxsi3): Likewise.
1822         (*thumb2_uminsi3): Likewise.
1823         (*thumb2_negdi2): Likewise.
1824         (*thumb2_abssi2): Likewise.
1825         (*thumb2_neg_abss): Likewise.
1826         (*thumb2_movsi_insn): Likewise.
1827         (tls_load_dot_plus_four): Likewise.
1828         (*thumb2_movhi_insn): Likewise.
1829         (*thumb2_mov_scc): Likewise.
1830         (*thumb2_mov_negs): Likewise.
1831         (*thumb2_mov_negs): Likewise.
1832         (*thumb2_mov_nots): Likewise.
1833         (*thumb2_mov_nots): Likewise.
1834         (*thumb2_movsicc_): Likewise.
1835         (*thumb2_movsfcc_soft_insn): Likewise.
1836         (*thumb2_indirect_jump): Likewise.
1837         (*thumb2_and_scc): Likewise.
1838         (*thumb2_ior_scc): Likewise.
1839         (*thumb2_ior_scc_strict_it): Likewise.
1840         (*thumb2_cond_move): Likewise.
1841         (*thumb2_cond_arith): Likewise.
1842         (*thumb2_cond_ari): Likewise.
1843         (*thumb2_cond_sub): Likewise.
1844         (*thumb2_negscc): Likewise.
1845         (*thumb2_movcond): Likewise.
1846         (thumb2_casesi_internal): Likewise.
1847         (thumb2_casesi_internal_pic): Likewise.
1848         (*thumb2_alusi3_short): Likewise.
1849         (*thumb2_mov<mode>_shortim): Likewise.
1850         (*thumb2_addsi_short): Likewise.
1851         (*thumb2_subsi_short): Likewise.
1852         (thumb2_addsi3_compare0): Likewise.
1853         (*thumb2_cbz): Likewise.
1854         (*thumb2_cbnz): Likewise.
1855         (*thumb2_one_cmplsi2_short): Likewise.
1856         (*thumb2_negsi2_short): Likewise.
1857         (*orsi_notsi_si): Likewise.
1858         * config/arm/arm1020e.md: Update with new attributes.
1859         * config/arm/arm1026ejs.md: Update with new attributes.
1860         * config/arm/arm1136jfs.md: Update with new attributes.
1861         * config/arm/arm926ejs.md: Update with new attributes.
1862         * config/arm/cortex-a15.md: Update with new attributes.
1863         * config/arm/cortex-a5.md: Update with new attributes.
1864         * config/arm/cortex-a53.md: Update with new attributes.
1865         * config/arm/cortex-a7.md: Update with new attributes.
1866         * config/arm/cortex-a8.md: Update with new attributes.
1867         * config/arm/cortex-a9.md: Update with new attributes.
1868         * config/arm/cortex-m4.md: Update with new attributes.
1869         * config/arm/cortex-r4.md: Update with new attributes.
1870         * config/arm/fa526.md: Update with new attributes.
1871         * config/arm/fa606te.md: Update with new attributes.
1872         * config/arm/fa626te.md: Update with new attributes.
1873         * config/arm/fa726te.md: Update with new attributes.
1875 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1877         * config/aarch64/aarch64-simd.md
1878         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
1879         <vwx> iterator to ensure correct register choice.
1880         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
1881         (aarch64_sqdmull_n<mode>): Likewise.
1882         (aarch64_sqdmull2_n<mode>_internal): Likewise.
1883         * config/aarch64/arm_neon.h
1884         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
1885         (vml<as><q>_n_<su>16): Likewise.
1886         (vml<as>l_high_lane<q>_<su>16): Likewise.
1887         (vml<as>l_high_n_<su>16): Likewise.
1888         (vml<as>l_lane<q>_<su>16): Likewise.
1889         (vml<as>l_n_<su>16): Likewise.
1890         (vmul<q>_lane<q>_<su>16): Likewise.
1891         (vmul<q>_n_<su>16): Likewise.
1892         (vmull_lane<q>_<su>16): Likewise.
1893         (vmull_n_<su>16): Likewise.
1894         (vmull_high_lane<q>_<su>16): Likewise.
1895         (vmull_high_n_<su>16): Likewise.
1896         (vqrdmulh<q>_n_s16): Likewise.
1898 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
1900         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
1901         have the correct lane parameter.
1903 2013-09-06  Richard Biener <rguenther@suse.de>
1905         * cfganal.c (control_dependences::~control_dependences):
1906         Properly free all of the vector.
1908 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
1910         PR target/58269
1911         * config/i386/i386.c (ix86_conditional_register_usage):
1912         Proper initialize extended SSE registers.
1914 2013-09-06  Jan Hubicka  <jh@suse.cz>
1916         PR tree-optimization/58311
1917         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
1919 2013-09-06  Jan Hubicka  <jh@suse.cz>
1921         * Makefile.in (tree-sra.o): Update dependencies.
1922         * tree-sra.c: Include ipa-utils.h
1923         (scan_function): Use recursive_call_p.
1924         (has_caller_p): New function.
1925         (cgraph_for_node_and_aliases): Count also callers of aliases.
1927 2013-09-06  Jan Hubicka  <jh@suse.cz>
1929         PR middle-end/58094
1930         * cgraph.h (symtab_semantically_equivalent_p): Declare.
1931         * tree-tailcall.c: Include ipa-utils.h.
1932         (find_tail_calls): Use it.
1933         * ipa-pure-const.c (check_call): Likewise.
1934         * ipa-utils.c (recursive_call_p): New function.
1935         * ipa-utils.h (recursive_call_p): Dclare.
1936         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
1937         (symtab_semantically_equivalent_p): New function.
1938         * Makefile.in (tree-tailcall.o): Update dependencies.
1940 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
1942         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
1943         non-inlinable part.
1945 2013-09-06  Richard Biener  <rguenther@suse.de>
1947         * lto-streamer.h (lto_global_var_decls): Remove.
1948         * Makefile.in (OBJS): Remove lto-symtab.o.
1949         (lto-symtab.o): Remove.
1950         (GTFILES): Remove lto-symtab.c
1951         * lto-symtab.c: Move to lto/
1953 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1955         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
1956         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
1957         (UNSPEC_FPINT_RINT): New constant definitions.
1958         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
1959         definition with 2 attributes.
1960         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
1961         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
1962         definitions.
1964 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1966         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
1967         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
1968         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
1969         attribute to "z196_alone".
1970         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
1971         "zEC12_simple".
1973 2013-09-06  Richard Biener  <rguenther@suse.de>
1975         * basic-block.h (class control_dependences): New.
1976         * tree-ssa-dce.c (control_dependence_map): Remove.
1977         (cd): New global.
1978         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
1979         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
1980         find_pdom, find_control_dependence, find_all_control_dependences):
1981         Move to cfganal.c.
1982         (mark_control_dependent_edges_necessary,
1983         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
1984         tree_dce_done, perform_tree_ssa_dce): Adjust.
1985         * cfganal.c (set_control_dependence_map_bit,
1986         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
1987         find_all_control_dependences): Move from tree-ssa-dce.c and
1988         implement as methods of control_dependences class.
1989         (control_dependences::control_dependences): New.
1990         (control_dependences::~control_dependences): Likewise.
1991         (control_dependences::get_edges_dependent_on): Likewise.
1992         (control_dependences::get_edge): Likewise.
1994 2013-09-04  Jan Hubicka  <jh@suse.cz>
1996         * tree.c (types_same_for_odr): Drop overactive check.
1997         * ipa-devirt.c (hash_type_name): Likewise.
1999 2013-09-04  Jan Hubicka  <jh@suse.cz>
2001         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
2002         (analyze_functions): ... here.
2004 2013-09-04  Jan Hubicka  <jh@suse.cz>
2006         PR middle-end/58201
2007         * cgraphunit.c (analyze_functions): Clear AUX fields
2008         after processing; initialize assembler name has.
2010 2013-09-05  Jeff Law  <law@redhat.com>
2012         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
2013         from thread_around_empty_block.  Record threading path into PATH.
2014         Recurse if threading through the initial block is successful.
2015         (thread_across_edge): Corresponding changes to slightly simplify.
2017 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
2019         * config/aarch64/aarch64.md
2020         (type): Remove frecpe, frecps, frecpx.
2021         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
2022         fix to be a TARGET_SIMD instruction.
2023         (aarch64_frecps): Remove.
2024         * config/aarch64/aarch64-simd.md
2025         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
2026         (aarch64_frecps<mode>): Handle all float/vector of float modes.
2028 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
2029             Sofiane Naci  <sofiane.naci@arm.com>
2031         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
2032         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
2033         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
2034         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
2035         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
2036         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
2037         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
2038         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
2039         into "clz, "rbit".  Rename "shift" to "shift_imm".
2040         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
2041         changes.  Update for attribute changes all occurrences of arlo_* and
2042         shift* types.
2043         * config/arm/arm-fixed.md: Update for attribute changes
2044         all occurrences of arlo_* types.
2045         * config/arm/thumb2.md: Update for attribute changes all occurrences
2046         of arlo_* types.
2047         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
2048         (cortexa7_older_only): Likewise.
2049         (cortexa7_younger):  Likewise.
2050         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
2051         (1020alu_shift_op): Likewise.
2052         (1020alu_shift_reg_op): Likewise.
2053         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
2054         (alu_shift_op): Likewise.
2055         (alu_shift_reg_op): Likewise.
2056         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
2057         (11_alu_shift_op): Likewise.
2058         (11_alu_shift_reg_op): Likewise.
2059         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
2060         (9_alu_shift_reg_op): Likewise.
2061         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
2062         attribute changes.
2063         (cortex_a15_alu_shift): Likewise.
2064         (cortex_a15_alu_shift_reg): Likewise.
2065         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
2066         attribute changes.
2067         (cortex_a5_alu_shift): Likewise.
2068         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
2069         attribute changes.
2070         (cortex_a53_alu_shift): Likewise.
2071         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
2072         attribute changes.
2073         (cortex_a7_alu_reg): Likewise.
2074         (cortex_a7_alu_shift): Likewise.
2075         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
2076         attribute changes.
2077         (cortex_a8_alu_shift): Likewise.
2078         (cortex_a8_alu_shift_reg): Likewise.
2079         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
2080         (cortex_a9_dp_shift): Likewise.
2081         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
2082         attribute changes.
2083         * config/arm/cortex-r4.md
2084         (cortex_r4_alu): Update for attribute changes.
2085         (cortex_r4_mov): Likewise.
2086         (cortex_r4_alu_shift_reg): Likewise.
2087         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
2088         (526_alu_shift_op): Likewise.
2089         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
2090         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
2091         (626te_alu_shift_op): Likewise.
2092         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
2093         (726te_alu_shift_op): Likewise.
2094         (726te_alu_shift_reg_op): Likewise.
2095         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
2096         (mp626_alu_shift_op): Likewise.
2097         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
2098         (pj4_alu_conds): Likewise.
2099         (pj4_shift): Likewise.
2100         (pj4_shift_conds): Likewise.
2101         (pj4_alu_shift): Likewise.
2102         (pj4_alu_shift_conds): Likewise.
2103         * config/aarch64/aarch64.md: Update for attribute change
2104         all occurrences of arlo_* and shift* types.
2106 2013-09-05  Mike Stump  <mikestump@comcast.net>
2108         * tree.h: Move documentation for tree_function_decl to tree-core.h
2109         with the declaration.
2111 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
2113         PR target/58139
2114         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
2115         looking for widest mode.
2117 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
2119         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
2121 2013-09-05  Richard Biener  <rguenther@suse.de>
2123         PR tree-optimization/58137
2124         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
2125         Do not create vectors of pointers.
2126         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
2127         types for the components of the vector initializer.
2128         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
2129         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
2131 2013-09-05  Martin Jambor  <mjambor@suse.cz>
2133         * ipa-prop.c (remove_described_reference): Accept missing references,
2134         return false if that hppens, otherwise return true.
2135         (cgraph_node_for_jfunc): New function.
2136         (try_decrement_rdesc_refcount): Likewise.
2137         (try_make_edge_direct_simple_call): Use them.
2138         (ipa_edge_removal_hook): Remove references from rdescs.
2139         (ipa_edge_duplication_hook): Clone rdescs and their references
2140         when the new edge has the same caller as the old one.
2141         * cgraph.c (cgraph_resolve_speculation): Remove speculative
2142         reference before removing any edges.
2144 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
2146         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
2147         initial store.
2148         * thumb2.md (thumb2_storewb_parisi): New pattern.
2150 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
2152         * config/aarch64/aarch64-option-extensions.def: Add
2153         AARCH64_OPT_EXTENSION of 'crc'.
2154         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
2155         (AARCH64_ISA_CRC): Ditto.
2156         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
2157         description of the CRC extension.
2159 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2161         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
2162         * config/rs6000/linux.h: Ditto.
2163         * alpha/linux.h: Ditto.
2164         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
2165         no_c99_libc_has_function.
2166         * config/c6x/uclinux-elf.h: Ditto.
2167         * config/lm32/uclinux-elf.h: Ditto.
2168         * config/m68k/uclinux.h: Ditto.
2169         * config/moxie/uclinux.h: Ditto.
2170         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
2171         (crisv32-*-linux*, cris-*-linux*): Ditto.
2172         * config/bfin/bfin.c: Include "tm_p.h".
2174 2013-09-05  Richard Biener  <rguenther@suse.de>
2176         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
2177         check for a definition without a basic-block.
2179 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
2180             Sofiane Naci <sofiane.naci@arm.com>
2182         * config/aarch64/aarch64.md
2183         (*movti_aarch64): Rename r_2_f and f_2_r.
2184         (*movsf_aarch64): Likewise.
2185         (*movdf_aarch64): Likewise.
2186         (*movtf_aarch64): Likewise.
2187         (aarch64_movdi_<mode>low): Likewise.
2188         (aarch64_movdi_<mode>high): Likewise.
2189         (aarch64_mov<mode>high_di): Likewise.
2190         (aarch64_mov<mode>low_di): Likewise.
2191         (aarch64_movtilow_tilow): Likewise.
2192         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
2193         values to config/arm/types.md
2194         (attribute "conds"): Update for attribute change.
2195         (anddi3_insn): Likewise.
2196         (iordi3_insn): Likewise.
2197         (xordi3_insn): Likewise.
2198         (one_cmpldi2): Likewise.
2199         * config/arm/types.md (type): Add Neon types.
2200         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
2201         use "type" attribute.
2202         (movmisalign<mode>_neon_store): Likewise.
2203         (movmisalign<mode>_neon_load): Likewise.
2204         (vec_set<mode>_internal): Likewise.
2205         (vec_setv2di_internal): Likewise.
2206         (vec_extract<mode>): Likewise.
2207         (vec_extractv2di): Likewise.
2208         (add<mode>3_neon): Likewise.
2209         (adddi3_neon): Likewise.
2210         (sub<mode>3_neon): Likewise.
2211         (subdi3_neon): Likewise.
2212         (mul<mode>3_neon): Likewise.
2213         (mul<mode>3add<mode>_neon): Likewise.
2214         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
2215         (fma<VCVTF:mode>4)): Likewise.
2216         (fma<VCVTF:mode>4_intrinsic): Likewise.
2217         (fmsub<VCVTF:mode>4)): Likewise.
2218         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
2219         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
2220         (ior<mode>3): Likewise.
2221         (and<mode>3): Likewise.
2222         (anddi3_neon): Likewise.
2223         (orn<mode>3_neon): Likewise.
2224         (orndi3_neon): Likewise.
2225         (bic<mode>3_neon): Likewise.
2226         (bicdi3_neon): Likewise.
2227         (xor<mode>3): Likewise.
2228         (one_cmpl<mode>2): Likewise.
2229         (abs<mode>2): Likewise.
2230         (neg<mode>2): Likewise.
2231         (umin<mode>3_neon): Likewise.
2232         (umax<mode>3_neon): Likewise.
2233         (smin<mode>3_neon): Likewise.
2234         (smax<mode>3_neon): Likewise.
2235         (vashl<mode>3): Likewise.
2236         (vashr<mode>3_imm): Likewise.
2237         (vlshr<mode>3_imm): Likewise.
2238         (ashl<mode>3_signed): Likewise.
2239         (ashl<mode>3_unsigned): Likewise.
2240         (neon_load_count): Likewise.
2241         (ashldi3_neon_noclobber): Likewise.
2242         (signed_shift_di3_neon): Likewise.
2243         (unsigned_shift_di3_neon): Likewise.
2244         (ashrdi3_neon_imm_noclobber): Likewise.
2245         (lshrdi3_neon_imm_noclobber): Likewise.
2246         (widen_ssum<mode>3): Likewise.
2247         (widen_usum<mode>3): Likewise.
2248         (quad_halves_<code>v4si): Likewise.
2249         (quad_halves_<code>v4sf): Likewise.
2250         (quad_halves_<code>v8hi): Likewise.
2251         (quad_halves_<code>v16qi): Likewise.
2252         (reduc_splus_v2di): Likewise.
2253         (neon_vpadd_internal<mode>): Likewise.
2254         (neon_vpsmin<mode>): Likewise.
2255         (neon_vpsmax<mode>): Likewise.
2256         (neon_vpumin<mode>): Likewise.
2257         (neon_vpumax<mode>): Likewise.
2258         (ss_add<mode>_neon): Likewise.
2259         (us_add<mode>_neon): Likewise.
2260         (ss_sub<mode>_neon): Likewise.
2261         (us_sub<mode>_neon): Likewise.
2262         (neon_vadd<mode>_unspec): Likewise.
2263         (neon_vaddl<mode>): Likewise.
2264         (neon_vaddw<mode>): Likewise.
2265         (neon_vhadd<mode>): Likewise.
2266         (neon_vqadd<mode>): Likewise.
2267         (neon_vaddhn<mode>): Likewise.
2268         (neon_vmul<mode>): Likewise.
2269         (neon_vmla<mode>): Likewise.
2270         (neon_vmlal<mode>): Likewise.
2271         (neon_vmls<mode>): Likewise.
2272         (neon_vmlsl<mode>): Likewise.
2273         (neon_vqdmulh<mode>): Likewise.
2274         (neon_vqdmlal<mode>): Likewise.
2275         (neon_vqdmlsl<mode>): Likewise.
2276         (neon_vmull<mode>): Likewise.
2277         (neon_vqdmull<mode>): Likewise.
2278         (neon_vsub<mode>_unspec): Likewise.
2279         (neon_vsubl<mode>): Likewise.
2280         (neon_vsubw<mode>): Likewise.
2281         (neon_vqsub<mode>): Likewise.
2282         (neon_vhsub<mode>): Likewise.
2283         (neon_vsubhn<mode>): Likewise.
2284         (neon_vceq<mode>): Likewise.
2285         (neon_vcge<mode>): Likewise.
2286         (neon_vcgeu<mode>): Likewise.
2287         (neon_vcgt<mode>): Likewise.
2288         (neon_vcgtu<mode>): Likewise.
2289         (neon_vcle<mode>): Likewise.
2290         (neon_vclt<mode>): Likewise.
2291         (neon_vcage<mode>): Likewise.
2292         (neon_vcagt<mode>): Likewise.
2293         (neon_vtst<mode>): Likewise.
2294         (neon_vabd<mode>): Likewise.
2295         (neon_vabdl<mode>): Likewise.
2296         (neon_vaba<mode>): Likewise.
2297         (neon_vabal<mode>): Likewise.
2298         (neon_vmax<mode>): Likewise.
2299         (neon_vmin<mode>): Likewise.
2300         (neon_vpaddl<mode>): Likewise.
2301         (neon_vpadal<mode>): Likewise.
2302         (neon_vpmax<mode>): Likewise.
2303         (neon_vpmin<mode>): Likewise.
2304         (neon_vrecps<mode>): Likewise.
2305         (neon_vrsqrts<mode>): Likewise.
2306         (neon_vqabs<mode>): Likewise.
2307         (neon_vqneg<mode>): Likewise.
2308         (neon_vcls<mode>): Likewise.
2309         (clz<mode>2): Likewise.
2310         (popcount<mode>2): Likewise.
2311         (neon_vrecpe): Likewise.
2312         (neon_vrsqrte): Likewise.
2313         (neon_vget_lane<mode>_sext_internal): Likewise.
2314         (neon_vget_lane<mode>_zext_internal): Likewise.
2315         (neon_vdup_n<mode>): Likewise.
2316         (neon_vdup_nv2di): Likewise.
2317         (neon_vdpu_lane<mode>_internal): Likewise.
2318         (neon_vswp<mode>): Likewise.
2319         (float<mode><V_cvtto>2): Likewise.
2320         (floatuns<mode><V_cvtto>2): Likewise.
2321         (fix_trunc<mode><V_cvtto>)2): Likewise
2322         (fixuns_trunc<mode><V_cvtto)2): Likewise.
2323         (neon_vcvt<mode>): Likewise.
2324         (neon_vcvtv4sfv4hf): Likewise.
2325         (neon_vcvtv4hfv4sf): Likewise.
2326         (neon_vcvt_n<mode>): Likewise.
2327         (neon_vmovn<mode>): Likewise.
2328         (neon_vqmovn<mode>): Likewise.
2329         (neon_vqmovun<mode>): Likewise.
2330         (neon_vmovl<mode>): Likewise.
2331         (neon_vmul_lane<mode>): Likewise.
2332         (neon_vmull_lane<mode>): Likewise.
2333         (neon_vqdmull_lane<mode>): Likewise.
2334         (neon_vqdmulh_lane<mode>): Likewise.
2335         (neon_vmla_lane<mode>): Likewise.
2336         (neon_vmlal_lane<mode>): Likewise.
2337         (neon_vqdmlal_lane<mode>): Likewise.
2338         (neon_vmls_lane<mode>): Likewise.
2339         (neon_vmlsl_lane<mode>): Likewise.
2340         (neon_vqdmlsl_lane<mode>): Likewise.
2341         (neon_vext<mode>): Likewise.
2342         (neon_vrev64<mode>): Likewise.
2343         (neon_vrev32<mode>): Likewise.
2344         (neon_vrev16<mode>): Likewise.
2345         (neon_vbsl<mode>_internal): Likewise.
2346         (neon_vshl<mode>): Likewise.
2347         (neon_vqshl<mode>): Likewise.
2348         (neon_vshr_n<mode>): Likewise.
2349         (neon_vshrn_n<mode>): Likewise.
2350         (neon_vqshrn_n<mode>): Likewise.
2351         (neon_vqshrun_n<mode>): Likewise.
2352         (neon_vshl_n<mode>): Likewise.
2353         (neon_vqshl_n<mode>): Likewise.
2354         (neon_vqshlu_n<mode>): Likewise.
2355         (neon_vshll_n<mode>): Likewise.
2356         (neon_vsra_n<mode>): Likewise.
2357         (neon_vsri_n<mode>): Likewise.
2358         (neon_vsli_n<mode>): Likewise.
2359         (neon_vtbl1v8qi): Likewise.
2360         (neon_vtbl2v8qi): Likewise.
2361         (neon_vtbl3v8qi): Likewise.
2362         (neon_vtbl4v8qi): Likewise.
2363         (neon_vtbx1v8qi): Likewise.
2364         (neon_vtbx2v8qi): Likewise.
2365         (neon_vtbx3v8qi): Likewise.
2366         (neon_vtbx4v8qi): Likewise.
2367         (neon_vtrn<mode>_internal): Likewise.
2368         (neon_vzip<mode>_internal): Likewise.
2369         (neon_vuzp<mode>_internal): Likewise.
2370         (neon_vld1<mode>): Likewise.
2371         (neon_vld1_lane<mode>): Likewise.
2372         (neon_vld1_dup<mode>): Likewise.
2373         (neon_vld1_dupv2di): Likewise.
2374         (neon_vst1<mode>): Likewise.
2375         (neon_vst1_lane<mode>): Likewise.
2376         (neon_vld2<mode>): Likewise.
2377         (neon_vld2_lane<mode>): Likewise.
2378         (neon_vld2_dup<mode>): Likewise.
2379         (neon_vst2<mode>): Likewise.
2380         (neon_vst2_lane<mode>): Likewise.
2381         (neon_vld3<mode>): Likewise.
2382         (neon_vld3qa<mode>): Likewise.
2383         (neon_vld3qb<mode>): Likewise.
2384         (neon_vld3_lane<mode>): Likewise.
2385         (neon_vld3_dup<mode>): Likewise.
2386         (neon_vst3<mode>): Likewise.
2387         (neon_vst3qa<mode>): Likewise.
2388         (neon_vst3qb<mode>): Likewise.
2389         (neon_vst3_lane<mode>): Likewise.
2390         (neon_vld4<mode>): Likewise.
2391         (neon_vld4qa<mode>): Likewise.
2392         (neon_vld4qb<mode>): Likewise.
2393         (neon_vld4_lane<mode>): Likewise.
2394         (neon_vld4_dup<mode>): Likewise.
2395         (neon_vst4<mode>): Likewise.
2396         (neon_vst4qa<mode>): Likewise.
2397         (neon_vst4qb<mode>): Likewise.
2398         (neon_vst4_lane<mode>): Likewise.
2399         (neon_vec_unpack<US>_lo_<mode>): Likewise.
2400         (neon_vec_unpack<US>_hi_<mode>): Likewise.
2401         (neon_vec_<US>mult_lo_<mode>): Likewise.
2402         (neon_vec_<US>mult_hi_<mode>): Likewise.
2403         (neon_vec_<US>shiftl_<mode>): Likewise.
2404         (neon_unpack<US>_<mode>): Likewise.
2405         (neon_vec_<US>mult_<mode>): Likewise.
2406         (vec_pack_trunc_<mode>): Likewise.
2407         (neon_vec_pack_trunk_<mode>): Likewise.
2408         (neon_vabd<mode>_2): Likewise.
2409         (neon_vabd<mode>_3): Likewise.
2410         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
2411         (thumb2_movsi_vfp): Likewise.
2412         (movdi_vfp): Likewise.
2413         (movdi_vfp_cortexa8): Likewise.
2414         (movhf_vfp_neon): Likewise.
2415         (movhf_vfp): Likewiwse.
2416         (movsf_vfp): Likewiwse.
2417         (thumb2_movsf_vfp): Likewiwse.
2418         (movdf_vfp): Likewise.
2419         (thumb2_movdf_vfp): Likewise.
2420         (movsfcc_vfp): Likewise.
2421         (thumb2_movsfcc_vfp): Likewise.
2422         (movdfcc_vfp): Likewise.
2423         (thumb2_movdfcc_vfp): Likewise.
2424         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
2425         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
2426         (v10_v2c): Likewise.
2427         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
2428         attribute change.
2429         (cortex_a15_neon_int_2): Likewise.
2430         (cortex_a15_neon_int_3): Likewise.
2431         (cortex_a15_neon_int_4): Likewise.
2432         (cortex_a15_neon_int_5): Likewise.
2433         (cortex_a15_neon_vqneg_vqabs): Likewise.
2434         (cortex_a15_neon_vmov): Likewise.
2435         (cortex_a15_neon_vaba): Likewise.
2436         (cortex_a15_neon_vaba_qqq): Likewise.
2437         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2438         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
2439         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
2440         scalar_64_32_long_scalar): Likewise.
2441         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2442         (cortex_a15_neon_mla_qqq_8_16): Likewise.
2443         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
2444         lotype_qdd_64_32_long): Likewise.
2445         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
2446         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
2447         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
2448         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
2449         (cortex_a15_neon_shift_1): Likewise.
2450         (cortex_a15_neon_shift_2): Likewise.
2451         (cortex_a15_neon_shift_3): Likewise.
2452         (cortex_a15_neon_vshl_ddd): Likewise.
2453         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
2454         (cortex_a15_neon_vsra_vrsra): Likewise.
2455         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
2456         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
2457         (cortex_a15_neon_fp_vmul_ddd): Likewise.
2458         (cortex_a15_neon_fp_vmul_qqd): Likewise.
2459         (cortex_a15_neon_fp_vmla_ddd): Likewise.
2460         (cortex_a15_neon_fp_vmla_qqq): Likewise.
2461         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
2462         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
2463         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
2464         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
2465         (cortex_a15_neon_bp_simple): Likewise.
2466         (cortex_a15_neon_bp_2cycle): Likewise.
2467         (cortex_a15_neon_bp_3cycle): Likewise.
2468         (cortex_a15_neon_vld1_1_2_regs): Likewise.
2469         (cortex_a15_neon_vld1_3_4_regs): Likewise.
2470         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
2471         (cortex_a15_neon_vld2_4_regs): Likewise.
2472         (cortex_a15_neon_vld3_vld4): Likewise.
2473         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
2474         (cortex_a15_neon_vst1_3_4_regs): Likewise.
2475         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
2476         (cortex_a15_neon_vst3_vst4): Likewise.
2477         (cortex_a15_neon_vld1_vld2_lane): Likewise.
2478         (cortex_a15_neon_vld3_vld4_lane" 10
2479         (cortex_a15_neon_vst1_vst2_lane): Likewise.
2480         (cortex_a15_neon_vst3_vst4_lane): Likewise.
2481         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
2482         (cortex_a15_neon_ldm_2): Likewise.0
2483         (cortex_a15_neon_stm_2): Likewise.
2484         (cortex_a15_neon_mcr): Likewise.
2485         (cortex_a15_neon_mcr_2_mcrr): Likewise.
2486         (cortex_a15_neon_mrc): Likewise.
2487         (cortex_a15_neon_mrrc): Likewise.
2488         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
2489         change.
2490         (cortex_a15_alu_shift): Likewise.
2491         (cortex_a15_alu_shift_reg): Likewise.
2492         (cortex_a15_mult32): Likewise.
2493         (cortex_a15_mult64): Likewise.
2494         (cortex_a15_block): Likewise.
2495         (cortex_a15_branch): Likewise.
2496         (cortex_a15_load1): Likewise.
2497         (cortex_a15_load3): Likewise.
2498         (cortex_a15_store1): Likewise.
2499         (cortex_a15_store3): Likewise.
2500         (cortex_a15_call): Likewise.
2501         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
2502         (cortex_a5_f2r): Likewise.
2503         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
2504         change.
2505         (cortex_a53_f2r): Likewise.
2506         * config/arm/cortex-a7.md
2507         (cortex_a7_branch): Update for attribute change.
2508         (cortex_a7_call): Likewise.
2509         (cortex_a7_alu_imm): Likewise.
2510         (cortex_a7_alu_reg): Likewise.
2511         (cortex_a7_alu_shift): Likewise.
2512         (cortex_a7_mul): Likewise.
2513         (cortex_a7_load1): Likewise.
2514         (cortex_a7_store1): Likewise.
2515         (cortex_a7_load2): Likewise.
2516         (cortex_a7_store2): Likewise.
2517         (cortex_a7_load3): Likewise.
2518         (cortex_a7_store3): Likewise.
2519         (cortex_a7_load4): Likewise.
2520         (cortex_a7_store4): Likewise.
2521         (cortex_a7_fpalu): Likewise.
2522         (cortex_a7_fconst): Likewise.
2523         (cortex_a7_fpmuls): Likewise.
2524         (cortex_a7_neon_mul): Likewise.
2525         (cortex_a7_fpmacs): Likewise.
2526         (cortex_a7_neon_mla: Likewise.
2527         (cortex_a7_fpmuld: Likewise.
2528         (cortex_a7_fpmacd: Likewise.
2529         (cortex_a7_fpfmad: Likewise.
2530         (cortex_a7_fdivs: Likewise.
2531         (cortex_a7_fdivd: Likewise.
2532         (cortex_a7_r2f: Likewise.
2533         (cortex_a7_f2r: Likewise.
2534         (cortex_a7_f_flags: Likewise.
2535         (cortex_a7_f_loads: Likewise.
2536         (cortex_a7_f_loadd: Likewise.
2537         (cortex_a7_f_stores: Likewise.
2538         (cortex_a7_f_stored: Likewise.
2539         (cortex_a7_neon): Likewise.
2540         * config/arm/cortex-a8-neon.md
2541         (cortex_a8_neon_mrc): Update for attribute change.
2542         (cortex_a8_neon_mrrc): Likewise.
2543         (cortex_a8_neon_int_1): Likewise.
2544         (cortex_a8_neon_int_2): Likewise.
2545         (cortex_a8_neon_int_3): Likewise.
2546         (cortex_a8_neon_int_4): Likewise.
2547         (cortex_a8_neon_int_5): Likewise.
2548         (cortex_a8_neon_vqneg_vqabs): Likewise.
2549         (cortex_a8_neon_vmov): Likewise.
2550         (cortex_a8_neon_vaba): Likewise.
2551         (cortex_a8_neon_vaba_qqq): Likewise.
2552         (cortex_a8_neon_vsma): Likewise.
2553         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2554         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
2555         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
2556         long_scalar): Likewise.
2557         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2558         (cortex_a8_neon_mla_qqq_8_16): Likewise.
2559         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
2560         long_scalar_qdd_64_32_long): Likewise.
2561         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
2562         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
2563         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
2564         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
2565         (cortex_a8_neon_shift_1): Likewise.
2566         (cortex_a8_neon_shift_2): Likewise.
2567         (cortex_a8_neon_shift_3): Likewise.
2568         (cortex_a8_neon_vshl_ddd): Likewise.
2569         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
2570         (cortex_a8_neon_vsra_vrsra): Likewise.
2571         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
2572         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
2573         (cortex_a8_neon_fp_vsum): Likewise.
2574         (cortex_a8_neon_fp_vmul_ddd): Likewise.
2575         (cortex_a8_neon_fp_vmul_qqd): Likewise.
2576         (cortex_a8_neon_fp_vmla_ddd): Likewise.
2577         (cortex_a8_neon_fp_vmla_qqq): Likewise.
2578         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
2579         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
2580         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
2581         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
2582         (cortex_a8_neon_bp_simple): Likewise.
2583         (cortex_a8_neon_bp_2cycle): Likewise.
2584         (cortex_a8_neon_bp_3cycle): Likewise.
2585         (cortex_a8_neon_ldr): Likewise.
2586         (cortex_a8_neon_str): Likewise.
2587         (cortex_a8_neon_vld1_1_2_regs): Likewise.
2588         (cortex_a8_neon_vld1_3_4_regs): Likewise.
2589         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
2590         (cortex_a8_neon_vld2_4_regs): Likewise.
2591         (cortex_a8_neon_vld3_vld4): Likewise.
2592         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
2593         (cortex_a8_neon_vst1_3_4_regs): Likewise.
2594         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
2595         (cortex_a8_neon_vst3_vst4): Likewise.
2596         (cortex_a8_neon_vld1_vld2_lane): Likewise.
2597         (cortex_a8_neon_vld3_vld4_lane): Likewise.
2598         (cortex_a8_neon_vst1_vst2_lane): Likewise.
2599         (cortex_a8_neon_vst3_vst4_lane): Likewise.
2600         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
2601         (cortex_a8_neon_mcr): Likewise.
2602         (cortex_a8_neon_mcr_2_mcrr): Likewise.
2603         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
2604         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
2605         change.
2606         (ca9_neon_mrrc): Likewise.
2607         (cortex_a9_neon_int_1): Likewise.
2608         (cortex_a9_neon_int_2): Likewise.
2609         (cortex_a9_neon_int_3): Likewise.
2610         (cortex_a9_neon_int_4): Likewise.
2611         (cortex_a9_neon_int_5): Likewise.
2612         (cortex_a9_neon_vqneg_vqabs): Likewise.
2613         (cortex_a9_neon_vmov): Likewise.
2614         (cortex_a9_neon_vaba): Likewise.
2615         (cortex_a9_neon_vaba_qqq): Likewise.
2616         (cortex_a9_neon_vsma): Likewise.
2617         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2618         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
2619         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
2620         long_scalar): Likewise.
2621         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2622         (cortex_a9_neon_mla_qqq_8_16): Likewise.
2623         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
2624         long_scalar_qdd_64_32_long): Likewise.
2625         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
2626         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
2627         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
2628         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
2629         (cortex_a9_neon_shift_1): Likewise.
2630         (cortex_a9_neon_shift_2): Likewise.
2631         (cortex_a9_neon_shift_3): Likewise.
2632         (cortex_a9_neon_vshl_ddd): Likewise.
2633         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
2634         (cortex_a9_neon_vsra_vrsra): Likewise.
2635         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
2636         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
2637         (cortex_a9_neon_fp_vsum): Likewise.
2638         (cortex_a9_neon_fp_vmul_ddd): Likewise.
2639         (cortex_a9_neon_fp_vmul_qqd): Likewise.
2640         (cortex_a9_neon_fp_vmla_ddd): Likewise.
2641         (cortex_a9_neon_fp_vmla_qqq): Likewise.
2642         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
2643         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
2644         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
2645         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
2646         (cortex_a9_neon_bp_simple): Likewise.
2647         (cortex_a9_neon_bp_2cycle): Likewise.
2648         (cortex_a9_neon_bp_3cycle): Likewise.
2649         (cortex_a9_neon_ldr): Likewise.
2650         (cortex_a9_neon_str): Likewise.
2651         (cortex_a9_neon_vld1_1_2_regs): Likewise.
2652         (cortex_a9_neon_vld1_3_4_regs): Likewise.
2653         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
2654         (cortex_a9_neon_vld2_4_regs): Likewise.
2655         (cortex_a9_neon_vld3_vld4): Likewise.
2656         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
2657         (cortex_a9_neon_vst1_3_4_regs): Likewise.
2658         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
2659         (cortex_a9_neon_vst3_vst4): Likewise.
2660         (cortex_a9_neon_vld1_vld2_lane): Likewise.
2661         (cortex_a9_neon_vld3_vld4_lane): Likewise.
2662         (cortex_a9_neon_vst1_vst2_lane): Likewise.
2663         (cortex_a9_neon_vst3_vst4_lane): Likewise.
2664         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
2665         (cortex_a9_neon_mcr): Likewise.
2666         (cortex_a9_neon_mcr_2_mcrr): Likewise.
2667         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
2668         (cortex_a9_fps): Likewise.
2669         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
2670         change.
2671         (cortex_m4_fmuls): Likewise.
2672         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
2673         change.
2674         (cortex_r4_mrc): Likewise.
2675         * config/arm/iterators.md: Update comment referring to neon_type.
2676         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
2677         (iwmmxt_movsi_insn): Likewise.
2678         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
2679         attribute change.
2680         (pj4_core_to_vfp): Likewise.
2681         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
2682         attribute change.
2683         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
2684         (vfp_fstore): Likewise.
2685         * doc/md.texi: Change references to neon_type to refer to type.
2687 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
2689         * tree.h (DECL_BUILT_IN): Fix typo in comment.
2691 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
2693         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
2694         lglobl if not weak.
2696 2013-09-04  Easwaran Raman  <eraman@google.com>
2698         PR middle-end/57370
2699         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
2700         (build_and_add_sum): Use it.
2701         (appears_later_in_bb): Simplify code.
2703 2013-09-04  Teresa Johnson  <tejohnson@google.com>
2705         * dumpfile.c (dump_finish): Don't close stderr/stdout.
2707 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
2709         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
2711 2013-09-04  Jan Hubicka  <jh@suse.cz>
2713         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
2714         * ipa-devirt.c: Include diganostic.h
2715         (odr_type_d): Add types and types_set.
2716         (hash_type_name): Work for types with vtables during LTO.
2717         (odr_hasher::remove): Fix comment; destroy types_set.
2718         (add_type_duplicate): New function,
2719         (get_odr_type): Use it.
2720         (dump_type_inheritance_graph): Dump type duplicates.
2721         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
2722         graph.
2723         * tree.c (types_same_for_odr): Give exact answers on types with
2724         virtual tables.
2726 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
2728         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
2729         explaining their differences.
2731 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
2733         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
2735 2013-09-03  Jeff Law  <law@redhat.com>
2737         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
2738         when not threading through a joiner block.  Pass joiner/no joiner
2739         state to register_jump_thread.
2740         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
2741         state from argument rather than implying on path length.
2742         Dump the entire jump thread path into debugging dump.
2743         * tree-flow.h (register_jump_thread): Update prototype.
2745 2013-08-29  Xinliang David Li  <davidxl@google.com>
2747         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
2748         Remove a trivial gcc_assert.
2750 2013-08-29  Xinliang David Li  <davidxl@google.com>
2752         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
2753         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
2754         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
2755         Delay base decl alignment adjustment.
2756         * tree-vectorizer.c (vect_destroy_datarefs): New function.
2757         * tree-vectorizer.h: New data structure.
2758         (set_dr_misalignment): New function.
2759         (dr_misalignment): Ditto.
2760         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
2761         (vectorizable_load): Ditto.
2762         (ensure_base_align): New function.
2763         (vectorize_loops): Add dbg_cnt support.
2764         (execute_vect_slp): Ditto.
2765         * dbgcnt.def: New debug counter.
2766         * Makefile: New dependency.
2768 2013-09-03  Meador Inge  <meadori@codesourcery.com>
2770         Revert:
2772         2013-08-30  Meador Inge  <meadori@codesourcery.com>
2774         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
2776 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
2778         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
2779         function descriptor.
2781 2013-09-03  Richard Biener  <rguenther@suse.de>
2783         * tree-affine.c (add_elt_to_tree): Fix association issue,
2784         avoid useless converts and make sure to always return a
2785         properly typed result.
2787 2013-09-03  Richard Biener  <rguenther@suse.de>
2789         PR middle-end/57656
2790         * fold-const.c (negate_expr_p): Fix division case.
2791         (negate_expr): Likewise.
2793 2013-09-03  Richard Biener  <rguenther@suse.de>
2795         PR lto/58285
2796         * tree-streamer-out.c: Include tm.h.
2797         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
2799 2013-09-03  Jan Hubicka  <jh@suse.cz>
2801         * tree-profile.c (tree_profiling): Cleanup CFG when done.
2803 2013-09-03  Alan Modra  <amodra@gmail.com>
2805         * config.gcc (powerpc*-*-linux*): Add support for little-endian
2806         multilibs to big-endian target and vice versa.
2807         * config/rs6000/t-linux64: Use := assignment on all vars.
2808         (MULTILIB_EXTRA_OPTS): Remove fPIC.
2809         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
2810         * config/rs6000/t-linux64le: New file.
2811         * config/rs6000/t-linux64bele: New file.
2812         * config/rs6000/t-linux64lebe: New file.
2814 2013-09-02  Jan Hubicka  <jh@suse.cz>
2816         * ipa-inline-transform.c (inline_transform): Do not
2817         optimize_inline_calls when not optimizing.
2819 2013-09-02  Jan Hubicka  <jh@suse.cz>
2821         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
2822         duplicated nodes for assembler names.
2823         * symtab.c (symtab_unregister_node): Do not attempt to unlink
2824         hard registers from assembler name hash.
2826 2013-09-02  Jan Hubicka  <jh@suse.cz>
2828         * ipa-split.c (execute_split_functions): Split externally visible
2829         functions called once.
2831 2013-09-02  Martin Jambor  <mjambor@suse.cz>
2833         PR ipa/58106
2834         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
2835         linked list.  When finding the correct duplicate, also consider also
2836         the caller in additon to its inlined_to node.
2838 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
2840         * config/aarch64/aarch64-simd-builtins.def
2841         (dup_lane_scalar): Remove.
2842         * config/aarch64/aarch64-simd.md
2843         (aarch64_simd_dup): Add 'w->w' alternative.
2844         (aarch64_dup_lane<mode>): Allow for VALL.
2845         (aarch64_dup_lane_scalar<mode>): Remove.
2846         (aarch64_dup_lane_<vswap_width_name><mode>): New.
2847         (aarch64_get_lane_signed<mode>): Add w->w altenative.
2848         (aarch64_get_lane_unsigned<mode>): Likewise.
2849         (aarch64_get_lane<mode>): Likewise.
2850         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
2851         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
2852         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
2853         (VCON): Change container of V2SF.
2854         (vswap_width_name): Likewise.
2855         * config/aarch64/arm_neon.h
2856         (__aarch64_vdup_lane_any): New.
2857         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
2858         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
2859         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
2861 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
2863         PR middle-end/56382
2864         * expr.c (emit_move_complex): Do not move complex FP values as parts if
2865         the source or the destination is a single hard register.
2867 2013-09-02  Richard Biener  <rguenther@suse.de>
2869         PR middle-end/57511
2870         * tree-scalar-evolution.c (instantiate_scev_name): Allow
2871         non-linear SCEVs.
2873 2013-09-02  Richard Biener  <rguenther@suse.de>
2875         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
2876         arithmetic to sizetype.
2878 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
2880         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
2881         Find auto-increment use both before and after candidate.
2883 2013-09-02  Marek Polacek  <polacek@redhat.com>
2885         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
2887 2013-09-01  Jan Hubicka  <jh@suse.cz>
2889         * Makefile.in: Add ipa-profile.o
2890         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
2891         * cgraph.c (struct cgraph_propagate_frequency_data,
2892         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
2893         ipa-profile.c; replace cgraph_ by ipa_ prefix.
2894         * cgraph.h (cgraph_propagate_frequency): Remove.
2895         * ipa-inline-analysis.c: Include ipa-utils.h;
2896         drop duplicated cfgloop.h.
2897         (inline_update_callee_summaries): Update.
2898         * ipa-profile.c: New file.
2899         * ipa-utils.h (ipa_propagate_frequency): Declare.
2900         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
2901         data-streamer.h, value-prof.h.
2902         (symtab_remove_unreachable_nodes): Update profile.
2903         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
2904         account_time_size, cmp_counts, dump_histogram,
2905         ipa_profile_generate_summary, ipa_profile_write_summary,
2906         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
2907         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
2908         Move to ipa-profile.c.
2910 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
2912         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
2914 2013-09-01  Jan Hubicka  <jh@suse.cz>
2916         * common.opt (fdevirtualize-speculatively): New function.
2917         * invoke.texi (fdevirtualize-speculatively): Document.
2918         * ipa-devirt.c: Include ipa-inline.h
2919         (likely_target_p): New function.
2920         (ipa_devirt): New function.
2921         (gate_ipa_devirt): New function.
2922         (pass_data_ipa_devirt): New static var.
2923         (pass_ipa_devirt): Likewise.
2924         (make_pass_ipa_devirt): New function.
2925         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
2926         (common_handle_option): Disable devirtualization when
2927         value range profiling is available.
2928         * passes.def (pass_ipa_devirt): Add.
2929         * timever.def (TV_IPA_DEVIRT): New timevar.
2930         * tree-pass.h (make_pass_ipa_devirt):
2932 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
2934         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
2935         include sanitize(undefined).
2937 2013-08-31  Diego Novillo  <dnovillo@google.com>
2939         * Makefile.in (TREE_CORE_H): Define.
2940         (TREE_H): Use.
2941         (GTFILES): Add tree-core.h.
2942         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
2943         size the array.
2944         * tree-core.h: New file.
2945         Move all data structures, enum, typedefs, global
2946         declarations and constants from ...
2947         * tree.h: ... here.
2949 2013-08-31  Jan Hubicka  <jh@suse.cz>
2951         * bulitins.c (expand_builtin): Do not early exit for gcov
2952         instrumented functions.
2954 2013-08-31  Marek Polacek  <polacek@redhat.com>
2956         * ubsan.c: Include tm_p.h.
2958 2013-08-31  Jan Hubicka  <jh@suse.cz>
2960         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
2961         warning.
2963         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
2964         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
2965         * tree-cfg.c (verify_gimple_label): ... here.
2966         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
2967         (ipa_merge_profiles): New function.
2968         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
2969         (lto_input_function_body): Likewise.
2970         * ipa-utils.h (ipa_merge_profiles): Declare.
2971         * lto-streamer.h (lto_input_function_body): Update prototype.
2972         (emit_label_in_global_context_p): Remove.
2973         * lto-symtab.c: Include ipa-utils.h
2974         (lto_cgraph_replace_node): Use ipa_merge_profiles.
2976 2013-08-31  Jan Hubicka  <jh@suse.cz>
2978         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
2980 2013-08-31  Jan Hubicka  <jh@suse.cz>
2982         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
2984 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
2986         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
2987         "cmp" RTX before signed_comparison_operator check to account
2988         for "code" changes.
2990 2013-08-30  Jan Hubicka  <jh@suse.cz>
2992         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
2993         (detect_type_change_1): Rename to ...
2994         (detect_type_change): ... this one; early return on non-polymorphic
2995         types.
2996         (detect_type_change_ssa): Add comp_type parameter; update
2997         use of detect_type_change.
2998         (compute_complex_assign_jump_func): Add param_type parameter;
2999         update use of detect_type_change_ssa.
3000         (compute_complex_ancestor_jump_func): Likewise.
3001         (ipa_get_callee_param_type): New function.
3002         (ipa_compute_jump_functions_for_edge): Compute parameter type;
3003         update calls to the jump function computation functions.
3005 2013-08-30  Teresa Johnson  <tejohnson@google.com>
3006             Steven Bosscher  <steven@gcc.gnu.org>
3008         * cfgrtl.c (fixup_new_cold_bb): New routine.
3009         (commit_edge_insertions): Invoke fixup_partitions.
3010         (find_partition_fixes): New routine.
3011         (fixup_partitions): Ditto.
3012         (verify_hot_cold_block_grouping): Update comments.
3013         (rtl_verify_edges): Invoke find_partition_fixes.
3014         (rtl_verify_bb_pointers): Update comments.
3015         (rtl_verify_bb_layout): Ditto.
3016         * basic-block.h (probably_never_executed_edge_p): Declare.
3017         (fixup_partitions): Ditto.
3018         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
3019         * bb-reorder.c (sanitize_hot_paths): New function.
3020         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
3021         sanitize_hot_paths.
3022         * predict.c (probably_never_executed_edge_p): New routine.
3023         * cfg.c (check_bb_profile): Add partition insanity warnings.
3025 2013-08-30  Meador Inge  <meadori@codesourcery.com>
3027         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
3029 2013-08-30  Marek Polacek  <polacek@redhat.com>
3031         * Makefile.in (ubsan.o): Add.
3032         (c-family/c-ubsan.o): Add.
3033         (builtins.o): Add ubsan.h dependency.
3034         * ubsan.h: New file.
3035         * ubsan.c: New file.
3036         * common.opt: Add -fsanitize=undefined option.
3037         (flag_sanitize): Add variable.
3038         (fsanitize=): Add option.  Add Driver.
3039         (fsanitize=thread): Remove option.
3040         (fsanitize=address): Likewise.
3041         (static-libubsan): New option.
3042         * doc/invoke.texi: Document the new flag and -static-libubsan.
3043         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
3044         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
3045         * builtin-attrs.def (ATTR_COLD): Define.
3046         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
3047         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
3048         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
3049         * flag-types.h (sanitize_code): New enum.
3050         * opts.c (common_handle_option): Parse command line arguments
3051         of -fsanitize=.  Add -fsanitize=unreachable option.
3052         * varasm.c (get_variable_section): Adjust.
3053         (assemble_noswitch_variable): Likewise.
3054         (assemble_variable): Likewise.
3055         (output_constant_def_contents): Likewise.
3056         (categorize_decl_for_section): Likewise.
3057         (place_block_symbol): Likewise.
3058         (output_object_block): Likewise.
3059         * builtins.def: Likewise.
3060         * toplev.c (compile_file): Likewise.
3061         (process_options): Likewise.
3062         * cppbuiltin.c: Likewise.
3063         * tsan.c (tsan_pass): Likewise.
3064         (tsan_gate): Likewise.
3065         (tsan_gate_O0): Likewise.
3066         * cfgexpand.c (partition_stack_vars): Likewise.
3067         (expand_stack_vars): Likewise.
3068         (defer_stack_allocation): Likewise.
3069         (expand_used_vars): Likewise.
3070         * cfgcleanup.c (old_insns_match_p): Likewise.
3071         * asan.c (asan_finish_file): Likewise.
3072         (asan_instrument): Likewise.
3073         (gate_asan): Likewise.
3074         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
3075         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
3076         (asan_global_struct): Use pointer_sized_int_node instead
3077         calling build_nonstandard_integer_type.
3078         (initialize_sanitizer_builtins): Likewise.
3079         (asan_finish_file): Likewise.
3080         * gcc.c: Document %{%:function(args):X}.
3081         (static_spec_functions): Add sanitize.
3082         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
3083         store funcval != NULL there.
3084         (do_spec_1): Adjust handle_spec_function caller.
3085         (handle_braces): Allow %:function(args) as condition.
3086         (sanitize_spec_function): New function.
3087         (ADD_STATIC_LIBUBSAN_LIBS): Define.
3088         (LIBUBSAN_SPEC): Likewise.
3089         (LIBUBSAN_EARLY_SPEC): Likewise.
3090         (SANITIZER_SPEC): Handle libubsan.
3091         (SANITIZER_EARLY_SPEC): Likewise.
3092         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
3093         instead of fsanitize=address.
3094         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
3095         instead of fsanitize=address*.
3096         * builtins.c: Include ubsan.h.
3097         (fold_builtin_0): Instrument __builtin_unreachable.
3098         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
3099         instead of flag_asan.
3100         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
3101         (pointer_sized_int_node): Define.
3102         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
3104 2013-08-30  Mike Stump  <mikestump@comcast.net>
3106         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
3107         with RE patterns.
3109 2013-08-29  Jan Hubicka  <jh@suse.cz>
3111         * cgraph.c (cgraph_function_body_availability): Handle weakref
3112         correctly.
3113         * passes.def: Remove pass_fixup_cfg.
3114         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
3115         track when we need to remove functions.
3116         (gate_ipa_inline): Execute inlining always; add comment why.
3117         (pass_data_ipa_inline): Remove TODO_remove_functions.
3118         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
3119         do not produce summaries.
3120         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
3121         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
3122         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
3123         constant pool and vtable.
3125 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
3127         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
3128         New arm_neon.h's internal macros to specify 64-bit constants.
3129         Avoid using stdint.h's macros.
3131 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
3133         * recog.c (verify_changes): Verify that changes[i].old is non-zero
3134         before applying REG_P.
3136 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
3138         PR tree-optimization/58277
3139         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
3140         after seeing too many stmts with vdef in between dombb and current
3141         bb, invalidate everything.
3143 2013-08-30  Richard Biener  <rguenther@suse.de>
3145         * fold-const.c (fold_single_bit_test): Fix overflow test.
3147 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
3149         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
3150         and which can live in a register, always retrieve the value on entry.
3151         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
3152         passed by invisible reference specially.
3153         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
3154         (vt_add_function_parameter): Correctly deal with a parameter passed by
3155         invisible reference.
3157 2013-08-30  Jan Hubicka  <jh@suse.cz>
3159         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
3161 2013-08-30  Richard Biener  <rguenther@suse.de>
3163         PR tree-optimization/58228
3164         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
3165         allow invariant loads in nested loop vectorization.
3167 2013-08-30  Richard Biener  <rguenther@suse.de>
3169         PR tree-optimization/58223
3170         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
3171         (has_anti_or_output_dependence): ... this and adjust to also
3172         look for output dependences.
3173         (mark_nodes_having_upstream_mem_writes): Adjust.
3174         (rdg_flag_uses): Likewise.
3176 2013-08-30  Richard Biener  <rguenther@suse.de>
3178         PR tree-optimization/58010
3179         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
3180         assert that we have a loop-closed PHI.
3182 2013-08-29  Jan Hubicka  <jh@suse.cz>
3184         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
3185         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
3186         * lto-section-in.c (lto_free_function_in_decl_state): New function.
3187         (lto_free_function_in_decl_state_for_node): New function.
3189 2013-08-29  Xinliang David Li  <davidxl@google.com>
3191         * loop-unroll.c (report_unroll_peel): Minor message change.
3192         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
3193         Emit alignment peeling message with default -fopt-info.
3194         (vect_loop_versioning): Emit loop version info message.
3195         * tree-vectorizer.c (vectorize_loops): Minor message change.
3196         (execute_vect_slp): Ditto.
3198 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
3200         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
3201         of the clone from the DECL_NAME of the original function.
3203 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
3205         * passes.c (register_pass): Add overload.
3206         * tree-pass.h (register_pass): Forward declare it.  Add comment.
3208 2013-08-29  Jan Hubicka  <jh@suse.cz>
3210         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
3211         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
3212         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
3213         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
3214         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
3215         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
3216         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
3217         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
3219 2013-08-29  Teresa Johnson  <tejohnson@google.com>
3221         * dumpfile.c (dump_loc): Output column number.
3222         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
3223         * doc/invoke.texi: Document optall -fopt-info flag.
3224         * profile.c (read_profile_edge_counts): Use new dump framework.
3225         (compute_branch_probabilities): Ditto.
3226         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
3227         when pass not in any opt group.
3228         * pass_manager.h (pass_manager::get_pass_profile): New method.
3229         * value-prof.c (check_counter): Use new dump framework.
3230         (check_ic_target): Ditto.
3231         * coverage.c (get_coverage_counts): Ditto.
3232         (coverage_init): Setup new dump framework.
3234 2013-08-29  Richard Biener  <rguenther@suse.de>
3236         PR tree-optimization/58246
3237         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
3238         handle the dominance check inside a basic-block.
3240 2013-08-29  Richard Biener  <rguenther@suse.de>
3242         PR middle-end/57287
3243         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
3244         of default defs that appear in abnormal PHI nodes.
3246 2013-08-29  Richard Biener  <rguenther@suse.de>
3248         PR tree-optimization/57685
3249         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
3250         single-use operands to avoid exponential complexity.
3252 2013-08-28  Dehao Chen  <dehao@google.com>
3254         * ipa-inline.c (edge_badness): Fix integer underflow.
3256 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
3258         * gtm-builtins.def (_ITM_free): Declare leaf.
3260 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
3262         PR target/58067
3263         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
3264         (*tls_local_dynamic_base_64_largepic): Likewise.
3265         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
3266         Remove predicate from call operand.
3267         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
3268         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
3270 2013-08-28  Jeff Law  <law@redhat.com>
3272         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
3273         checks for the number of predecessors and successors allowed.
3274         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
3275         which require copying a joiner block if there is a request which
3276         is a subpath that requires no joiner block copying.
3278 2013-08-28  Jan Hubicka  <jh@suse.cz>
3280         * lto-streamer-out.c (DFS_write_tree_body): Drop
3281         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
3282         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
3283         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
3284         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
3285         Do not read DECL_ERROR_ISSUED.
3286         (unpack_ts_decl_with_vis_value_fields): Do not read
3287         DECL_DEFER_OUTPUT.
3288         (lto_input_ts_binfo_tree_pointers): Do not read
3289         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
3290         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
3291         write DECL_ERROR_ISSUED..
3292         (pack_ts_decl_with_vis_value_fields): Do not write
3293         DECL_DEFER_OUTPUT.
3294         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
3295         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
3296         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
3297         * tree.h (tree_decl_common): Update comment.
3298         (DECL_ERROR_ISSUED): Remove.
3300 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
3302         PR middle-end/58257
3303         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
3305 2013-08-28  Jan Hubicka  <jh@suse.cz>
3307         * builtins.def (free): Declare leaf.
3309 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
3311         * gdbhooks.py: New.
3312         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
3313         * configure: Regenerate.
3315 2013-08-27  Martin Jambor  <mjambor@suse.cz>
3317         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
3318         (ipa_ancestor_jf_data): Likewise.
3319         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
3320         (ipa_get_jf_pass_through_type_preserved): New function.
3321         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
3322         (ipa_get_jf_ancestor_type_preserved): New function.
3323         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
3324         (ipa_get_jf_ancestor_result): Likewise.
3325         (propagate_vals_accross_pass_through): Use
3326         ipa_get_jf_pass_through_result to do all the value mappings.
3327         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
3328         type_preserved flag.
3329         (ipa_set_jf_cst_copy): New function.
3330         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
3331         (ipa_set_jf_arith_pass_through): Likewise.
3332         (ipa_set_ancestor_jf): Likewise.
3333         (compute_complex_assign_jump_func): Set type_preserved instead of
3334         punting.
3335         (ipa_compute_jump_functions_for_edge): Likewise.
3336         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
3337         (update_jump_functions_after_inlining): Update type_preserved.
3338         Explicitely create jump functions when combining one with pass_through.
3339         (ipa_write_jump_function): Stream the type_preserved flags.
3340         (ipa_read_jump_function): Likewise.
3342 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
3343             Aldy Hernandez  <aldyh@redhat.com>
3345         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
3346         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
3347         * function.h (struct function): Add has_force_vect_loops and
3348         has_simduid_loops.
3349         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
3350         * gimple.c (gimple_build_omp_critical): Add KIND argument and
3351         handle it.
3352         * gimple.def: Update CLAUSES comments.
3353         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
3354         (gimple_build_omp_for): Add argument to prototype.
3355         (gimple_omp_for_kind): New.
3356         (gimple_omp_for_set_kind): New.
3357         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
3358         GOVD_DATA_SHARE_CLASS.
3359         (enum omp_region_type): Add ORT_SIMD.
3360         (gimple_add_tmp_var): Handle ORT_SIMD.
3361         (gimplify_var_or_parm_decl): Same.
3362         (is_gimple_stmt): Same.
3363         (omp_firstprivatize_variable): Same.
3364         (omp_add_variable): Only use splay_tree_insert if lookup failed.
3365         (omp_notice_variable): Handle ORT_SIMD.
3366         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
3367         (omp_check_private): Handle ORT_SIMD.
3368         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
3369         OMP_CLAUSE_SAFELEN.
3370         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
3371         Handle OMP_CLAUSE_LASTPRIVATE.
3372         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
3373         OMP_CLAUSE_SAFELEN.
3374         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
3375         (gimplify_expr): Handle OMP_SIMD.
3376         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
3377         (expand_GOMP_SIMD_VF): New.
3378         (expand_GOMP_SIMD_LAST_LANE): New.
3379         * internal-fn.def (GOMP_SIMD_LANE): New.
3380         (GOMP_SIMD_VF): New.
3381         (GOMP_SIMD_LAST_LANE): New.
3382         * omp-low.c: Include target.h.
3383         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
3384         OMP_CLAUSE_SAFELEN.
3385         (check_omp_nesting_restrictions): Same.
3386         (omp_max_vf): New.
3387         (lower_rec_simd_input_clauses): New.
3388         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
3389         OMP_CLAUSE_LINEAR.
3390         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
3391         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
3392         (expand_omp_build_assign): New.
3393         (expand_omp_for_init_counts): New.
3394         (expand_omp_for_init_vars): New.
3395         (extract_omp_for_update_vars): New.
3396         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
3397         and rewrite accordingly.
3398         (expand_omp_simd): New.
3399         (expand_omp_for): Use expand_omp_simd.
3400         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
3401         (lower_omp_for): Do not lower the body.
3402         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
3403         in their own loops.
3404         * tree-flow.h (find_omp_clause): Remove prototype.
3405         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
3406         forcing vectorization of the loop, or if flag_tree_vectorize.
3407         (gate_tree_if_conversion): Similarly.
3408         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
3409         gimple_build_omp_for.
3410         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
3411         * tree-parloops (create_parallel_loop): Pass kind argument to
3412         gimple_build_omp_for.
3413         * tree-pretty-print.c (dump_omp_clause): Add cases for
3414         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
3415         OMP_CLAUSE__SIMDUID_.
3416         (dump_generic_node): Handle OMP_SIMD.
3417         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
3418         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
3419         unroll OMP_SIMD loops here.
3420         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
3421         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
3422         loop->safelen.
3423         (vect_analyze_data_refs): Handle simd loops.
3424         * tree-vect-loop.c (vectorizable_live_operation): Handle
3425         IFN_GOMP_SIMD*.
3426         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
3427         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
3428         (vectorizable_load): Same.
3429         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
3430         (struct simduid_to_vf): New.
3431         (simduid_to_vf::hash): New.
3432         (simduid_to-vf::equal): New.
3433         (struct simd_array_to_simduid): New.
3434         (simd_array_to_simduid::hash): New.
3435         (simd_array_to_simduid::equal): New.
3436         (adjust_simduid_builtins): New.
3437         (struct note_simd_array_uses_struct): New.
3438         (note_simd_array_uses_cb): New.
3439         (note_simd_array_uses): New.
3440         (vectorize_loops): Handle simd hints and adjust simd builtins
3441         accordingly.
3442         * tree-vectorizer.h (struct _stmt_vec_info): Add
3443         simd_lane_access_p field.
3444         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
3445         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
3446         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
3447         (omp_clause_code_name): Same.
3448         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
3449         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
3450         * tree.def (OMP_SIMD): New entry.
3451         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
3452         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
3453         (OMP_CLAUSE_DECL): Adjust range for new clauses.
3454         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
3455         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
3456         (OMP_CLAUSE_LINEAR_STEP): New.
3457         (OMP_CLAUSE_SAFELEN_EXPR): New.
3458         (OMP_CLAUSE__SIMDUID__DECL): New.
3459         (find_omp_clause): New prototype.
3461 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
3463         * config/i386/driver-i386.c (host_detect_local_cpu): Update
3464         Haswell processor detection.
3466 2013-08-27  Christian Widmer  <shadow@umbrox.de>
3468         PR target/57927
3469         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
3470         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
3471         AVX2 capable processors.
3473 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
3475         * config/aarch64/arm_neon.h: Replace all inline asm implementations
3476         of vget_low_* with implementations in terms of other intrinsics.
3478 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
3480         PR middle-end/57219
3481         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
3482         values to -1, 0 and 1.
3484 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
3486         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
3487         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
3488         (<optab><mode>3_insn): Remove.
3489         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
3490         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
3491         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
3492         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
3493         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
3494         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
3495         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
3496         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
3497         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
3498         (ror<mode>3_insn): Likewise.
3499         * config/aarch64/predicates.md (aarch64_simd_register): New.
3501 2013-08-27  Richard Biener  <rguenther@suse.de>
3503         PR tree-optimization/57521
3504         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
3505         one edge is non-critical.
3506         (find_phi_replacement_condition): Make sure to use a non-critical
3507         edge.  Cleanup and remove old bug workarounds.
3508         (bb_postdominates_preds): Remove.
3509         (if_convertible_loop_p_1): Do not compute post-dominators.
3510         (combine_blocks): Do not free post-dominators.
3511         (main_tree_if_conversion): Likewise.
3512         (pass_data_if_conversion): Add TODO_verify_ssa.
3514 2013-08-27  DJ Delorie  <dj@redhat.com>
3516         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
3518 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
3520         * function.c (assign_parm_find_data_types): Set passed_mode and
3521         nominal_mode to the TYPE_MODE of nominal_type for the built
3522         pointer type in case of the struct-pass-by-reference.
3524 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
3526         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
3527         (UINT16_TYPE): Change default to "unsigned int".
3529         * config/avr/avr.opt (mfract-convert-truncate): New option.
3530         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
3531         is set, round negative fractional integers according to n1169
3532         when converting to integer types.
3534 2013-08-26  Jan Hubicka  <jh@suse.cz>
3536         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
3537         methods can not be called indirectly when their address is not taken.
3539 2013-08-26  Jan Hubicka  <jh@suse.cz>
3541         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
3542         ctor_for_folding.
3544 2013-08-26  Jan Hubicka  <jh@suse.cz>
3546         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
3547         can be unshared.
3549 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
3551         * reload.c (find_valid_class): Allow classes that do not include
3552         FIRST_PSEUDO_REGISTER - 1.
3554 2013-08-26  Jan Hubicka  <jh@suse.cz>
3556         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
3557         fix edge count/frequency when speculation failed; fix type check
3558         for the direct call.
3560 2013-08-26  Jan Hubicka  <jh@suse.cz>
3562         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
3564 2013-08-26  Jan Hubicka  <jh@suse.cz>
3566         * ipa-inline-transform.c (inline_transform): Be ready for basic block
3567         to be changed by edge redirection.
3569 2013-08-26  Jan Hubicka  <jh@suse.cz>
3571         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
3572         formating; add sanity check.
3573         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
3574         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
3575         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
3576         (resolve_noninline_speculation): Update callee keys, too.
3578 2013-08-26  Jan Hubicka  <jh@suse.cz>
3580         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
3581         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
3583 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
3585         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
3586         into proper place.
3588 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
3590         * config/i386/i386.c (ix86_debug_options): Remove prototype.
3591         (x86_64_elf_select_section): Ditto.
3592         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
3593         arguments.
3594         (ix86_pass_by_reference): Ditto.
3595         (output_set_got): Ditto.
3596         (ix86_unary_operator_ok): Ditto.
3597         (ix86_expand_builtin): Ditto.
3599 2013-08-23  Jan Hubicka  <jh@suse.cz>
3601         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
3603 2013-08-23  Jan Hubicka  <jh@suse.cz>
3605         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
3606         (tree_decl_with_vis): Add FINAL field.
3608 2013-08-23  Jeff Law  <law@redhat.com>
3610         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
3611         the debugging dump when the expression is fully redundant.
3613 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3615         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
3616         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
3617         * pretty-print.c (pp_formatted_text_data): Likewise.
3618         (pp_write_text_to_stream): Likewise.
3619         (pp_write_text_as_dot_label_to_stream): Likewise.
3620         (pp_append_r): Likewise.
3621         (pp_format): Likewise.
3622         (pp_flush): Likewise.
3623         (pp_clear_output_area): Likewise.
3624         (pp_append_text): Likewise.
3625         (pp_formatted_text): Likewise.
3626         (pp_remaining_character_count_for_line): Likewise.
3627         (pp_newline): Likewise.
3628         (pp_character): Likewise.
3629         (output_buffer::~output_buffer): Define.
3630         (pretty_printer::~pretty_printer): Destruct output buffer.
3631         * pretty-print.h (output_buffer::~output_buffer): Declare.
3632         (pretty_printer::~pretty_printer): Declare virtual.
3634 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
3636         PR other/57324
3637         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
3638         HOST_WIDE_INT_M1U): New macros.
3639         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
3640         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
3641         unsigned -1 for lshift.
3642         * cse.c (cse_insn): Likewise.
3643         * double-int.c (rshift_double, lshift_double): Likewise.
3644         * builtins.c (fold_builtin_bitop): Likewise.
3645         * combine.c (force_to_mode): Likewise.
3646         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
3647         * simplify-rtx.c (simplify_const_unary_operation,
3648         simplify_const_binary_operation): Likewise.
3649         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
3650         check_va_list_escapes): Likewise.
3651         * rtlanal.c (nonzero_bits1): Likewise.
3652         * expmed.c (expand_smod_pow2): Likewise.
3653         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
3655 2013-08-23  Jan Hubicka  <jh@suse.cz>
3657         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
3658         as having address taken.
3660 2013-08-23  Jan Hubicka  <jh@suse.cz>
3662         * ipa-utils.h (method_class_type): Declare.
3663         * ipa-devirt.c (method_class_type): Export.
3665         * cgraphunit.c (analyze_functions): Do basic devirtualization;
3666         do not walk base classes of anonymous types.
3668 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
3670         PR rtl-optimization/58220
3671         PR regression/58221
3672         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
3673         handle SEQUENCE insns properly.
3675 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3677         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
3678         definition.
3679         (pp_newline_and_indent): Likewise.
3680         (pp_separate_with): Likewise.
3681         * pretty-print.c (pp_newline_and_flush): Define.
3682         (pp_newline_and_indent): Likewise.
3683         (pp_separate_with): Likewise.
3685 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
3687         PR target/58218
3688         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
3689         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
3691 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
3693         * config/i386/predicates.md (ext_sse_reg_operand): New.
3694         * config/i386/i386.md (*movti_internal): Use
3695         predicate to determine if EVEX is needed.
3696         (*movsi_internal): Ditto.
3697         (*movdf_internal): Ditto.
3698         (*movsf_internal): Ditto.
3699         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
3701 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
3703         PR tree-optimization/58209
3704         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
3705         (find_tail_calls): Give up for pointer result types if m is non-NULL.
3706         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
3707         emit POINTER_PLUS_EXPR.
3708         (create_tailcall_accumulator): For pointer result type accumulate in
3709         sizetype type.
3711 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
3713         * configure.ac: Add backslashes missing from the last change.
3714         * configure: Regenerate.
3716 2013-08-22  Jan Hubicka  <jh@susue.cz>
3718         * ipa.c (function_and_variable_visibility): First remember function
3719         was global and then make it local.
3721 2013-08-22  Julian Brown  <julian@codesourcery.com>
3723         * configure.ac: Add aarch64 to list of arches which use "nop" in
3724         debug_line test.
3725         * configure: Regenerate.
3727 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3729         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
3730         gnu_libc_has_function.
3731         * config/s390/tpf.h: Likewise.
3733 2013-08-22  Jan Hubicka  <jh@susue.cz>
3735         * timevar.c (validate_phases): Add cast.
3737 2013-08-22  Jan Hubicka  <jh@susue.cz>
3739         * timevar.c (validate_phases): Use size_t for memory.
3740         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
3742 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3744         * pretty-print.h (output_buffer::output_buffer): Declare.
3745         (pretty_printer::pretty_printer): Likewise.
3746         (pp_construct): Remove.
3747         * pretty-print.c (output_buffer::output_buffer): Define.
3748         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
3749         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
3750         (print_gimple_expr): Likewise.
3751         (print_gimple_seq): Likewise.
3752         (gimple_dump_bb): Likewise.
3753         * sched-vis.c (dump_value_slim): Likewise.
3754         (dump_insn_slim): Likewise.
3755         (dump_rtl_slim): Likewise.
3756         (str_pattern_slim): Likewise.
3757         * tree-mudflap.c (mf_varname_tree): Likewise.
3758         * graph.c (print_graph_cfg): Likewise.
3759         (start_graph_dump): Likewise.
3760         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
3761         placement-new.
3762         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
3763         pretty printer initialization.
3764         * coretypes.h (diagnostic_context): Remove superflous type alias
3765         declaration.
3766         (pretty_printer): Likewise.  Declare directly as a class.
3767         (pretty_print_info): Remove declaration as class.
3768         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
3769         and pp_clear_output_area.
3770         (asan_add_global): Likewise.
3772 2013-08-22  Jan Hubicka  <jh@suse.cz>
3774         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
3775         * ipa-utils.h (update_type_inheritance_graph): Declare.
3776         (possible_polymorphic_call_target_p): Declare.
3777         (possible_polymorphic_call_target_p): New.
3778         * ipa-devirt.c: Update toplevel comments.
3779         (cached_polymorphic_call_targets): Move up.
3780         (odr_type_d): Move ID down.
3781         (polymorphic_type_binfo_p): Update comment.
3782         (odr_hasher::remove): Likewise;
3783         (get_odr_type): Set anonymous_namespace.
3784         (dump_odr_type): Dump it.
3785         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
3786         (maybe_record_node): Record node in cached_polymorphic_call_targets.
3787         (record_binfo): Add comment.
3788         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
3789         built.
3790         (devirt_node_removal_hook): Do not iCE when cache is freed.
3791         (possible_polymorphic_call_target_p): New predicate.
3792         (update_type_inheritance_graph): New function.
3794 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3795             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3796             Sergey Lega  <sergey.s.lega@intel.com>
3797             Anna Tikhonova  <anna.tikhonova@intel.com>
3798             Ilya Tocar  <ilya.tocar@intel.com>
3799             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3800             Ilya Verbin  <ilya.verbin@intel.com>
3801             Kirill Yukhin  <kirill.yukhin@intel.com>
3802             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3804         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
3805         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
3806         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
3807         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
3808         (OPTION_MASK_ISA_AVX2_UNSET): Update.
3809         (OPTION_MASK_ISA_AVX512F_UNSET): New.
3810         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
3811         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
3812         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
3813         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
3814         OPT_mavx512pf, OPT_mavx512er cases.
3815         * config/i386/constraints.md (v): New constraint.
3816         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
3817         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
3818         (bit_AVX512CD): New.
3819         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
3820         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
3821         * config/i386/i386-c.c (ix86_target_macros_internal):
3822         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
3823         __AVX512PF__.
3824         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
3825         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
3826         * config/i386/i386.c (regclass_map, dbx_register_map)
3827         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
3828         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
3829         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
3830         -mavx512pf options.
3831         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
3832         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
3833         -mavx512pf options.  Fix formatting.
3834         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
3835         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
3836         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
3837         -mavx512cd, -mavx512pf options.
3838         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
3839         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
3840         (ix86_preferred_output_reload_class): Replace SSE_REGS with
3841         ALL_SSE_REGS.
3842         (ix86_hard_regno_mode_ok): Support 512-bit registers.
3843         (ix86_set_reg_reg_cost): Ditto.
3844         (x86_order_regs_for_local_alloc): Ditto.
3845         (MAX_VECT_LEN): Extend to 64-byte.
3846         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
3847         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
3848         (TARGET_AVX512ER, TARGET_AVX512CD): New.
3849         (BIGGEST_ALIGNMENT): Extend to 512-bits.
3850         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
3851         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
3852         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
3853         (SSE_REG_MODE_P): Support new modes.
3854         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
3855         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
3856         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
3857         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
3858         (REG_CLASS_CONTENTS): Add new registers.
3859         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
3860         (EXT_REX_SSE_REGNO_P): New.
3861         (HI_REGISTER_NAMES): Add new registers.
3862         * config/i386/i386.md: Define constants for new registers.
3863         (mode): Add new 512-bit modes.
3864         (prefix): Support evex prefix.
3865         (isa): Support avx512f, noavx512f, fma_avx512f.
3866         (ssemodesuffix): Add new 512-bit modes.
3867         (movxi): New.
3868         (*movxi_internal_avx512f): Ditto.
3869         (*movdi_internal): Replace constraint "x" with the new constraint "v".
3870         Support MODE_XI.
3871         (*movsi_internal): Likewise.
3872         (*movdf_internal): Likewise.
3873         (*movsf_internal): Likewise.
3874         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
3875         (<code><mode>3): Likewise.
3876         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
3877         New.
3878         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
3879         with the new constraint "v".
3880         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
3881         modes.
3882         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
3883         with the new constraint "v".
3884         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
3885         (<sse2>_storedqu<avxsizesuffix>): Likewise.
3886         (*<plusminus_insn><mode>3): Likewise.
3887         (<sse>_vm<plusminus_insn><mode>3): Likewise.
3888         (*mul<mode>3): Likewise.
3889         (<sse>_vmmul<mode>3): Likewise.
3890         (<sse>_div<mode>3): Likewise.
3891         (<sse>_vmdiv<mode>3): Likewise.
3892         (<sse>_sqrt<mode>2): Likewise.
3893         (<sse>_vmsqrt<mode>2): Likewise.
3894         (*<code><mode>3_finite): Likewise.
3895         (*<code><mode>3) <smaxmin>: Likewise.
3896         (<sse>_vm<code><mode>3): Likewise.
3897         (*<code><mode>3) <any_logic>: Likewise.
3898         (*fma_fmadd_<mode>): Likewise.
3899         (*fma_fmsub_<mode>): Likewise.
3900         (*fma_fnmadd_<mode>): Likewise.
3901         (*fma_fnmsub_<mode>): Likewise.
3902         (*fma_fmaddsub_<mode>): Likewise.
3903         (*fma_fmsubadd_<mode>): Likewise.
3904         (*fmai_fmadd_<mode>): Likewise.
3905         (*fmai_fmsub_<mode>): Likewise.
3906         (*fmai_fnmadd_<mode>): Likewise.
3907         (*fmai_fnmsub_<mode>): Likewise.
3908         (sse_cvtsi2ss): Likewise.
3909         (sse_cvtsi2ssq): Likewise.
3910         (sse_cvtss2si): Likewise.
3911         (sse_cvtss2si_2): Likewise.
3912         (sse_cvtss2siq): Likewise.
3913         (sse_cvtss2siq_2): Likewise.
3914         (sse_cvttss2si): Likewise.
3915         (sse_cvtss2siq_2): Likewise.
3916         (float<sseintvecmodelower><mode>2): Likewise.
3917         (sse2_cvtsd2si_2): Likewise.
3918         (sse2_cvtsd2siq_2): Likewise.
3919         (*<plusminus_insn><mode>3): Likewise.
3920         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
3921         (*<sse4_1_avx2>_mul<mode>3): Likewise.
3922         (ashr<mode>3): Likewise.
3923         (<shift_insn><mode>3): Likewise.
3924         (avx2_<code><mode>3): Likewise.
3925         (*avx2_<code><mode>3): Likewise.
3926         (*andnot<mode>3): Likewise.
3927         (*<code><mode>3) <any_logic>: Likewise.
3928         (abs<mode>2): Likewise.
3929         (avx2_permvar<mode>): Likewise.
3930         (avx2_perm<mode>_1): Likewise.
3931         (*avx_vpermilp<mode>): Likewise.
3932         (avx_vpermilvar<mode>3): Likewise.
3933         (avx2_ashrv<mode>): Likewise.
3934         (avx2_<shift_insn>v<mode>): Likewise.
3935         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
3936         -mavx512cd.
3937         * doc/rtl.texi: Document XImode.
3939 2013-08-21  Jeff Law  <law@redhat.com>
3941         * tree-flow.h (register_jump_thread): Pass vector of edges
3942         instead of each important edge.
3943         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
3944         thread path into a vector and pass that to register_jump_thread.
3945         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
3946         passed in edge vector to the current 3-edge form.
3948         Revert:
3949         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
3951         * dce.c (fini_dce): Call df_analyze again just in case
3952         delete_unmarked_insns removed anything.
3954 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
3956         * reload.h (struct reg_equivs): Rename to ..
3957         (struct reg_equivs_s): .. this.
3959 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
3961         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
3963 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3965         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
3967 2013-08-21  Jeff Law  <law@redhat.com>
3969         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
3970         simplify assignments too.  If the RHS collapses to a singleton
3971         range, then return the value for the range.
3973 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
3975         * config/i386/sse.md (V16): Rename to...
3976         (VMOVE): this.
3977         (mov<mode>): Update iterator name.
3978         (*mov<mode>_internal): Ditto.
3979         (push<mode>1): Ditto.
3980         (movmisalign<mode>): Ditto.
3982 2013-08-20  Jan Hubicka  <jh@suse.cz>
3984         PR bootstrap/58186
3985         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
3986         entry for direct edges.
3987         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
3989 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3991         Revert my last two changes, r201865 and r201864:
3993         Revert r201865:
3994         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
3996         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
3997         instances can own GC refs.
3999         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
4000         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
4001         (gcc::context::gt_pch_nx): Likewise.
4002         (gcc::context::gt_pch_nx):  Likewise.
4003         * ggc.h (gt_ggc_mx <T>): New.
4004         (gt_pch_nx_with_op <T>): New.
4005         (gt_pch_nx <T>): New.
4006         * passes.c (opt_pass::gt_ggc_mx): New.
4007         (opt_pass::gt_pch_nx): New.
4008         (opt_pass::gt_pch_nx_with_op): New.
4009         (pass_manager::gt_ggc_mx): New.
4010         (pass_manager::gt_pch_nx): New.
4011         (pass_manager::gt_pch_nx_with_op): New.
4012         (pass_manager::operator new): Use
4013         ggc_internal_cleared_alloc_stat rather than xcalloc.
4014         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
4015         (pass_manager::gt_ggc_mx): New.
4016         (pass_manager::gt_pch_nx): New.
4017         (pass_manager::gt_pch_nx_with_op): New.
4018         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
4019         (opt_pass::operator new): New.
4020         (opt_pass::gt_ggc_mx): New.
4021         (opt_pass::gt_pch_nx): New.
4022         (opt_pass::gt_pch_nx_with_op): New.
4024         Revert r201864:
4025         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
4027         * Makefile.in (GTFILES): Add context.h.
4028         * context.c (gcc::context::operator new): New.
4029         (gcc::context::gt_ggc_mx): New.
4030         (gcc::context::gt_pch_nx): New.
4031         (gcc::context::gt_pch_nx): New.
4032         * context.h (gcc::context): Add GTY((user)) marking.
4033         (gcc::context::operator new): New.
4034         (gcc::context::gt_ggc_mx): New.
4035         (gcc::context::gt_pch_nx): New.
4036         (gcc::context::gt_pch_nx): New.
4037         (g): Add GTY marking.
4038         (gt_ggc_mx (gcc::context *)): New.
4039         (gt_pch_nx (gcc::context *)): New.
4040         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
4041         void *cookie)): New.
4042         * gengtype.c (open_base_files) <ifiles>: Add context.h.
4044 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
4046         * dce.c (fini_dce): Call df_analyze again just in case
4047         delete_unmarked_insns removed anything.
4049 2013-08-20  Teresa Johnson  <tejohnson@google.com>
4051         PR rtl-optimizations/57451
4052         * final.c (reemit_insn_block_notes): Prevent lexical blocks
4053         from crossing split section boundaries.
4055 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
4057         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
4058         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
4059         with MULTLIB_DEFAULTS.
4061 2013-08-20  Nick Clifton  <nickc@redhat.com>
4063         * target.def (narrow_volatile_bitfield): Note that the default
4064         value is false, not !TARGET_STRICT_ALIGN.
4065         * doc/tm.texi: Regenerate.
4067 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
4069         Fix LIB_SPEC for systems without libpthread.
4071         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
4072         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
4073         for Android.
4074         * config/i386/linux-common.h: Likewise.
4075         * config/mips/linux-common.h: Likewise.
4077 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
4079         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
4080         checks.
4082 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
4084         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
4085         instances can own GC refs.
4087         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
4088         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
4089         (gcc::context::gt_pch_nx): Likewise.
4090         (gcc::context::gt_pch_nx):  Likewise.
4091         * ggc.h (gt_ggc_mx <T>): New.
4092         (gt_pch_nx_with_op <T>): New.
4093         (gt_pch_nx <T>): New.
4094         * passes.c (opt_pass::gt_ggc_mx): New.
4095         (opt_pass::gt_pch_nx): New.
4096         (opt_pass::gt_pch_nx_with_op): New.
4097         (pass_manager::gt_ggc_mx): New.
4098         (pass_manager::gt_pch_nx): New.
4099         (pass_manager::gt_pch_nx_with_op): New.
4100         (pass_manager::operator new): Use
4101         ggc_internal_cleared_alloc_stat rather than xcalloc.
4102         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
4103         (pass_manager::gt_ggc_mx): New.
4104         (pass_manager::gt_pch_nx): New.
4105         (pass_manager::gt_pch_nx_with_op): New.
4106         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
4107         (opt_pass::operator new): New.
4108         (opt_pass::gt_ggc_mx): New.
4109         (opt_pass::gt_pch_nx): New.
4110         (opt_pass::gt_pch_nx_with_op): New.
4112 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
4114         * Makefile.in (GTFILES): Add context.h.
4115         * context.c (gcc::context::operator new): New.
4116         (gcc::context::gt_ggc_mx): New.
4117         (gcc::context::gt_pch_nx): New.
4118         (gcc::context::gt_pch_nx): New.
4119         * context.h (gcc::context): Add GTY((user)) marking.
4120         (gcc::context::operator new): New.
4121         (gcc::context::gt_ggc_mx): New.
4122         (gcc::context::gt_pch_nx): New.
4123         (gcc::context::gt_pch_nx): New.
4124         (g): Add GTY marking.
4125         (gt_ggc_mx (gcc::context *)): New.
4126         (gt_pch_nx (gcc::context *)): New.
4127         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
4128         void *cookie)): New.
4129         * gengtype.c (open_base_files) <ifiles>: Add context.h.
4131 2013-08-20  Alan Modra  <amodra@gmail.com>
4133         PR target/57865
4134         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
4135         (rs6000_emit_epilogue): Likewise.
4137 2013-08-19  Dehao Chen  <dehao@google.com>
4139         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
4141 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
4142             Jakub Jelinek  <jakub@redhat.com>
4144         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
4145         (BUILT_IN_FABSD64): Likewise.
4146         (BUILT_IN_FABSD128): Likewise.
4147         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
4148         (fold_builtin_1): Likewise.
4149         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
4150         destination and source operands.
4151         (*abstd2_fpr): Likewise.
4152         (*nabstd2_fpr): Likewise.
4154 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
4156         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
4157         JUMP_P and INSN_P.
4159 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
4161         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
4162         incomplete.
4164 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4166         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
4167         * builtins.c (default_libc_has_function): New.
4168         (gnu_libc_has_function): Ditto.
4169         (no_c99_libc_has_function): Ditto.
4170         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
4171         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
4172         (fold_builtin_sincos): Likewise.
4173         (fold_builtin_cexp): Likewise.
4174         * builtins.def (DEF_C94_BUILTIN): Likewise.
4175         (DEF_C99_BUILTIN): Likewise.
4176         (DEF_C99_C90RES_BUILTIN): Likewise.
4177         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
4178         definitions to using this define.
4179         * config/darwin-protos.h (darwin_libc_has_function): New.
4180         * config/darwin.c (darwin_libc_has_function): Ditto.
4181         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
4182         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
4183         * config/darwin.h: Ditto.
4184         * config/elfos.h: Ditto.
4185         * config/freebsd.h: Ditto.
4186         * config/i386/cygming.h: Ditto.
4187         * config/i386/djgpp.h: Ditto.
4188         * config/i386/i386-interix.h: Ditto.
4189         * config/microblaze/microblaze.h: Ditto.
4190         * config/mmix/mmix.h: Ditto.
4191         * config/gnu-user.h: Ditto.
4192         * config/ia64/hpux.h: Ditto.
4193         * config/pa/pa-hpux.h: Ditto.
4194         * config/pdp11/pdp11.h: Ditto.
4195         * config/picochip/picochip.h: Ditto.
4196         * config/linux.h: Ditto.
4197         * config/netbsd.h: Ditto.
4198         * config/openbsd.h: Ditto.
4199         * config/rs6000/aix43.h: Ditto.
4200         * config/rs6000/aix51.h: Ditto.
4201         * config/rs6000/aix52.h: Ditto.
4202         * config/rs6000/aix53.h: Ditto.
4203         * config/rs6000/aix61.h: Ditto.
4204         * config/rs6000/darwin.h: Ditto.
4205         * config/rs6000/linux.h: Ditto.
4206         * config/rs6000/linux64.h: Ditto.
4207         * config/s390/tpf.h: Ditto.
4208         * config/sol2-10.h: Ditto.
4209         * config/sol2.h: Ditto.
4210         * config/vms/vms.h: Ditto.
4211         * config/vxworks.h: Ditto.
4212         * config/linux-android.c (linux_android_libc_has_function):
4213         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
4214         * config/linux-protos.h (linux_android_libc_has_function):
4215         New declaration.
4216         * config/i386/i386.c (ix86_libc_has_function): New.
4217         * config/i386/i386-protos.h
4218         (ix86_libc_has_function): New declaration.
4219         * config/i386/i386.md
4220         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
4221         ("isinf<mode>2): Likewise.
4222         * convert.c (convert_to_integer): Using new target hook
4223         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
4224         TARGET_C99_FUNCTIONS.
4225         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
4226         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
4227         * coretypes.h (function_class): New enum for different
4228         classes of functions.
4229         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
4230         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
4231         (TARGET_HAS_SINCOS): Likewise.
4232         (TARGET_LIBC_HAS_FUNCTION): New.
4233         * doc/tm.texi: Regenerated.
4234         * targhooks.h (default_libc_has_function): New declaration.
4235         (no_c99_libc_has_function): Ditto.
4236         (gnu_libc_has_function): Ditto.
4237         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
4238         and TARGET_HAS_SINCOS.
4240 2013-08-18  Jan Hubicka  <jh@suse.cz>
4242         * Makeifle-in (ipa-devirt.o): New.
4243         (GTFILES): Add ipa-utils.h and ipa-devirt.c
4244         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
4245         (analyze_functions): Look into possible targets of polymorphic call.
4246         * dumpfile.c (dump_files): Add type-inheritance dump.
4247         * dumpfile.h (TDI_inheritance): New.
4248         * ipa-devirt.c: New file.
4249         * ipa-utils.h (odr_type_d): Forward declare.
4250         (odr_type): New type.
4251         (build_type_inheritance_graph): Declare.
4252         (possible_polymorphic_call_targets): Declare and introduce inline
4253         variant when only edge is pased.
4254         (dump_possible_polymorphic_call_targets): Likewise.
4255         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
4256         * tree.c (type_in_anonymous_namespace_p): Break out from ...
4257         (types_same_for_odr): ... here.
4258         * tree.h (type_in_anonymous_namespace_p): Declare.
4260 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
4262         PR tree-optimization/58006
4263         * tree-parloops.c (take_address_of): Don't ICE if get_name
4264         returns NULL.
4265         (eliminate_local_variables_stmt): Remove clobber stmts.
4267 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
4269         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
4270         error message is issued for an alias to undefined symbol.
4272 2013-08-18  Jan Hubicka  <jh@suse.cz>
4274         * cgraph.c (cgraph_create_indirect_edge): Discover
4275         polymorphic calls and record basic info into indirect_info.
4276         * gimple-fold.c (gimple_fold_call): When doing BINFO based
4277         devirtualization, ignore objc function calls.
4278         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
4279         call with no parm index info.
4280         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
4281         * tree.c (virtual_method_call_p): New function.
4282         * tree.h (virtual_method_call_p): Declare.
4284 2013-08-16  Jan Hubicka  <jh@suse.cz>
4286         PR middle-end/58179
4287         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
4289 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
4291         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
4292         attribute.
4294 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
4296         * gengtype.c (type_for_name): Add special-case support for
4297         locating types within the "gcc::" namespace.
4298         (open_base_files): Emit a "using namespace gcc" directive.
4300 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
4302         PR target/58160
4303         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
4304         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
4306         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
4307         array instead of each individual operand as a separate argument.
4308         (emit_fusion_gpr_load): Likewise.
4309         (expand_fusion_gpr_load): Add new function declaration.
4311         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
4312         signature to have the operands passed as an array, instead of as
4313         separate arguments.  Allow ZERO_EXTEND to be in the memory
4314         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
4315         depend on the register live/dead flags when peepholes are run.
4316         (expand_fusion_gpr_load): New function to be called from the
4317         peephole2 pass, to change the register that addis sets to be the
4318         target register.
4319         (emit_fusion_gpr_load): Change the calling signature to have the
4320         operands passed as an array, instead of as separate arguments.
4321         Allow ZERO_EXTEND to be in the memory address, and also
4322         SIGN_EXTEND if -mpower8-fusion-sign.
4324         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
4325         unspec enumeration.
4326         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
4327         adjust the register addis loads up in the peephole2 pass.  Do not
4328         depend on the register live/dead state when the peephole pass is done.
4330 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
4332         * gengtype.c (create_user_defined_type): Ensure that the kind
4333         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
4334         declaration is seen before the GTY((user)) marking.
4336 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4338         PR target/58105
4339         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
4341 2013-08-16  Jan Hubicka  <jh@suse.cz>
4343         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
4344         arugment expected_type.
4345         (gimple_fold_call): Use it.
4346         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
4347         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
4348         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
4349         (try_make_edge_direct_virtual_call): Likewise.
4350         * tree.c (obj_type_ref_class): New.
4351         * tree.h (obj_type_ref_class): Use it.
4353 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
4355         * sched-vis.c (rtl_slim_pp_initialized): Remove.
4356         (rtl_slim_pp): Likewise.
4357         (init_rtl_slim_pretty_print): Likewise.
4358         (dump_value_slim):  Don't call it.  Use local pretty printer.
4359         (dump_insn_slim): Likewise.
4360         (dump_rtl_slim): Likewise.
4361         (str_pattern_slim): Likewise.
4362         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
4363         Simplify.
4365 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
4367         PR tree-optimization/58164
4368         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
4369         walk gimple_goto_dest of GIMPLE_GOTO.
4371         PR tree-optimization/58165
4372         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
4373         bi_call must be the last stmt in a bb, don't split_block, instead
4374         use fallthru edge from it and give up if there is none.
4375         Release conds vector when returning early.
4377 2013-08-14  Xinliang David Li  <davidxl@google.com>
4379         * config/i386/i386.c (ix86_option_override_internal):
4380         Remove unused variable and field.
4382 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4384         PR target/57949
4385         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
4386         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
4387         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
4388         and Linux, correct BLKmode alignment when 128-bit alignment is
4389         required and compatibility flag is not set.
4390         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
4391         for zero-size arguments when compatibility flag is not set.
4393 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
4395         PR tree-optimization/58145
4396         * tree-sra.c (build_ref_for_offset): If prev_base has
4397         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
4399 2013-08-14  Xinliang David Li  <davidxl@google.com>
4401         * config/i386/i386.c (ix86_option_override_internal):
4402         Fix uninitialized variable error.
4404 2013-08-14  Xinliang David Li  <davidxl@google.com>
4406         * config/i386/i386.opt: Define two new options.
4407         * config/i386/x86-tune.def: Add arch selector field in macros.
4408         * config/i386/i386.h: Adjust macro definition.
4409         * config/i386/i386.c (ix86_option_override_internal):
4410         Refactor the code.
4411         (parse_mtune_ctrl_str): New function.
4412         (set_ix86_tune_features): New function.
4413         (ix86_function_specific_restore): Call the new helper function.
4415 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
4417         PR rtl-optimization/57662
4418         * sel-sched.c (code_motion_process_successors): When the current insn
4419         is removed after the recursive traversal, break from the loop.
4420         Add comments and debug printouts.
4422 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
4423             Alexandre Oliva  <aoliva@redhat.com>
4425         PR target/58067
4426         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
4427         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
4428         there also UNSPEC_PLTOFF.
4430 2013-08-14  Marek Polacek  <polacek@redhat.com>
4432         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
4433         NUM_CONDITIONS bit positions.
4435 2013-08-13  Cary Coutant  <ccoutant@google.com>
4437         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
4438         (attr_checksum): Hash vector contents instead of pointer.
4439         (attr_checksum_ordered): Likewise.
4441 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
4443         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
4444         when Pmode != word_mode.  Add length_address attribute.
4445         (sse3_monitor_<mode>): Merge from sse3_monitor and
4446         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
4447         Pmode != word_mode.  Update insn length attribute.
4448         * config/i386/i386.c (ix86_option_override_internal): Update
4449         ix86_gen_monitor selection for merged sse3_monitor insn.
4451 2013-08-13  Julian Brown  <julian@codesourcery.com>
4453         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
4454         perform invalid legitimization on greater-than-word-size modes for
4455         TARGET_E500_DOUBLE.
4457 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
4459         * ira.c (setup_class_translate_array): Use aclass instead of cl
4460         for classes not fully covered by allocno classes.
4462 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
4464         PR tree-optimization/57661
4465         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
4466         * tree-inline.c (tree_function_versioning): Initialize it.
4467         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
4468         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
4469         that is not being copied.
4471         PR sanitizer/56417
4472         * asan.c (instrument_strlen_call): Fix typo in comment.
4473         Use char * type even for the lhs of POINTER_PLUS_EXPR.
4475 2013-08-13  Steve Ellcey  <sellcey@mips.com>
4477         * config/mips/mips.md (prefetch): Use lw instead of ld on
4478         loongson in 32bit mode.
4480 2013-08-13  Nick Clifton  <nickc@redhat.com>
4482         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
4484 2013-08-13  Jan Hubicka  <jh@suse.cz>
4486         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
4487         introduced edge; fix typo in sanity check.
4488         (cgraph_resolve_speculation): Export; improve diagnostic.
4489         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
4490         speculation at type mismatch.
4491         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
4492         (cgraph_resolve_speculation): Declare.
4493         (symtab_can_be_discarded): New function.
4494         * value-prof.c (gimple_ic_transform): Remove actual transform code.
4495         * ipa-inline-transform.c (speculation_removed): New global var.
4496         (clone_inlined_nodes): See if speculation can be removed.
4497         (inline_call): If speculations was removed, we growths may not match.
4498         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
4499         (speculation_useful_p): New function.
4500         (resolve_noninline_speculation): New function.
4501         (inline_small_functions): Resolve useless speculations.
4502         * ipa-inline.h (speculation_useful_p): Declare
4503         * ipa.c (can_replace_by_local_alias): Simplify.
4504         (ipa_profile): Produce speculative calls in non-lto, too;
4505         add simple cost model; produce local aliases.
4507 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
4509         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
4510         PASS_MANAGER_H.
4512 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
4514         * config/i386/i386.c (ix86_function_versions): Use error + inform.
4516 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
4518         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
4519         iterator instead of X87MODEF.
4521 2013-08-12  Perez Read  <netfirewall@gmail.com>
4523         PR target/58132
4524         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
4525         operand 0 for intel asm alternative.
4526         (*movabs<mode>_2): Ditto for operand 1.
4528 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
4530         * config/aarch64/arm_none.h
4531         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
4533 2013-08-12  Nick Clifton  <nickc@redhat.com>
4535         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
4537 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
4539         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
4540         expand for QI/HImode operand to produce more effictive code for
4541         unsigned char(short) --> float(double) conversion.
4543 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
4545         * doc/invoke.texi: Mention that -ftls-model does not force the final
4546         model.
4548 2013-08-12  Marek Polacek  <polacek@redhat.com>
4549             Marc Glisse  <marc.glisse@inria.fr>
4551         PR tree-optimization/57980
4552         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
4553         when creating -1 constant.
4555 2013-08-10  Jan Hubicka  <jh@suse.cz>
4557         Workaround binutils PR14342.
4558         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
4559         (gimple_init_edge_profiler): Likewise.
4560         (gimple_gen_ic_func_profiler): Likewise.
4562 2013-08-09  Jan Hubicka  <jh@suse.cz>
4564         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
4566 2013-08-09  Xinliang David Li  <davidxl@google.com>
4568         * config/i386/stringop.def: New file.
4569         * config/i386/stringop.opt: New file.
4570         * config/i386/i386-opts.h: Include stringopt.def.
4571         * config/i386/i386.opt: Include stringopt.opt.
4572         * config/i386/i386.c (ix86_option_override_internal):
4573         Override default size based stringop inline strategies with options.
4574         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
4575         New function.
4577 2013-08-09  Jan Hubicka  <jh@suse.cz>
4579         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
4581 2013-08-09  Jan Hubicka  <jh@suse.cz>
4583         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
4584         CGRAPH_FREQ_MAX.
4585         (dump_cgraph_node): Dump profile-id.
4586         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
4587         and common_target_probability.
4588         * lto-cgraph.c (lto_output_edge): Stream common targets.
4589         (lto_output_node): Stream profile ids.
4590         (input_node): Stream profile ids.
4591         (input_edge): Stream common targets.
4592         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
4593         * ipa.c: Include value-prof.h
4594         (ipa_profile_generate_summary): Turn indirect call statement histograms
4595         into common targets.
4596         (ipa_profile): Turn common targets into speculative edges.
4598 2013-08-09  Jan Hubicka  <jh@suse.cz>
4600         * cgraph.h (cgraph_node): Add profile_id.
4601         * value-prof.c (cgraph_node_map): Turn into pointer_map.
4602         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
4603         (del_node_map): Update.
4604         (find_func_by_funcdef_no): Replace by ...
4605         (find_func_by_profile_id): ... this one.
4606         (gimple_ic_transform): Do not remove useful histograms when
4607         speculation is not done; dump info when indirect call removal
4608         can happen at LTO.
4609         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
4610         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
4611         (__gcov_indirect_call_profiler_v2): .. this one.
4612         * profile.h (init_node_map): Update.
4613         * coverage.c (coverage_compute_profile_id): New function.
4614         * coverage.h (coverage_compute_profile_id): Declare.
4615         * tree-profile.c (init_ic_make_global_vars): Make
4616         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
4617         (gimple_init_edge_profiler): Update prototype of
4618         __gcov_indirect_call_profiler.
4619         (gimple_gen_ic_func_profiler): Simplify.
4620         (tree_profiling): Use init_node_map
4622 2013-08-09  Jan Hubicka  <jh@suse.cz>
4624         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
4625         non-speculative refs.
4626         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
4627         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
4628         (cgraph_set_call_stmt): Likewise.
4629         (cgraph_create_edge_1): Fix release checking compilatoin;
4630         clear lto_stmt_uid.
4631         (cgraph_free_edge): Free indirect info.
4632         (cgraph_turn_edge_to_speculative): New function.
4633         (cgraph_speculative_call_info): New function.
4634         (cgraph_make_edge_direct): Return direct edge; handle speculation.
4635         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
4636         (dump_cgraph_node): Dump speculation.
4637         (verify_edge_count_and_frequency): Accept speculative edges.
4638         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
4639         (verify_cgraph_node): Handle speculation.
4640         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
4641         (cgraph_set_call_stmt): Update prototype.
4642         (cgraph_make_edge_direct): Update prototype.
4643         (cgraph_speculative_call_info): Declare.
4644         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
4645         to change; update call of ipa_find_references.
4646         * ipa-ref.c (ipa_record_reference): Fix return value; clear
4647         lto_stmt_uid and speculative flags.
4648         (ipa_dump_references): Dump speculation.
4649         (ipa_clone_references): Clone speculative flag.
4650         (ipa_clone_referring): Likewise.
4651         (ipa_clone_ref): New function.
4652         (ipa_find_reference): Look into lto_stmt_uids
4653         (ipa_clear_stmts_in_references): Do not clear speculative calls.
4654         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
4655         (ipa_find_reference): Update declaration.
4656         (ipa_clone_ref): Declare.
4657         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
4658         stream speculative flag.
4659         (lto_output_ref): Stream statements uids and speculation.
4660         (input_ref): Likewise.
4661         (input_edge): Stream speuclation.
4662         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
4663         (cgraph_set_call_stmt_including_clones): Handle speculation.
4664         * ipa-inline.c (heap_edge_removal_hook): New function.
4665         (inline_small_functions): Register it.
4666         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
4667         also initialize refs.
4668         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
4669         edge to change.
4670         (try_make_edge_direct_simple_call): Likewise.
4671         (try_make_edge_direct_simple_call): Likewise.
4672         (update_indirect_edges_after_inlining): Likewise.
4673         (remove_described_reference): Look proper lto_stmt_uid.
4674         (propagate_controlled_uses): Likewise.
4675         (propagate_controlled_uses): Liekwise.
4676         * tree-inline.c (copy_bb): Copy speculative edges.
4677         (redirect_all_calls): New function.
4678         (copy_cfg_body): Do redirection after loop info is updated.
4679         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
4681 2013-08-09  Jan Hubicka  <jh@suse.cz>
4683         * lto-streamer-out.c (output_function): Renumber PHIs.
4684         * lto-streamer-in.c (input_function): Likewise.
4686 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
4688         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
4689         (get_lane_unsigned): Likewise.
4690         (dup_lane_scalar): Likewise.
4691         (get_lane): enable for VALL.
4692         * config/aarch64/aarch64-simd.md
4693         (aarch64_dup_lane_scalar<mode>): Remove.
4694         (aarch64_get_lane_signed<mode>): Likewise.
4695         (aarch64_get_lane_unsigned<mode>): Likewise.
4696         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
4697         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
4698         (aarch64_get_lane<mode>): Enable for all vector modes.
4699         (aarch64_get_lanedi): Remove misleading constraints.
4700         * config/aarch64/arm_neon.h
4701         (__aarch64_vget_lane_any): Define.
4702         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
4703         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
4704         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
4705         * config/aarch64/iterators.md (VDQQH): New.
4706         (VDQQHS): Likewise.
4707         (vwcore): Likewise.
4709 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
4711         * configure.ac: Add GAS check for LEON instructions on SPARC.
4712         * configure: Regenerate.
4713         * config.in: Likewise.
4714         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
4715         sparc*-*-* block.
4716         * config/sparc/sparc.opt (LEON, LEON3): New masks.
4717         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
4718         for LEON or LEON3.
4719         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
4720         (AS_LEON_FLAG): New macro.
4721         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
4722         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
4723         Deal with LEON and LEON3 for the memory model.
4724         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
4725         (atomic_compare_and_swap<mode>_1): Likewise.
4726         (*atomic_compare_and_swap<mode>_1): Likewise.
4728 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
4730         * config/arm/neon.md (vcond): Fix floating-point vector
4731         comparisons against 0.
4733 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
4735         * lra-constraints.c (emit_spill_move): Remove assert.
4736         (process_alt_operands): Add more debugging
4737         output.  Increase reject for spilling into memory.  Decrease
4738         reject for reloading scratch.
4739         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
4741 2013-08-08  Steve Ellcey  <sellcey@mips.com>
4743         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
4744         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
4745         micromips incompatible.  Add nan2008.
4746         (MULTILIB_DIRNAMES): Add nan2008.
4747         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
4748         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
4749         and micromips incompatible.  Add nan2008.
4750         (MULTILIB_DIRNAMES): Add nan2008.
4751         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
4753 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
4755         PR rtl-optimization/58079
4756         * combine.c (combine_simplify_rtx): Avoid using SUBST if
4757         simplify_comparison has widened a comparison with an integer.
4759 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4761         * config/arm/neon.md (movmisalign<mode>): Disable when we
4762         don't allow unaligned accesses.
4763         (*movmisalign<mode>_neon_store): Likewise.
4764         (*movmisalign<mode>_neon_load): Likewise.
4765         (*movmisalign<mode>_neon_store): Likewise.
4766         (*movmisalign<mode>_neon_load): Likewise.
4768 2013-08-08  Jan Hubicka  <jh@suse.cz>
4770         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
4771         (make_pass_rebuild_cgraph_edges): Also clear references.
4772         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
4773         * ipa-inline-transform.c (inline_transform): Remove all references
4774         after inlining.
4775         * cgraphunit.c (expand_function): Remove all references after
4776         expansion.
4777         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
4778         (ipa_find_reference): Rewrite to iterator.
4779         (remove_stmt_references): Likewise.
4780         (ipa_clear_stmts_in_references): New function.
4781         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
4782         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
4783         clear references.
4784         * ipa-split.c (split_function): Remove references in split function.
4786 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
4788         PR target/57431
4789         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
4790         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
4792 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
4794         PR target/56979
4795         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
4796         suggested mode for the assignment isn't compatible with the
4797         registers required.
4799 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4801         PR target/58065
4802         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
4804 2013-08-07  Xinliang David Li  <davidxl@google.com>
4806         * config/i386/i386.opt: New option -mtune-ctrl=.
4807         * config/i386/x86-tune.def: New file.
4808         * config/i386/i386.h: include x86-tune.def.
4809         * config/i386/i386.c (ix86_option_override_internal):
4810         Parsing -mtune-ctrl= option and set tune features.
4812 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
4814         PR other/12081
4815         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
4816         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
4817         to gen_2arg_fn_t.
4819 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
4821         * rtl.h (update_alignments): Declare.
4822         * final.c (grow_label_align): New function extracted from...
4823         (shorten_branches): ...here.  Call it.
4824         (update_alignments): New function.
4825         * reorg.c (sibling_labels): New variable.
4826         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
4827         the new label along with it onto the sibling_labels vector.
4828         (fill_simple_delay_slots): Adjust call to get_label_before.
4829         (fill_slots_from_thread): Likewise.
4830         (relax_delay_slots): Likewise.
4831         (make_return_insns): Likewise.
4832         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
4834 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
4836         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
4837         document its semantics.
4838         (diagnostic_report_diagnostic): Adjust accordingly.
4840 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
4842         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
4843         (sparc_option_override): ...and port to new C++ pass API.
4844         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
4846 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
4848         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
4850 2013-08-06  Caroline Tice  <cmtice@google.com>
4852         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
4853         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
4854         * tree-pass.h: Add pass_vtable_verify.
4855         * varasm.c (assemble_variable): Add code to properly set the comdat
4856         section and name for the .vtable_map_vars section.
4857         (assemble_vtyv_preinit_initializer): New function.
4858         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
4859         LINK_ONCE flag.
4860         * output.h: Add function decl for assemble_vtv_preinit_initializer.
4861         * vtable-verify.c: New file.
4862         * vtable-verify.h: New file.
4863         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
4864         initialiation levels.
4865         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
4866         * passes.def: Insert pass_vtable_verify.
4867         * aclocal.m4: Reorder includes.
4868         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
4869         -fvtv-counts options.
4870         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
4871         as appropriate, if -fvtable-verify=... is used.
4872         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
4873         -fvtable-verify=... is used.
4874         * Makefile.in (OBJS):  Add vtable-verify.o to list.
4875         (vtable-verify.o): Add new build rule.
4876         (GTFILES): Add vtable-verify.c to list.
4877         * common.opt (fvtable-verify=): New flag.
4878         (vtv_priority): Values for fvtable-verify= flag.
4879         (fvtv-counts): New flag.
4880         (fvtv-debug): New flag.
4881         * tree.h (save_vtable_map_decl): New extern function decl.
4883 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
4885         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
4886         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
4887         (pass_data_rl78_devirt): ...new pass_data instance and...
4888         (make_pass_rl78_devirt): ...new function.
4889         (rl78_asm_file_start): Port pass registration to new C++ API.
4891 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
4893         * coretypes.h (rtl_opt_pass): Add.
4894         (gcc::context): Add.
4895         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
4896         (epiphany_init): Port to new C++ pass API.
4897         (epiphany_optimize_mode_switching): Likewise.
4898         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
4899         (pass_manager::get_pass_mode_switching): New.
4900         (pass_manager::get_pass_peephole2): New.
4901         * mode-switching.c (pass_mode_switching): Add clone method.
4902         * recog.c (pass_peephole2): Add clone method.
4903         (pass_split_all_insns): Add clone method.
4905 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
4907         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
4908         (mips_option_override): ...here, porting to new C++ API for passes.
4910 2013-08-06  Jan Hubicka  <jh@suse.cz>
4912         * cgraph.c (cgraph_get_body): New function based on lto.c
4913         implementation.
4914         * cgraph.h (cgraph_get_body): Declare.
4915         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
4916         LTO paths.
4917         * cgraphunit.c (expand_function): Get body prior expanding.
4918         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
4919         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
4920         really need.
4921         * passes.c (do_per_function_toporder): Get body.
4922         * tree-inline.c (expand_call_inline): Get body prior inlining it.
4923         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
4925 2013-08-06  Martin Jambor  <mjambor@suse.cz>
4927         PR fortran/57987
4928         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
4929         is not re-finalized.  Rename second parameter to no_collect.
4931 2013-08-06  Martin Jambor  <mjambor@suse.cz>
4933         PR middle-end/58041
4934         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
4935         MEM_REF has proper alignment information.
4937 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
4939         PR other/12081
4940         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
4941         class insn_gen_fn.
4942         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
4943         rtx (*) (rtx, ...) with insn_gen_fn.
4944         * genoutput.c (output_insn_data): Cast gen_? function pointers to
4945         insn_gen_fn::stored_funcptr.  Add initializer braces.
4947 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
4949         Rewrite how instances of passes are cloned to remove assumptions
4950         about their sizes (thus allowing pass subclasses to have
4951         additional data fields, albeit non-GC-managed ones at this point).
4953         * passes.c (make_pass_instance): Now that passes have clone
4954         methods, rewrite this function to eliminate XNEW and memcpy
4955         calls that used hardcoded sizes.  Since this function no longer
4956         creates pass instances, rename it to...
4957         (add_pass_instance): ...this.  Document the old way that passes were
4958         numbered and flagged, and rework this function to continue using it.
4959         (next_pass_1): Add an initial_pass argument for use by
4960         add_pass_instance.
4961         (position_pass): When adding multiple instances of a pass, use
4962         the pass's clone method, rather than relying on the XNEW/memcpy
4963         within the former make_pass_instance (now add_pass_instance).
4964         (pass_manager::pass_manager): When invoking next_pass_1, also supply
4965         the initial instance of the current pass within the pass manager.
4967 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
4969         This is the automated part of the conversion of passes from C
4970         structs to C++ classes.
4972         Patch autogenerated by refactor_passes.py from
4973         https://github.com/davidmalcolm/gcc-refactoring-scripts
4974         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
4976         * asan.c (pass_asan): Convert from a global struct to a subclass of
4977         gimple_opt_pass along with...
4978         (pass_data_asan): ...new pass_data instance and...
4979         (make_pass_asan): ...new function.
4980         (pass_asan_O0): Convert from a global struct to a subclass of
4981         gimple_opt_pass along with...
4982         (pass_data_asan_O0): ...new pass_data instance and...
4983         (make_pass_asan_O0): ...new function.
4984         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
4985         subclass of rtl_opt_pass along with...
4986         (pass_data_inc_dec): ...new pass_data instance and...
4987         (make_pass_inc_dec): ...new function.
4988         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
4989         a subclass of rtl_opt_pass along with...
4990         (pass_data_reorder_blocks): ...new pass_data instance and...
4991         (make_pass_reorder_blocks): ...new function.
4992         (pass_duplicate_computed_gotos): Convert from a global struct to a
4993         subclass of rtl_opt_pass along with...
4994         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
4995         (make_pass_duplicate_computed_gotos): ...new function.
4996         (pass_partition_blocks): Convert from a global struct to a subclass of
4997         rtl_opt_pass along with...
4998         (pass_data_partition_blocks): ...new pass_data instance and...
4999         (make_pass_partition_blocks): ...new function.
5000         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
5001         struct to a subclass of rtl_opt_pass along with...
5002         (pass_data_branch_target_load_optimize1): ...new pass_data instance
5003         and...
5004         (make_pass_branch_target_load_optimize1): ...new function.
5005         (pass_branch_target_load_optimize2): Convert from a global struct to a
5006         subclass of rtl_opt_pass along with...
5007         (pass_data_branch_target_load_optimize2): ...new pass_data instance
5008         and...
5009         (make_pass_branch_target_load_optimize2): ...new function.
5010         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
5011         of rtl_opt_pass along with...
5012         (pass_data_jump): ...new pass_data instance and...
5013         (make_pass_jump): ...new function.
5014         (pass_jump2): Convert from a global struct to a subclass of
5015         rtl_opt_pass along with...
5016         (pass_data_jump2): ...new pass_data instance and...
5017         (make_pass_jump2): ...new function.
5018         * cfgexpand.c (pass_expand): Convert from a global struct to a
5019         subclass of rtl_opt_pass along with...
5020         (pass_data_expand): ...new pass_data instance and...
5021         (make_pass_expand): ...new function.
5022         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
5023         of rtl_opt_pass along with...
5024         (pass_data_free_cfg): ...new pass_data instance and...
5025         (make_pass_free_cfg): ...new function.
5026         (pass_into_cfg_layout_mode): Convert from a global struct to a
5027         subclass of rtl_opt_pass along with...
5028         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
5029         (make_pass_into_cfg_layout_mode): ...new function.
5030         (pass_outof_cfg_layout_mode): Convert from a global struct to a
5031         subclass of rtl_opt_pass along with...
5032         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
5033         (make_pass_outof_cfg_layout_mode): ...new function.
5034         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
5035         struct to a subclass of gimple_opt_pass along with...
5036         (pass_data_build_cgraph_edges): ...new pass_data instance and...
5037         (make_pass_build_cgraph_edges): ...new function.
5038         (pass_rebuild_cgraph_edges): Convert from a global struct to a
5039         subclass of gimple_opt_pass along with...
5040         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
5041         (make_pass_rebuild_cgraph_edges): ...new function.
5042         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
5043         subclass of gimple_opt_pass along with...
5044         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
5045         and...
5046         (make_pass_remove_cgraph_callee_edges): ...new function.
5047         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
5048         struct to a subclass of rtl_opt_pass along with...
5049         (pass_data_stack_adjustments): ...new pass_data instance and...
5050         (make_pass_stack_adjustments): ...new function.
5051         * combine.c (pass_combine): Convert from a global struct to a subclass
5052         of rtl_opt_pass along with...
5053         (pass_data_combine): ...new pass_data instance and...
5054         (make_pass_combine): ...new function.
5055         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
5056         global struct to a subclass of rtl_opt_pass along with...
5057         (pass_data_compare_elim_after_reload): ...new pass_data instance
5058         and...
5059         (make_pass_compare_elim_after_reload): ...new function.
5060         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
5061         of rtl_opt_pass along with...
5062         (pass_data_rtl_cprop): ...new pass_data instance and...
5063         (make_pass_rtl_cprop): ...new function.
5064         * cse.c (pass_cse): Convert from a global struct to a subclass of
5065         rtl_opt_pass along with...
5066         (pass_data_cse): ...new pass_data instance and...
5067         (make_pass_cse): ...new function.
5068         (pass_cse2): Convert from a global struct to a subclass of
5069         rtl_opt_pass along with...
5070         (pass_data_cse2): ...new pass_data instance and...
5071         (make_pass_cse2): ...new function.
5072         (pass_cse_after_global_opts): Convert from a global struct to a
5073         subclass of rtl_opt_pass along with...
5074         (pass_data_cse_after_global_opts): ...new pass_data instance and...
5075         (make_pass_cse_after_global_opts): ...new function.
5076         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
5077         of rtl_opt_pass along with...
5078         (pass_data_ud_rtl_dce): ...new pass_data instance and...
5079         (make_pass_ud_rtl_dce): ...new function.
5080         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
5081         rtl_opt_pass along with...
5082         (pass_data_fast_rtl_dce): ...new pass_data instance and...
5083         (make_pass_fast_rtl_dce): ...new function.
5084         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
5085         a subclass of rtl_opt_pass along with...
5086         (pass_data_df_initialize_opt): ...new pass_data instance and...
5087         (make_pass_df_initialize_opt): ...new function.
5088         (pass_df_initialize_no_opt): Convert from a global struct to a
5089         subclass of rtl_opt_pass along with...
5090         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
5091         (make_pass_df_initialize_no_opt): ...new function.
5092         (pass_df_finish): Convert from a global struct to a subclass of
5093         rtl_opt_pass along with...
5094         (pass_data_df_finish): ...new pass_data instance and...
5095         (make_pass_df_finish): ...new function.
5096         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
5097         rtl_opt_pass along with...
5098         (pass_data_rtl_dse1): ...new pass_data instance and...
5099         (make_pass_rtl_dse1): ...new function.
5100         (pass_rtl_dse2): Convert from a global struct to a subclass of
5101         rtl_opt_pass along with...
5102         (pass_data_rtl_dse2): ...new pass_data instance and...
5103         (make_pass_rtl_dse2): ...new function.
5104         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
5105         subclass of rtl_opt_pass along with...
5106         (pass_data_dwarf2_frame): ...new pass_data instance and...
5107         (make_pass_dwarf2_frame): ...new function.
5108         * except.c (pass_set_nothrow_function_flags): Convert from a global
5109         struct to a subclass of rtl_opt_pass along with...
5110         (pass_data_set_nothrow_function_flags): ...new pass_data instance
5111         and...
5112         (make_pass_set_nothrow_function_flags): ...new function.
5113         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
5114         subclass of rtl_opt_pass along with...
5115         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
5116         and...
5117         (make_pass_convert_to_eh_region_ranges): ...new function.
5118         * final.c (pass_compute_alignments): Convert from a global struct to a
5119         subclass of rtl_opt_pass along with...
5120         (pass_data_compute_alignments): ...new pass_data instance and...
5121         (make_pass_compute_alignments): ...new function.
5122         (pass_final): Convert from a global struct to a subclass of
5123         rtl_opt_pass along with...
5124         (pass_data_final): ...new pass_data instance and...
5125         (make_pass_final): ...new function.
5126         (pass_shorten_branches): Convert from a global struct to a subclass of
5127         rtl_opt_pass along with...
5128         (pass_data_shorten_branches): ...new pass_data instance and...
5129         (make_pass_shorten_branches): ...new function.
5130         (pass_clean_state): Convert from a global struct to a subclass of
5131         rtl_opt_pass along with...
5132         (pass_data_clean_state): ...new pass_data instance and...
5133         (make_pass_clean_state): ...new function.
5134         * function.c (pass_instantiate_virtual_regs): Convert from a global
5135         struct to a subclass of rtl_opt_pass along with...
5136         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
5137         (make_pass_instantiate_virtual_regs): ...new function.
5138         (pass_leaf_regs): Convert from a global struct to a subclass of
5139         rtl_opt_pass along with...
5140         (pass_data_leaf_regs): ...new pass_data instance and...
5141         (make_pass_leaf_regs): ...new function.
5142         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
5143         subclass of rtl_opt_pass along with...
5144         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
5145         and...
5146         (make_pass_thread_prologue_and_epilogue): ...new function.
5147         (pass_match_asm_constraints): Convert from a global struct to a
5148         subclass of rtl_opt_pass along with...
5149         (pass_data_match_asm_constraints): ...new pass_data instance and...
5150         (make_pass_match_asm_constraints): ...new function.
5151         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
5152         subclass of rtl_opt_pass along with...
5153         (pass_data_rtl_fwprop): ...new pass_data instance and...
5154         (make_pass_rtl_fwprop): ...new function.
5155         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
5156         rtl_opt_pass along with...
5157         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
5158         (make_pass_rtl_fwprop_addr): ...new function.
5159         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
5160         rtl_opt_pass along with...
5161         (pass_data_rtl_pre): ...new pass_data instance and...
5162         (make_pass_rtl_pre): ...new function.
5163         (pass_rtl_hoist): Convert from a global struct to a subclass of
5164         rtl_opt_pass along with...
5165         (pass_data_rtl_hoist): ...new pass_data instance and...
5166         (make_pass_rtl_hoist): ...new function.
5167         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
5168         subclass of gimple_opt_pass along with...
5169         (pass_data_lower_cf): ...new pass_data instance and...
5170         (make_pass_lower_cf): ...new function.
5171         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
5172         from a global struct to a subclass of gimple_opt_pass along with...
5173         (pass_data_strength_reduction): ...new pass_data instance and...
5174         (make_pass_strength_reduction): ...new function.
5175         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
5176         of rtl_opt_pass along with...
5177         (pass_data_rtl_ifcvt): ...new pass_data instance and...
5178         (make_pass_rtl_ifcvt): ...new function.
5179         (pass_if_after_combine): Convert from a global struct to a subclass of
5180         rtl_opt_pass along with...
5181         (pass_data_if_after_combine): ...new pass_data instance and...
5182         (make_pass_if_after_combine): ...new function.
5183         (pass_if_after_reload): Convert from a global struct to a subclass of
5184         rtl_opt_pass along with...
5185         (pass_data_if_after_reload): ...new pass_data instance and...
5186         (make_pass_if_after_reload): ...new function.
5187         * init-regs.c (pass_initialize_regs): Convert from a global struct to
5188         a subclass of rtl_opt_pass along with...
5189         (pass_data_initialize_regs): ...new pass_data instance and...
5190         (make_pass_initialize_regs): ...new function.
5191         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
5192         of ipa_opt_pass_d along with...
5193         (pass_data_ipa_cp): ...new pass_data instance and...
5194         (make_pass_ipa_cp): ...new function.
5195         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
5196         global struct to a subclass of gimple_opt_pass along with...
5197         (pass_data_inline_parameters): ...new pass_data instance and...
5198         (make_pass_inline_parameters): ...new function.
5199         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
5200         subclass of gimple_opt_pass along with...
5201         (pass_data_early_inline): ...new pass_data instance and...
5202         (make_pass_early_inline): ...new function.
5203         (pass_ipa_inline): Convert from a global struct to a subclass of
5204         ipa_opt_pass_d along with...
5205         (pass_data_ipa_inline): ...new pass_data instance and...
5206         (make_pass_ipa_inline): ...new function.
5207         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
5208         struct to a subclass of gimple_opt_pass along with...
5209         (pass_data_local_pure_const): ...new pass_data instance and...
5210         (make_pass_local_pure_const): ...new function.
5211         (pass_ipa_pure_const): Convert from a global struct to a subclass of
5212         ipa_opt_pass_d along with...
5213         (pass_data_ipa_pure_const): ...new pass_data instance and...
5214         (make_pass_ipa_pure_const): ...new function.
5215         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
5216         to a subclass of ipa_opt_pass_d along with...
5217         (pass_data_ipa_reference): ...new pass_data instance and...
5218         (make_pass_ipa_reference): ...new function.
5219         * ipa-split.c (pass_split_functions): Convert from a global struct to
5220         a subclass of gimple_opt_pass along with...
5221         (pass_data_split_functions): ...new pass_data instance and...
5222         (make_pass_split_functions): ...new function.
5223         (pass_feedback_split_functions): Convert from a global struct to a
5224         subclass of gimple_opt_pass along with...
5225         (pass_data_feedback_split_functions): ...new pass_data instance and...
5226         (make_pass_feedback_split_functions): ...new function.
5227         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
5228         global struct to a subclass of simple_ipa_opt_pass along with...
5229         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
5230         instance and...
5231         (make_pass_ipa_function_and_variable_visibility): ...new function.
5232         (pass_ipa_free_inline_summary): Convert from a global struct to a
5233         subclass of simple_ipa_opt_pass along with...
5234         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
5235         (make_pass_ipa_free_inline_summary): ...new function.
5236         (pass_ipa_whole_program_visibility): Convert from a global struct to a
5237         subclass of ipa_opt_pass_d along with...
5238         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
5239         and...
5240         (make_pass_ipa_whole_program_visibility): ...new function.
5241         (pass_ipa_profile): Convert from a global struct to a subclass of
5242         ipa_opt_pass_d along with...
5243         (pass_data_ipa_profile): ...new pass_data instance and...
5244         (make_pass_ipa_profile): ...new function.
5245         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
5246         ipa_opt_pass_d along with...
5247         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
5248         (make_pass_ipa_cdtor_merge): ...new function.
5249         * ira.c (pass_ira): Convert from a global struct to a subclass of
5250         rtl_opt_pass along with...
5251         (pass_data_ira): ...new pass_data instance and...
5252         (make_pass_ira): ...new function.
5253         (pass_reload): Convert from a global struct to a subclass of
5254         rtl_opt_pass along with...
5255         (pass_data_reload): ...new pass_data instance and...
5256         (make_pass_reload): ...new function.
5257         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
5258         subclass of rtl_opt_pass along with...
5259         (pass_data_cleanup_barriers): ...new pass_data instance and...
5260         (make_pass_cleanup_barriers): ...new function.
5261         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
5262         of rtl_opt_pass along with...
5263         (pass_data_loop2): ...new pass_data instance and...
5264         (make_pass_loop2): ...new function.
5265         (pass_rtl_loop_init): Convert from a global struct to a subclass of
5266         rtl_opt_pass along with...
5267         (pass_data_rtl_loop_init): ...new pass_data instance and...
5268         (make_pass_rtl_loop_init): ...new function.
5269         (pass_rtl_loop_done): Convert from a global struct to a subclass of
5270         rtl_opt_pass along with...
5271         (pass_data_rtl_loop_done): ...new pass_data instance and...
5272         (make_pass_rtl_loop_done): ...new function.
5273         (pass_rtl_move_loop_invariants): Convert from a global struct to a
5274         subclass of rtl_opt_pass along with...
5275         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
5276         (make_pass_rtl_move_loop_invariants): ...new function.
5277         (pass_rtl_unswitch): Convert from a global struct to a subclass of
5278         rtl_opt_pass along with...
5279         (pass_data_rtl_unswitch): ...new pass_data instance and...
5280         (make_pass_rtl_unswitch): ...new function.
5281         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
5282         subclass of rtl_opt_pass along with...
5283         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
5284         and...
5285         (make_pass_rtl_unroll_and_peel_loops): ...new function.
5286         (pass_rtl_doloop): Convert from a global struct to a subclass of
5287         rtl_opt_pass along with...
5288         (pass_data_rtl_doloop): ...new pass_data instance and...
5289         (make_pass_rtl_doloop): ...new function.
5290         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
5291         a subclass of rtl_opt_pass along with...
5292         (pass_data_lower_subreg): ...new pass_data instance and...
5293         (make_pass_lower_subreg): ...new function.
5294         (pass_lower_subreg2): Convert from a global struct to a subclass of
5295         rtl_opt_pass along with...
5296         (pass_data_lower_subreg2): ...new pass_data instance and...
5297         (make_pass_lower_subreg2): ...new function.
5298         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
5299         struct to a subclass of ipa_opt_pass_d along with...
5300         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
5301         (make_pass_ipa_lto_gimple_out): ...new function.
5302         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
5303         of ipa_opt_pass_d along with...
5304         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
5305         (make_pass_ipa_lto_finish_out): ...new function.
5306         * mode-switching.c (pass_mode_switching): Convert from a global struct
5307         to a subclass of rtl_opt_pass along with...
5308         (pass_data_mode_switching): ...new pass_data instance and...
5309         (make_pass_mode_switching): ...new function.
5310         * modulo-sched.c (pass_sms): Convert from a global struct to a
5311         subclass of rtl_opt_pass along with...
5312         (pass_data_sms): ...new pass_data instance and...
5313         (make_pass_sms): ...new function.
5314         * omp-low.c (pass_expand_omp): Convert from a global struct to a
5315         subclass of gimple_opt_pass along with...
5316         (pass_data_expand_omp): ...new pass_data instance and...
5317         (make_pass_expand_omp): ...new function.
5318         (pass_lower_omp): Convert from a global struct to a subclass of
5319         gimple_opt_pass along with...
5320         (pass_data_lower_omp): ...new pass_data instance and...
5321         (make_pass_lower_omp): ...new function.
5322         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
5323         of gimple_opt_pass along with...
5324         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
5325         (make_pass_diagnose_omp_blocks): ...new function.
5326         * passes.c (pass_early_local_passes): Convert from a global struct to
5327         a subclass of simple_ipa_opt_pass along with...
5328         (pass_data_early_local_passes): ...new pass_data instance and...
5329         (make_pass_early_local_passes): ...new function.
5330         (pass_all_early_optimizations): Convert from a global struct to a
5331         subclass of gimple_opt_pass along with...
5332         (pass_data_all_early_optimizations): ...new pass_data instance and...
5333         (make_pass_all_early_optimizations): ...new function.
5334         (pass_all_optimizations): Convert from a global struct to a subclass
5335         of gimple_opt_pass along with...
5336         (pass_data_all_optimizations): ...new pass_data instance and...
5337         (make_pass_all_optimizations): ...new function.
5338         (pass_all_optimizations_g): Convert from a global struct to a subclass
5339         of gimple_opt_pass along with...
5340         (pass_data_all_optimizations_g): ...new pass_data instance and...
5341         (make_pass_all_optimizations_g): ...new function.
5342         (pass_rest_of_compilation): Convert from a global struct to a subclass
5343         of rtl_opt_pass along with...
5344         (pass_data_rest_of_compilation): ...new pass_data instance and...
5345         (make_pass_rest_of_compilation): ...new function.
5346         (pass_postreload): Convert from a global struct to a subclass of
5347         rtl_opt_pass along with...
5348         (pass_data_postreload): ...new pass_data instance and...
5349         (make_pass_postreload): ...new function.
5350         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
5351         subclass of rtl_opt_pass along with...
5352         (pass_data_gcse2): ...new pass_data instance and...
5353         (make_pass_gcse2): ...new function.
5354         * postreload.c (pass_postreload_cse): Convert from a global struct to
5355         a subclass of rtl_opt_pass along with...
5356         (pass_data_postreload_cse): ...new pass_data instance and...
5357         (make_pass_postreload_cse): ...new function.
5358         * predict.c (pass_profile): Convert from a global struct to a subclass
5359         of gimple_opt_pass along with...
5360         (pass_data_profile): ...new pass_data instance and...
5361         (make_pass_profile): ...new function.
5362         (pass_strip_predict_hints): Convert from a global struct to a subclass
5363         of gimple_opt_pass along with...
5364         (pass_data_strip_predict_hints): ...new pass_data instance and...
5365         (make_pass_strip_predict_hints): ...new function.
5366         * recog.c (pass_peephole2): Convert from a global struct to a subclass
5367         of rtl_opt_pass along with...
5368         (pass_data_peephole2): ...new pass_data instance and...
5369         (make_pass_peephole2): ...new function.
5370         (pass_split_all_insns): Convert from a global struct to a subclass of
5371         rtl_opt_pass along with...
5372         (pass_data_split_all_insns): ...new pass_data instance and...
5373         (make_pass_split_all_insns): ...new function.
5374         (pass_split_after_reload): Convert from a global struct to a subclass
5375         of rtl_opt_pass along with...
5376         (pass_data_split_after_reload): ...new pass_data instance and...
5377         (make_pass_split_after_reload): ...new function.
5378         (pass_split_before_regstack): Convert from a global struct to a
5379         subclass of rtl_opt_pass along with...
5380         (pass_data_split_before_regstack): ...new pass_data instance and...
5381         (make_pass_split_before_regstack): ...new function.
5382         (pass_split_before_sched2): Convert from a global struct to a subclass
5383         of rtl_opt_pass along with...
5384         (pass_data_split_before_sched2): ...new pass_data instance and...
5385         (make_pass_split_before_sched2): ...new function.
5386         (pass_split_for_shorten_branches): Convert from a global struct to a
5387         subclass of rtl_opt_pass along with...
5388         (pass_data_split_for_shorten_branches): ...new pass_data instance
5389         and...
5390         (make_pass_split_for_shorten_branches): ...new function.
5391         * ree.c (pass_ree): Convert from a global struct to a subclass of
5392         rtl_opt_pass along with...
5393         (pass_data_ree): ...new pass_data instance and...
5394         (make_pass_ree): ...new function.
5395         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
5396         subclass of rtl_opt_pass along with...
5397         (pass_data_stack_regs): ...new pass_data instance and...
5398         (make_pass_stack_regs): ...new function.
5399         (pass_stack_regs_run): Convert from a global struct to a subclass of
5400         rtl_opt_pass along with...
5401         (pass_data_stack_regs_run): ...new pass_data instance and...
5402         (make_pass_stack_regs_run): ...new function.
5403         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
5404         subclass of rtl_opt_pass along with...
5405         (pass_data_cprop_hardreg): ...new pass_data instance and...
5406         (make_pass_cprop_hardreg): ...new function.
5407         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
5408         subclass of rtl_opt_pass along with...
5409         (pass_data_reginfo_init): ...new pass_data instance and...
5410         (make_pass_reginfo_init): ...new function.
5411         * regmove.c (pass_regmove): Convert from a global struct to a subclass
5412         of rtl_opt_pass along with...
5413         (pass_data_regmove): ...new pass_data instance and...
5414         (make_pass_regmove): ...new function.
5415         * regrename.c (pass_regrename): Convert from a global struct to a
5416         subclass of rtl_opt_pass along with...
5417         (pass_data_regrename): ...new pass_data instance and...
5418         (make_pass_regrename): ...new function.
5419         * reorg.c (pass_delay_slots): Convert from a global struct to a
5420         subclass of rtl_opt_pass along with...
5421         (pass_data_delay_slots): ...new pass_data instance and...
5422         (make_pass_delay_slots): ...new function.
5423         (pass_machine_reorg): Convert from a global struct to a subclass of
5424         rtl_opt_pass along with...
5425         (pass_data_machine_reorg): ...new pass_data instance and...
5426         (make_pass_machine_reorg): ...new function.
5427         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
5428         of rtl_opt_pass along with...
5429         (pass_data_sched): ...new pass_data instance and...
5430         (make_pass_sched): ...new function.
5431         (pass_sched2): Convert from a global struct to a subclass of
5432         rtl_opt_pass along with...
5433         (pass_data_sched2): ...new pass_data instance and...
5434         (make_pass_sched2): ...new function.
5435         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
5436         to a subclass of rtl_opt_pass along with...
5437         (pass_data_stack_ptr_mod): ...new pass_data instance and...
5438         (make_pass_stack_ptr_mod): ...new function.
5439         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
5440         to a subclass of rtl_opt_pass along with...
5441         (pass_data_rtl_store_motion): ...new pass_data instance and...
5442         (make_pass_rtl_store_motion): ...new function.
5443         * tracer.c (pass_tracer): Convert from a global struct to a subclass
5444         of gimple_opt_pass along with...
5445         (pass_data_tracer): ...new pass_data instance and...
5446         (make_pass_tracer): ...new function.
5447         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
5448         to a subclass of gimple_opt_pass along with...
5449         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
5450         (make_pass_diagnose_tm_blocks): ...new function.
5451         (pass_lower_tm): Convert from a global struct to a subclass of
5452         gimple_opt_pass along with...
5453         (pass_data_lower_tm): ...new pass_data instance and...
5454         (make_pass_lower_tm): ...new function.
5455         (pass_tm_init): Convert from a global struct to a subclass of
5456         gimple_opt_pass along with...
5457         (pass_data_tm_init): ...new pass_data instance and...
5458         (make_pass_tm_init): ...new function.
5459         (pass_tm_mark): Convert from a global struct to a subclass of
5460         gimple_opt_pass along with...
5461         (pass_data_tm_mark): ...new pass_data instance and...
5462         (make_pass_tm_mark): ...new function.
5463         (pass_tm_edges): Convert from a global struct to a subclass of
5464         gimple_opt_pass along with...
5465         (pass_data_tm_edges): ...new pass_data instance and...
5466         (make_pass_tm_edges): ...new function.
5467         (pass_tm_memopt): Convert from a global struct to a subclass of
5468         gimple_opt_pass along with...
5469         (pass_data_tm_memopt): ...new pass_data instance and...
5470         (make_pass_tm_memopt): ...new function.
5471         (pass_ipa_tm): Convert from a global struct to a subclass of
5472         simple_ipa_opt_pass along with...
5473         (pass_data_ipa_tm): ...new pass_data instance and...
5474         (make_pass_ipa_tm): ...new function.
5475         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
5476         subclass of gimple_opt_pass along with...
5477         (pass_data_call_cdce): ...new pass_data instance and...
5478         (make_pass_call_cdce): ...new function.
5479         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
5480         subclass of gimple_opt_pass along with...
5481         (pass_data_build_cfg): ...new pass_data instance and...
5482         (make_pass_build_cfg): ...new function.
5483         (pass_split_crit_edges): Convert from a global struct to a subclass of
5484         gimple_opt_pass along with...
5485         (pass_data_split_crit_edges): ...new pass_data instance and...
5486         (make_pass_split_crit_edges): ...new function.
5487         (pass_warn_function_return): Convert from a global struct to a
5488         subclass of gimple_opt_pass along with...
5489         (pass_data_warn_function_return): ...new pass_data instance and...
5490         (make_pass_warn_function_return): ...new function.
5491         (pass_warn_function_noreturn): Convert from a global struct to a
5492         subclass of gimple_opt_pass along with...
5493         (pass_data_warn_function_noreturn): ...new pass_data instance and...
5494         (make_pass_warn_function_noreturn): ...new function.
5495         (pass_warn_unused_result): Convert from a global struct to a subclass
5496         of gimple_opt_pass along with...
5497         (pass_data_warn_unused_result): ...new pass_data instance and...
5498         (make_pass_warn_unused_result): ...new function.
5499         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
5500         a subclass of gimple_opt_pass along with...
5501         (pass_data_merge_phi): ...new pass_data instance and...
5502         (make_pass_merge_phi): ...new function.
5503         * tree-complex.c (pass_lower_complex): Convert from a global struct to
5504         a subclass of gimple_opt_pass along with...
5505         (pass_data_lower_complex): ...new pass_data instance and...
5506         (make_pass_lower_complex): ...new function.
5507         (pass_lower_complex_O0): Convert from a global struct to a subclass of
5508         gimple_opt_pass along with...
5509         (pass_data_lower_complex_O0): ...new pass_data instance and...
5510         (make_pass_lower_complex_O0): ...new function.
5511         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
5512         subclass of gimple_opt_pass along with...
5513         (pass_data_lower_eh): ...new pass_data instance and...
5514         (make_pass_lower_eh): ...new function.
5515         (pass_refactor_eh): Convert from a global struct to a subclass of
5516         gimple_opt_pass along with...
5517         (pass_data_refactor_eh): ...new pass_data instance and...
5518         (make_pass_refactor_eh): ...new function.
5519         (pass_lower_resx): Convert from a global struct to a subclass of
5520         gimple_opt_pass along with...
5521         (pass_data_lower_resx): ...new pass_data instance and...
5522         (make_pass_lower_resx): ...new function.
5523         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
5524         of gimple_opt_pass along with...
5525         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
5526         (make_pass_lower_eh_dispatch): ...new function.
5527         (pass_cleanup_eh): Convert from a global struct to a subclass of
5528         gimple_opt_pass along with...
5529         (pass_data_cleanup_eh): ...new pass_data instance and...
5530         (make_pass_cleanup_eh): ...new function.
5531         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
5532         to a subclass of simple_ipa_opt_pass along with...
5533         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
5534         (make_pass_ipa_lower_emutls): ...new function.
5535         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
5536         a subclass of gimple_opt_pass along with...
5537         (pass_data_if_conversion): ...new pass_data instance and...
5538         (make_pass_if_conversion): ...new function.
5539         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
5540         subclass of gimple_opt_pass along with...
5541         (pass_data_build_ssa): ...new pass_data instance and...
5542         (make_pass_build_ssa): ...new function.
5543         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
5544         global struct to a subclass of gimple_opt_pass along with...
5545         (pass_data_loop_distribution): ...new pass_data instance and...
5546         (make_pass_loop_distribution): ...new function.
5547         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
5548         subclass of gimple_opt_pass along with...
5549         (pass_data_mudflap_1): ...new pass_data instance and...
5550         (make_pass_mudflap_1): ...new function.
5551         (pass_mudflap_2): Convert from a global struct to a subclass of
5552         gimple_opt_pass along with...
5553         (pass_data_mudflap_2): ...new pass_data instance and...
5554         (make_pass_mudflap_2): ...new function.
5555         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
5556         subclass of gimple_opt_pass along with...
5557         (pass_data_mudflap_1): ...new pass_data instance and...
5558         (make_pass_mudflap_1): ...new function.
5559         (pass_mudflap_2): Convert from a global struct to a subclass of
5560         gimple_opt_pass along with...
5561         (pass_data_mudflap_2): ...new pass_data instance and...
5562         (make_pass_mudflap_2): ...new function.
5563         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
5564         gimple_opt_pass along with...
5565         (pass_data_nrv): ...new pass_data instance and...
5566         (make_pass_nrv): ...new function.
5567         (pass_return_slot): Convert from a global struct to a subclass of
5568         gimple_opt_pass along with...
5569         (pass_data_return_slot): ...new pass_data instance and...
5570         (make_pass_return_slot): ...new function.
5571         * tree-object-size.c (pass_object_sizes): Convert from a global struct
5572         to a subclass of gimple_opt_pass along with...
5573         (pass_data_object_sizes): ...new pass_data instance and...
5574         (make_pass_object_sizes): ...new function.
5575         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
5576         global struct to a subclass of gimple_opt_pass along with...
5577         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
5578         and...
5579         (make_pass_cleanup_cfg_post_optimizing): ...new function.
5580         (pass_fixup_cfg): Convert from a global struct to a subclass of
5581         gimple_opt_pass along with...
5582         (pass_data_fixup_cfg): ...new pass_data instance and...
5583         (make_pass_fixup_cfg): ...new function.
5584         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
5585         (make_pass_mudflap_1): ...new function.
5586         (pass_mudflap_2): Replace declaration with that of...
5587         (make_pass_mudflap_2): ...new function.
5588         (pass_asan): Replace declaration with that of...
5589         (make_pass_asan): ...new function.
5590         (pass_asan_O0): Replace declaration with that of...
5591         (make_pass_asan_O0): ...new function.
5592         (pass_tsan): Replace declaration with that of...
5593         (make_pass_tsan): ...new function.
5594         (pass_tsan_O0): Replace declaration with that of...
5595         (make_pass_tsan_O0): ...new function.
5596         (pass_lower_cf): Replace declaration with that of...
5597         (make_pass_lower_cf): ...new function.
5598         (pass_refactor_eh): Replace declaration with that of...
5599         (make_pass_refactor_eh): ...new function.
5600         (pass_lower_eh): Replace declaration with that of...
5601         (make_pass_lower_eh): ...new function.
5602         (pass_lower_eh_dispatch): Replace declaration with that of...
5603         (make_pass_lower_eh_dispatch): ...new function.
5604         (pass_lower_resx): Replace declaration with that of...
5605         (make_pass_lower_resx): ...new function.
5606         (pass_build_cfg): Replace declaration with that of...
5607         (make_pass_build_cfg): ...new function.
5608         (pass_early_tree_profile): Replace declaration with that of...
5609         (make_pass_early_tree_profile): ...new function.
5610         (pass_cleanup_eh): Replace declaration with that of...
5611         (make_pass_cleanup_eh): ...new function.
5612         (pass_sra): Replace declaration with that of...
5613         (make_pass_sra): ...new function.
5614         (pass_sra_early): Replace declaration with that of...
5615         (make_pass_sra_early): ...new function.
5616         (pass_early_ipa_sra): Replace declaration with that of...
5617         (make_pass_early_ipa_sra): ...new function.
5618         (pass_tail_recursion): Replace declaration with that of...
5619         (make_pass_tail_recursion): ...new function.
5620         (pass_tail_calls): Replace declaration with that of...
5621         (make_pass_tail_calls): ...new function.
5622         (pass_tree_loop): Replace declaration with that of...
5623         (make_pass_tree_loop): ...new function.
5624         (pass_tree_loop_init): Replace declaration with that of...
5625         (make_pass_tree_loop_init): ...new function.
5626         (pass_lim): Replace declaration with that of...
5627         (make_pass_lim): ...new function.
5628         (pass_tree_unswitch): Replace declaration with that of...
5629         (make_pass_tree_unswitch): ...new function.
5630         (pass_predcom): Replace declaration with that of...
5631         (make_pass_predcom): ...new function.
5632         (pass_iv_canon): Replace declaration with that of...
5633         (make_pass_iv_canon): ...new function.
5634         (pass_scev_cprop): Replace declaration with that of...
5635         (make_pass_scev_cprop): ...new function.
5636         (pass_empty_loop): Replace declaration with that of...
5637         (make_pass_empty_loop): ...new function.
5638         (pass_record_bounds): Replace declaration with that of...
5639         (make_pass_record_bounds): ...new function.
5640         (pass_graphite): Replace declaration with that of...
5641         (make_pass_graphite): ...new function.
5642         (pass_graphite_transforms): Replace declaration with that of...
5643         (make_pass_graphite_transforms): ...new function.
5644         (pass_if_conversion): Replace declaration with that of...
5645         (make_pass_if_conversion): ...new function.
5646         (pass_loop_distribution): Replace declaration with that of...
5647         (make_pass_loop_distribution): ...new function.
5648         (pass_vectorize): Replace declaration with that of...
5649         (make_pass_vectorize): ...new function.
5650         (pass_slp_vectorize): Replace declaration with that of...
5651         (make_pass_slp_vectorize): ...new function.
5652         (pass_complete_unroll): Replace declaration with that of...
5653         (make_pass_complete_unroll): ...new function.
5654         (pass_complete_unrolli): Replace declaration with that of...
5655         (make_pass_complete_unrolli): ...new function.
5656         (pass_parallelize_loops): Replace declaration with that of...
5657         (make_pass_parallelize_loops): ...new function.
5658         (pass_loop_prefetch): Replace declaration with that of...
5659         (make_pass_loop_prefetch): ...new function.
5660         (pass_iv_optimize): Replace declaration with that of...
5661         (make_pass_iv_optimize): ...new function.
5662         (pass_tree_loop_done): Replace declaration with that of...
5663         (make_pass_tree_loop_done): ...new function.
5664         (pass_ch): Replace declaration with that of...
5665         (make_pass_ch): ...new function.
5666         (pass_ccp): Replace declaration with that of...
5667         (make_pass_ccp): ...new function.
5668         (pass_phi_only_cprop): Replace declaration with that of...
5669         (make_pass_phi_only_cprop): ...new function.
5670         (pass_build_ssa): Replace declaration with that of...
5671         (make_pass_build_ssa): ...new function.
5672         (pass_build_alias): Replace declaration with that of...
5673         (make_pass_build_alias): ...new function.
5674         (pass_build_ealias): Replace declaration with that of...
5675         (make_pass_build_ealias): ...new function.
5676         (pass_dominator): Replace declaration with that of...
5677         (make_pass_dominator): ...new function.
5678         (pass_dce): Replace declaration with that of...
5679         (make_pass_dce): ...new function.
5680         (pass_dce_loop): Replace declaration with that of...
5681         (make_pass_dce_loop): ...new function.
5682         (pass_cd_dce): Replace declaration with that of...
5683         (make_pass_cd_dce): ...new function.
5684         (pass_call_cdce): Replace declaration with that of...
5685         (make_pass_call_cdce): ...new function.
5686         (pass_merge_phi): Replace declaration with that of...
5687         (make_pass_merge_phi): ...new function.
5688         (pass_split_crit_edges): Replace declaration with that of...
5689         (make_pass_split_crit_edges): ...new function.
5690         (pass_pre): Replace declaration with that of...
5691         (make_pass_pre): ...new function.
5692         (pass_profile): Replace declaration with that of...
5693         (make_pass_profile): ...new function.
5694         (pass_strip_predict_hints): Replace declaration with that of...
5695         (make_pass_strip_predict_hints): ...new function.
5696         (pass_lower_complex_O0): Replace declaration with that of...
5697         (make_pass_lower_complex_O0): ...new function.
5698         (pass_lower_complex): Replace declaration with that of...
5699         (make_pass_lower_complex): ...new function.
5700         (pass_lower_vector): Replace declaration with that of...
5701         (make_pass_lower_vector): ...new function.
5702         (pass_lower_vector_ssa): Replace declaration with that of...
5703         (make_pass_lower_vector_ssa): ...new function.
5704         (pass_lower_omp): Replace declaration with that of...
5705         (make_pass_lower_omp): ...new function.
5706         (pass_diagnose_omp_blocks): Replace declaration with that of...
5707         (make_pass_diagnose_omp_blocks): ...new function.
5708         (pass_expand_omp): Replace declaration with that of...
5709         (make_pass_expand_omp): ...new function.
5710         (pass_expand_omp_ssa): Replace declaration with that of...
5711         (make_pass_expand_omp_ssa): ...new function.
5712         (pass_object_sizes): Replace declaration with that of...
5713         (make_pass_object_sizes): ...new function.
5714         (pass_strlen): Replace declaration with that of...
5715         (make_pass_strlen): ...new function.
5716         (pass_fold_builtins): Replace declaration with that of...
5717         (make_pass_fold_builtins): ...new function.
5718         (pass_stdarg): Replace declaration with that of...
5719         (make_pass_stdarg): ...new function.
5720         (pass_early_warn_uninitialized): Replace declaration with that of...
5721         (make_pass_early_warn_uninitialized): ...new function.
5722         (pass_late_warn_uninitialized): Replace declaration with that of...
5723         (make_pass_late_warn_uninitialized): ...new function.
5724         (pass_cse_reciprocals): Replace declaration with that of...
5725         (make_pass_cse_reciprocals): ...new function.
5726         (pass_cse_sincos): Replace declaration with that of...
5727         (make_pass_cse_sincos): ...new function.
5728         (pass_optimize_bswap): Replace declaration with that of...
5729         (make_pass_optimize_bswap): ...new function.
5730         (pass_optimize_widening_mul): Replace declaration with that of...
5731         (make_pass_optimize_widening_mul): ...new function.
5732         (pass_warn_function_return): Replace declaration with that of...
5733         (make_pass_warn_function_return): ...new function.
5734         (pass_warn_function_noreturn): Replace declaration with that of...
5735         (make_pass_warn_function_noreturn): ...new function.
5736         (pass_cselim): Replace declaration with that of...
5737         (make_pass_cselim): ...new function.
5738         (pass_phiopt): Replace declaration with that of...
5739         (make_pass_phiopt): ...new function.
5740         (pass_forwprop): Replace declaration with that of...
5741         (make_pass_forwprop): ...new function.
5742         (pass_phiprop): Replace declaration with that of...
5743         (make_pass_phiprop): ...new function.
5744         (pass_tree_ifcombine): Replace declaration with that of...
5745         (make_pass_tree_ifcombine): ...new function.
5746         (pass_dse): Replace declaration with that of...
5747         (make_pass_dse): ...new function.
5748         (pass_nrv): Replace declaration with that of...
5749         (make_pass_nrv): ...new function.
5750         (pass_rename_ssa_copies): Replace declaration with that of...
5751         (make_pass_rename_ssa_copies): ...new function.
5752         (pass_sink_code): Replace declaration with that of...
5753         (make_pass_sink_code): ...new function.
5754         (pass_fre): Replace declaration with that of...
5755         (make_pass_fre): ...new function.
5756         (pass_check_data_deps): Replace declaration with that of...
5757         (make_pass_check_data_deps): ...new function.
5758         (pass_copy_prop): Replace declaration with that of...
5759         (make_pass_copy_prop): ...new function.
5760         (pass_vrp): Replace declaration with that of...
5761         (make_pass_vrp): ...new function.
5762         (pass_uncprop): Replace declaration with that of...
5763         (make_pass_uncprop): ...new function.
5764         (pass_return_slot): Replace declaration with that of...
5765         (make_pass_return_slot): ...new function.
5766         (pass_reassoc): Replace declaration with that of...
5767         (make_pass_reassoc): ...new function.
5768         (pass_rebuild_cgraph_edges): Replace declaration with that of...
5769         (make_pass_rebuild_cgraph_edges): ...new function.
5770         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
5771         (make_pass_remove_cgraph_callee_edges): ...new function.
5772         (pass_build_cgraph_edges): Replace declaration with that of...
5773         (make_pass_build_cgraph_edges): ...new function.
5774         (pass_local_pure_const): Replace declaration with that of...
5775         (make_pass_local_pure_const): ...new function.
5776         (pass_tracer): Replace declaration with that of...
5777         (make_pass_tracer): ...new function.
5778         (pass_warn_unused_result): Replace declaration with that of...
5779         (make_pass_warn_unused_result): ...new function.
5780         (pass_diagnose_tm_blocks): Replace declaration with that of...
5781         (make_pass_diagnose_tm_blocks): ...new function.
5782         (pass_lower_tm): Replace declaration with that of...
5783         (make_pass_lower_tm): ...new function.
5784         (pass_tm_init): Replace declaration with that of...
5785         (make_pass_tm_init): ...new function.
5786         (pass_tm_mark): Replace declaration with that of...
5787         (make_pass_tm_mark): ...new function.
5788         (pass_tm_memopt): Replace declaration with that of...
5789         (make_pass_tm_memopt): ...new function.
5790         (pass_tm_edges): Replace declaration with that of...
5791         (make_pass_tm_edges): ...new function.
5792         (pass_split_functions): Replace declaration with that of...
5793         (make_pass_split_functions): ...new function.
5794         (pass_feedback_split_functions): Replace declaration with that of...
5795         (make_pass_feedback_split_functions): ...new function.
5796         (pass_strength_reduction): Replace declaration with that of...
5797         (make_pass_strength_reduction): ...new function.
5798         (pass_ipa_lower_emutls): Replace declaration with that of...
5799         (make_pass_ipa_lower_emutls): ...new function.
5800         (pass_ipa_function_and_variable_visibility): Replace declaration with
5801         that of...
5802         (make_pass_ipa_function_and_variable_visibility): ...new function.
5803         (pass_ipa_tree_profile): Replace declaration with that of...
5804         (make_pass_ipa_tree_profile): ...new function.
5805         (pass_early_local_passes): Replace declaration with that of...
5806         (make_pass_early_local_passes): ...new function.
5807         (pass_ipa_whole_program_visibility): Replace declaration with that
5808         of...
5809         (make_pass_ipa_whole_program_visibility): ...new function.
5810         (pass_ipa_lto_gimple_out): Replace declaration with that of...
5811         (make_pass_ipa_lto_gimple_out): ...new function.
5812         (pass_ipa_increase_alignment): Replace declaration with that of...
5813         (make_pass_ipa_increase_alignment): ...new function.
5814         (pass_ipa_inline): Replace declaration with that of...
5815         (make_pass_ipa_inline): ...new function.
5816         (pass_ipa_free_lang_data): Replace declaration with that of...
5817         (make_pass_ipa_free_lang_data): ...new function.
5818         (pass_ipa_free_inline_summary): Replace declaration with that of...
5819         (make_pass_ipa_free_inline_summary): ...new function.
5820         (pass_ipa_cp): Replace declaration with that of...
5821         (make_pass_ipa_cp): ...new function.
5822         (pass_ipa_reference): Replace declaration with that of...
5823         (make_pass_ipa_reference): ...new function.
5824         (pass_ipa_pure_const): Replace declaration with that of...
5825         (make_pass_ipa_pure_const): ...new function.
5826         (pass_ipa_pta): Replace declaration with that of...
5827         (make_pass_ipa_pta): ...new function.
5828         (pass_ipa_lto_finish_out): Replace declaration with that of...
5829         (make_pass_ipa_lto_finish_out): ...new function.
5830         (pass_ipa_tm): Replace declaration with that of...
5831         (make_pass_ipa_tm): ...new function.
5832         (pass_ipa_profile): Replace declaration with that of...
5833         (make_pass_ipa_profile): ...new function.
5834         (pass_ipa_cdtor_merge): Replace declaration with that of...
5835         (make_pass_ipa_cdtor_merge): ...new function.
5836         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
5837         of...
5838         (make_pass_cleanup_cfg_post_optimizing): ...new function.
5839         (pass_init_datastructures): Replace declaration with that of...
5840         (make_pass_init_datastructures): ...new function.
5841         (pass_fixup_cfg): Replace declaration with that of...
5842         (make_pass_fixup_cfg): ...new function.
5843         (pass_expand): Replace declaration with that of...
5844         (make_pass_expand): ...new function.
5845         (pass_instantiate_virtual_regs): Replace declaration with that of...
5846         (make_pass_instantiate_virtual_regs): ...new function.
5847         (pass_rtl_fwprop): Replace declaration with that of...
5848         (make_pass_rtl_fwprop): ...new function.
5849         (pass_rtl_fwprop_addr): Replace declaration with that of...
5850         (make_pass_rtl_fwprop_addr): ...new function.
5851         (pass_jump): Replace declaration with that of...
5852         (make_pass_jump): ...new function.
5853         (pass_jump2): Replace declaration with that of...
5854         (make_pass_jump2): ...new function.
5855         (pass_lower_subreg): Replace declaration with that of...
5856         (make_pass_lower_subreg): ...new function.
5857         (pass_cse): Replace declaration with that of...
5858         (make_pass_cse): ...new function.
5859         (pass_fast_rtl_dce): Replace declaration with that of...
5860         (make_pass_fast_rtl_dce): ...new function.
5861         (pass_ud_rtl_dce): Replace declaration with that of...
5862         (make_pass_ud_rtl_dce): ...new function.
5863         (pass_rtl_dce): Replace declaration with that of...
5864         (make_pass_rtl_dce): ...new function.
5865         (pass_rtl_dse1): Replace declaration with that of...
5866         (make_pass_rtl_dse1): ...new function.
5867         (pass_rtl_dse2): Replace declaration with that of...
5868         (make_pass_rtl_dse2): ...new function.
5869         (pass_rtl_dse3): Replace declaration with that of...
5870         (make_pass_rtl_dse3): ...new function.
5871         (pass_rtl_cprop): Replace declaration with that of...
5872         (make_pass_rtl_cprop): ...new function.
5873         (pass_rtl_pre): Replace declaration with that of...
5874         (make_pass_rtl_pre): ...new function.
5875         (pass_rtl_hoist): Replace declaration with that of...
5876         (make_pass_rtl_hoist): ...new function.
5877         (pass_rtl_store_motion): Replace declaration with that of...
5878         (make_pass_rtl_store_motion): ...new function.
5879         (pass_cse_after_global_opts): Replace declaration with that of...
5880         (make_pass_cse_after_global_opts): ...new function.
5881         (pass_rtl_ifcvt): Replace declaration with that of...
5882         (make_pass_rtl_ifcvt): ...new function.
5883         (pass_into_cfg_layout_mode): Replace declaration with that of...
5884         (make_pass_into_cfg_layout_mode): ...new function.
5885         (pass_outof_cfg_layout_mode): Replace declaration with that of...
5886         (make_pass_outof_cfg_layout_mode): ...new function.
5887         (pass_loop2): Replace declaration with that of...
5888         (make_pass_loop2): ...new function.
5889         (pass_rtl_loop_init): Replace declaration with that of...
5890         (make_pass_rtl_loop_init): ...new function.
5891         (pass_rtl_move_loop_invariants): Replace declaration with that of...
5892         (make_pass_rtl_move_loop_invariants): ...new function.
5893         (pass_rtl_unswitch): Replace declaration with that of...
5894         (make_pass_rtl_unswitch): ...new function.
5895         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
5896         (make_pass_rtl_unroll_and_peel_loops): ...new function.
5897         (pass_rtl_doloop): Replace declaration with that of...
5898         (make_pass_rtl_doloop): ...new function.
5899         (pass_rtl_loop_done): Replace declaration with that of...
5900         (make_pass_rtl_loop_done): ...new function.
5901         (pass_web): Replace declaration with that of...
5902         (make_pass_web): ...new function.
5903         (pass_cse2): Replace declaration with that of...
5904         (make_pass_cse2): ...new function.
5905         (pass_df_initialize_opt): Replace declaration with that of...
5906         (make_pass_df_initialize_opt): ...new function.
5907         (pass_df_initialize_no_opt): Replace declaration with that of...
5908         (make_pass_df_initialize_no_opt): ...new function.
5909         (pass_reginfo_init): Replace declaration with that of...
5910         (make_pass_reginfo_init): ...new function.
5911         (pass_inc_dec): Replace declaration with that of...
5912         (make_pass_inc_dec): ...new function.
5913         (pass_stack_ptr_mod): Replace declaration with that of...
5914         (make_pass_stack_ptr_mod): ...new function.
5915         (pass_initialize_regs): Replace declaration with that of...
5916         (make_pass_initialize_regs): ...new function.
5917         (pass_combine): Replace declaration with that of...
5918         (make_pass_combine): ...new function.
5919         (pass_if_after_combine): Replace declaration with that of...
5920         (make_pass_if_after_combine): ...new function.
5921         (pass_ree): Replace declaration with that of...
5922         (make_pass_ree): ...new function.
5923         (pass_partition_blocks): Replace declaration with that of...
5924         (make_pass_partition_blocks): ...new function.
5925         (pass_match_asm_constraints): Replace declaration with that of...
5926         (make_pass_match_asm_constraints): ...new function.
5927         (pass_regmove): Replace declaration with that of...
5928         (make_pass_regmove): ...new function.
5929         (pass_split_all_insns): Replace declaration with that of...
5930         (make_pass_split_all_insns): ...new function.
5931         (pass_fast_rtl_byte_dce): Replace declaration with that of...
5932         (make_pass_fast_rtl_byte_dce): ...new function.
5933         (pass_lower_subreg2): Replace declaration with that of...
5934         (make_pass_lower_subreg2): ...new function.
5935         (pass_mode_switching): Replace declaration with that of...
5936         (make_pass_mode_switching): ...new function.
5937         (pass_sms): Replace declaration with that of...
5938         (make_pass_sms): ...new function.
5939         (pass_sched): Replace declaration with that of...
5940         (make_pass_sched): ...new function.
5941         (pass_ira): Replace declaration with that of...
5942         (make_pass_ira): ...new function.
5943         (pass_reload): Replace declaration with that of...
5944         (make_pass_reload): ...new function.
5945         (pass_clean_state): Replace declaration with that of...
5946         (make_pass_clean_state): ...new function.
5947         (pass_branch_prob): Replace declaration with that of...
5948         (make_pass_branch_prob): ...new function.
5949         (pass_value_profile_transformations): Replace declaration with that
5950         of...
5951         (make_pass_value_profile_transformations): ...new function.
5952         (pass_postreload_cse): Replace declaration with that of...
5953         (make_pass_postreload_cse): ...new function.
5954         (pass_gcse2): Replace declaration with that of...
5955         (make_pass_gcse2): ...new function.
5956         (pass_split_after_reload): Replace declaration with that of...
5957         (make_pass_split_after_reload): ...new function.
5958         (pass_branch_target_load_optimize1): Replace declaration with that
5959         of...
5960         (make_pass_branch_target_load_optimize1): ...new function.
5961         (pass_thread_prologue_and_epilogue): Replace declaration with that
5962         of...
5963         (make_pass_thread_prologue_and_epilogue): ...new function.
5964         (pass_stack_adjustments): Replace declaration with that of...
5965         (make_pass_stack_adjustments): ...new function.
5966         (pass_peephole2): Replace declaration with that of...
5967         (make_pass_peephole2): ...new function.
5968         (pass_if_after_reload): Replace declaration with that of...
5969         (make_pass_if_after_reload): ...new function.
5970         (pass_regrename): Replace declaration with that of...
5971         (make_pass_regrename): ...new function.
5972         (pass_cprop_hardreg): Replace declaration with that of...
5973         (make_pass_cprop_hardreg): ...new function.
5974         (pass_reorder_blocks): Replace declaration with that of...
5975         (make_pass_reorder_blocks): ...new function.
5976         (pass_branch_target_load_optimize2): Replace declaration with that
5977         of...
5978         (make_pass_branch_target_load_optimize2): ...new function.
5979         (pass_leaf_regs): Replace declaration with that of...
5980         (make_pass_leaf_regs): ...new function.
5981         (pass_split_before_sched2): Replace declaration with that of...
5982         (make_pass_split_before_sched2): ...new function.
5983         (pass_compare_elim_after_reload): Replace declaration with that of...
5984         (make_pass_compare_elim_after_reload): ...new function.
5985         (pass_sched2): Replace declaration with that of...
5986         (make_pass_sched2): ...new function.
5987         (pass_stack_regs): Replace declaration with that of...
5988         (make_pass_stack_regs): ...new function.
5989         (pass_stack_regs_run): Replace declaration with that of...
5990         (make_pass_stack_regs_run): ...new function.
5991         (pass_df_finish): Replace declaration with that of...
5992         (make_pass_df_finish): ...new function.
5993         (pass_compute_alignments): Replace declaration with that of...
5994         (make_pass_compute_alignments): ...new function.
5995         (pass_duplicate_computed_gotos): Replace declaration with that of...
5996         (make_pass_duplicate_computed_gotos): ...new function.
5997         (pass_variable_tracking): Replace declaration with that of...
5998         (make_pass_variable_tracking): ...new function.
5999         (pass_free_cfg): Replace declaration with that of...
6000         (make_pass_free_cfg): ...new function.
6001         (pass_machine_reorg): Replace declaration with that of...
6002         (make_pass_machine_reorg): ...new function.
6003         (pass_cleanup_barriers): Replace declaration with that of...
6004         (make_pass_cleanup_barriers): ...new function.
6005         (pass_delay_slots): Replace declaration with that of...
6006         (make_pass_delay_slots): ...new function.
6007         (pass_split_for_shorten_branches): Replace declaration with that of...
6008         (make_pass_split_for_shorten_branches): ...new function.
6009         (pass_split_before_regstack): Replace declaration with that of...
6010         (make_pass_split_before_regstack): ...new function.
6011         (pass_convert_to_eh_region_ranges): Replace declaration with that
6012         of...
6013         (make_pass_convert_to_eh_region_ranges): ...new function.
6014         (pass_shorten_branches): Replace declaration with that of...
6015         (make_pass_shorten_branches): ...new function.
6016         (pass_set_nothrow_function_flags): Replace declaration with that of...
6017         (make_pass_set_nothrow_function_flags): ...new function.
6018         (pass_dwarf2_frame): Replace declaration with that of...
6019         (make_pass_dwarf2_frame): ...new function.
6020         (pass_final): Replace declaration with that of...
6021         (make_pass_final): ...new function.
6022         (pass_rtl_seqabstr): Replace declaration with that of...
6023         (make_pass_rtl_seqabstr): ...new function.
6024         (pass_release_ssa_names): Replace declaration with that of...
6025         (make_pass_release_ssa_names): ...new function.
6026         (pass_early_inline): Replace declaration with that of...
6027         (make_pass_early_inline): ...new function.
6028         (pass_inline_parameters): Replace declaration with that of...
6029         (make_pass_inline_parameters): ...new function.
6030         (pass_update_address_taken): Replace declaration with that of...
6031         (make_pass_update_address_taken): ...new function.
6032         (pass_convert_switch): Replace declaration with that of...
6033         (make_pass_convert_switch): ...new function.
6034         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
6035         to a subclass of simple_ipa_opt_pass along with...
6036         (pass_data_ipa_tree_profile): ...new pass_data instance and...
6037         (make_pass_ipa_tree_profile): ...new function.
6038         * tree-sra.c (pass_sra_early): Convert from a global struct to a
6039         subclass of gimple_opt_pass along with...
6040         (pass_data_sra_early): ...new pass_data instance and...
6041         (make_pass_sra_early): ...new function.
6042         (pass_sra): Convert from a global struct to a subclass of
6043         gimple_opt_pass along with...
6044         (pass_data_sra): ...new pass_data instance and...
6045         (make_pass_sra): ...new function.
6046         (pass_early_ipa_sra): Convert from a global struct to a subclass of
6047         gimple_opt_pass along with...
6048         (pass_data_early_ipa_sra): ...new pass_data instance and...
6049         (make_pass_early_ipa_sra): ...new function.
6050         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
6051         subclass of gimple_opt_pass along with...
6052         (pass_data_ccp): ...new pass_data instance and...
6053         (make_pass_ccp): ...new function.
6054         (pass_fold_builtins): Convert from a global struct to a subclass of
6055         gimple_opt_pass along with...
6056         (pass_data_fold_builtins): ...new pass_data instance and...
6057         (make_pass_fold_builtins): ...new function.
6058         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
6059         subclass of gimple_opt_pass along with...
6060         (pass_data_copy_prop): ...new pass_data instance and...
6061         (make_pass_copy_prop): ...new function.
6062         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
6063         global struct to a subclass of gimple_opt_pass along with...
6064         (pass_data_rename_ssa_copies): ...new pass_data instance and...
6065         (make_pass_rename_ssa_copies): ...new function.
6066         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
6067         subclass of gimple_opt_pass along with...
6068         (pass_data_dce): ...new pass_data instance and...
6069         (make_pass_dce): ...new function.
6070         (pass_dce_loop): Convert from a global struct to a subclass of
6071         gimple_opt_pass along with...
6072         (pass_data_dce_loop): ...new pass_data instance and...
6073         (make_pass_dce_loop): ...new function.
6074         (pass_cd_dce): Convert from a global struct to a subclass of
6075         gimple_opt_pass along with...
6076         (pass_data_cd_dce): ...new pass_data instance and...
6077         (make_pass_cd_dce): ...new function.
6078         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
6079         subclass of gimple_opt_pass along with...
6080         (pass_data_dominator): ...new pass_data instance and...
6081         (make_pass_dominator): ...new function.
6082         (pass_phi_only_cprop): Convert from a global struct to a subclass of
6083         gimple_opt_pass along with...
6084         (pass_data_phi_only_cprop): ...new pass_data instance and...
6085         (make_pass_phi_only_cprop): ...new function.
6086         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
6087         subclass of gimple_opt_pass along with...
6088         (pass_data_dse): ...new pass_data instance and...
6089         (make_pass_dse): ...new function.
6090         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
6091         a subclass of gimple_opt_pass along with...
6092         (pass_data_forwprop): ...new pass_data instance and...
6093         (make_pass_forwprop): ...new function.
6094         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
6095         struct to a subclass of gimple_opt_pass along with...
6096         (pass_data_tree_ifcombine): ...new pass_data instance and...
6097         (make_pass_tree_ifcombine): ...new function.
6098         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
6099         subclass of gimple_opt_pass along with...
6100         (pass_data_ch): ...new pass_data instance and...
6101         (make_pass_ch): ...new function.
6102         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
6103         subclass of gimple_opt_pass along with...
6104         (pass_data_tree_loop): ...new pass_data instance and...
6105         (make_pass_tree_loop): ...new function.
6106         (pass_tree_loop_init): Convert from a global struct to a subclass of
6107         gimple_opt_pass along with...
6108         (pass_data_tree_loop_init): ...new pass_data instance and...
6109         (make_pass_tree_loop_init): ...new function.
6110         (pass_lim): Convert from a global struct to a subclass of
6111         gimple_opt_pass along with...
6112         (pass_data_lim): ...new pass_data instance and...
6113         (make_pass_lim): ...new function.
6114         (pass_tree_unswitch): Convert from a global struct to a subclass of
6115         gimple_opt_pass along with...
6116         (pass_data_tree_unswitch): ...new pass_data instance and...
6117         (make_pass_tree_unswitch): ...new function.
6118         (pass_predcom): Convert from a global struct to a subclass of
6119         gimple_opt_pass along with...
6120         (pass_data_predcom): ...new pass_data instance and...
6121         (make_pass_predcom): ...new function.
6122         (pass_vectorize): Convert from a global struct to a subclass of
6123         gimple_opt_pass along with...
6124         (pass_data_vectorize): ...new pass_data instance and...
6125         (make_pass_vectorize): ...new function.
6126         (pass_graphite): Convert from a global struct to a subclass of
6127         gimple_opt_pass along with...
6128         (pass_data_graphite): ...new pass_data instance and...
6129         (make_pass_graphite): ...new function.
6130         (pass_graphite_transforms): Convert from a global struct to a subclass
6131         of gimple_opt_pass along with...
6132         (pass_data_graphite_transforms): ...new pass_data instance and...
6133         (make_pass_graphite_transforms): ...new function.
6134         (pass_check_data_deps): Convert from a global struct to a subclass of
6135         gimple_opt_pass along with...
6136         (pass_data_check_data_deps): ...new pass_data instance and...
6137         (make_pass_check_data_deps): ...new function.
6138         (pass_iv_canon): Convert from a global struct to a subclass of
6139         gimple_opt_pass along with...
6140         (pass_data_iv_canon): ...new pass_data instance and...
6141         (make_pass_iv_canon): ...new function.
6142         (pass_scev_cprop): Convert from a global struct to a subclass of
6143         gimple_opt_pass along with...
6144         (pass_data_scev_cprop): ...new pass_data instance and...
6145         (make_pass_scev_cprop): ...new function.
6146         (pass_record_bounds): Convert from a global struct to a subclass of
6147         gimple_opt_pass along with...
6148         (pass_data_record_bounds): ...new pass_data instance and...
6149         (make_pass_record_bounds): ...new function.
6150         (pass_complete_unroll): Convert from a global struct to a subclass of
6151         gimple_opt_pass along with...
6152         (pass_data_complete_unroll): ...new pass_data instance and...
6153         (make_pass_complete_unroll): ...new function.
6154         (pass_complete_unrolli): Convert from a global struct to a subclass of
6155         gimple_opt_pass along with...
6156         (pass_data_complete_unrolli): ...new pass_data instance and...
6157         (make_pass_complete_unrolli): ...new function.
6158         (pass_parallelize_loops): Convert from a global struct to a subclass
6159         of gimple_opt_pass along with...
6160         (pass_data_parallelize_loops): ...new pass_data instance and...
6161         (make_pass_parallelize_loops): ...new function.
6162         (pass_loop_prefetch): Convert from a global struct to a subclass of
6163         gimple_opt_pass along with...
6164         (pass_data_loop_prefetch): ...new pass_data instance and...
6165         (make_pass_loop_prefetch): ...new function.
6166         (pass_iv_optimize): Convert from a global struct to a subclass of
6167         gimple_opt_pass along with...
6168         (pass_data_iv_optimize): ...new pass_data instance and...
6169         (make_pass_iv_optimize): ...new function.
6170         (pass_tree_loop_done): Convert from a global struct to a subclass of
6171         gimple_opt_pass along with...
6172         (pass_data_tree_loop_done): ...new pass_data instance and...
6173         (make_pass_tree_loop_done): ...new function.
6174         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
6175         struct to a subclass of gimple_opt_pass along with...
6176         (pass_data_cse_reciprocals): ...new pass_data instance and...
6177         (make_pass_cse_reciprocals): ...new function.
6178         (pass_cse_sincos): Convert from a global struct to a subclass of
6179         gimple_opt_pass along with...
6180         (pass_data_cse_sincos): ...new pass_data instance and...
6181         (make_pass_cse_sincos): ...new function.
6182         (pass_optimize_bswap): Convert from a global struct to a subclass of
6183         gimple_opt_pass along with...
6184         (pass_data_optimize_bswap): ...new pass_data instance and...
6185         (make_pass_optimize_bswap): ...new function.
6186         (pass_optimize_widening_mul): Convert from a global struct to a
6187         subclass of gimple_opt_pass along with...
6188         (pass_data_optimize_widening_mul): ...new pass_data instance and...
6189         (make_pass_optimize_widening_mul): ...new function.
6190         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
6191         subclass of gimple_opt_pass along with...
6192         (pass_data_phiopt): ...new pass_data instance and...
6193         (make_pass_phiopt): ...new function.
6194         (pass_cselim): Convert from a global struct to a subclass of
6195         gimple_opt_pass along with...
6196         (pass_data_cselim): ...new pass_data instance and...
6197         (make_pass_cselim): ...new function.
6198         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
6199         subclass of gimple_opt_pass along with...
6200         (pass_data_phiprop): ...new pass_data instance and...
6201         (make_pass_phiprop): ...new function.
6202         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
6203         subclass of gimple_opt_pass along with...
6204         (pass_data_pre): ...new pass_data instance and...
6205         (make_pass_pre): ...new function.
6206         (pass_fre): Convert from a global struct to a subclass of
6207         gimple_opt_pass along with...
6208         (pass_data_fre): ...new pass_data instance and...
6209         (make_pass_fre): ...new function.
6210         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
6211         subclass of gimple_opt_pass along with...
6212         (pass_data_reassoc): ...new pass_data instance and...
6213         (make_pass_reassoc): ...new function.
6214         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
6215         subclass of gimple_opt_pass along with...
6216         (pass_data_sink_code): ...new pass_data instance and...
6217         (make_pass_sink_code): ...new function.
6218         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
6219         subclass of gimple_opt_pass along with...
6220         (pass_data_strlen): ...new pass_data instance and...
6221         (make_pass_strlen): ...new function.
6222         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
6223         struct to a subclass of gimple_opt_pass along with...
6224         (pass_data_build_alias): ...new pass_data instance and...
6225         (make_pass_build_alias): ...new function.
6226         (pass_build_ealias): Convert from a global struct to a subclass of
6227         gimple_opt_pass along with...
6228         (pass_data_build_ealias): ...new pass_data instance and...
6229         (make_pass_build_ealias): ...new function.
6230         (pass_ipa_pta): Convert from a global struct to a subclass of
6231         simple_ipa_opt_pass along with...
6232         (pass_data_ipa_pta): ...new pass_data instance and...
6233         (make_pass_ipa_pta): ...new function.
6234         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
6235         subclass of gimple_opt_pass along with...
6236         (pass_data_uncprop): ...new pass_data instance and...
6237         (make_pass_uncprop): ...new function.
6238         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
6239         global struct to a subclass of gimple_opt_pass along with...
6240         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
6241         (make_pass_late_warn_uninitialized): ...new function.
6242         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
6243         to a subclass of gimple_opt_pass along with...
6244         (pass_data_init_datastructures): ...new pass_data instance and...
6245         (make_pass_init_datastructures): ...new function.
6246         (pass_early_warn_uninitialized): Convert from a global struct to a
6247         subclass of gimple_opt_pass along with...
6248         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
6249         (make_pass_early_warn_uninitialized): ...new function.
6250         (pass_update_address_taken): Convert from a global struct to a
6251         subclass of gimple_opt_pass along with...
6252         (pass_data_update_address_taken): ...new pass_data instance and...
6253         (make_pass_update_address_taken): ...new function.
6254         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
6255         struct to a subclass of gimple_opt_pass along with...
6256         (pass_data_release_ssa_names): ...new pass_data instance and...
6257         (make_pass_release_ssa_names): ...new function.
6258         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
6259         subclass of gimple_opt_pass along with...
6260         (pass_data_stdarg): ...new pass_data instance and...
6261         (make_pass_stdarg): ...new function.
6262         * tree-switch-conversion.c (pass_convert_switch): Convert from a
6263         global struct to a subclass of gimple_opt_pass along with...
6264         (pass_data_convert_switch): ...new pass_data instance and...
6265         (make_pass_convert_switch): ...new function.
6266         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
6267         to a subclass of gimple_opt_pass along with...
6268         (pass_data_tail_recursion): ...new pass_data instance and...
6269         (make_pass_tail_recursion): ...new function.
6270         (pass_tail_calls): Convert from a global struct to a subclass of
6271         gimple_opt_pass along with...
6272         (pass_data_tail_calls): ...new pass_data instance and...
6273         (make_pass_tail_calls): ...new function.
6274         * tree-vect-generic.c (pass_lower_vector): Convert from a global
6275         struct to a subclass of gimple_opt_pass along with...
6276         (pass_data_lower_vector): ...new pass_data instance and...
6277         (make_pass_lower_vector): ...new function.
6278         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
6279         gimple_opt_pass along with...
6280         (pass_data_lower_vector_ssa): ...new pass_data instance and...
6281         (make_pass_lower_vector_ssa): ...new function.
6282         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
6283         to a subclass of gimple_opt_pass along with...
6284         (pass_data_slp_vectorize): ...new pass_data instance and...
6285         (make_pass_slp_vectorize): ...new function.
6286         (pass_ipa_increase_alignment): Convert from a global struct to a
6287         subclass of simple_ipa_opt_pass along with...
6288         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
6289         (make_pass_ipa_increase_alignment): ...new function.
6290         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
6291         gimple_opt_pass along with...
6292         (pass_data_vrp): ...new pass_data instance and...
6293         (make_pass_vrp): ...new function.
6294         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
6295         subclass of simple_ipa_opt_pass along with...
6296         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
6297         (make_pass_ipa_free_lang_data): ...new function.
6298         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
6299         gimple_opt_pass along with...
6300         (pass_data_tsan): ...new pass_data instance and...
6301         (make_pass_tsan): ...new function.
6302         (pass_tsan_O0): Convert from a global struct to a subclass of
6303         gimple_opt_pass along with...
6304         (pass_data_tsan_O0): ...new pass_data instance and...
6305         (make_pass_tsan_O0): ...new function.
6306         * var-tracking.c (pass_variable_tracking): Convert from a global
6307         struct to a subclass of rtl_opt_pass along with...
6308         (pass_data_variable_tracking): ...new pass_data instance and...
6309         (make_pass_variable_tracking): ...new function.
6310         * web.c (pass_web): Convert from a global struct to a subclass of
6311         rtl_opt_pass along with...
6312         (pass_data_web): ...new pass_data instance and...
6313         (make_pass_web): ...new function.
6314         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
6315         declaration with that of...
6316         (make_pass_mode_switch_use): ...new function.
6317         (pass_resolve_sw_modes): Replace declaration with that of...
6318         (make_pass_resolve_sw_modes): ...new function.
6319         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
6320         from a global struct to a subclass of rtl_opt_pass along with...
6321         (pass_data_mode_switch_use): ...new pass_data instance and...
6322         (make_pass_mode_switch_use): ...new function.
6323         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
6324         from a global struct to a subclass of rtl_opt_pass along with...
6325         (pass_data_resolve_sw_modes): ...new pass_data instance and...
6326         (make_pass_resolve_sw_modes): ...new function.
6327         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
6328         struct to a subclass of rtl_opt_pass along with...
6329         (pass_data_insert_vzeroupper): ...new pass_data instance and...
6330         (make_pass_insert_vzeroupper): ...new function.
6331         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
6332         global struct to a subclass of rtl_opt_pass along with...
6333         (pass_data_work_around_errata): ...new pass_data instance and...
6334         (make_pass_work_around_errata): ...new function.
6335         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
6336         struct to a subclass of rtl_opt_pass along with...
6337         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
6338         (make_pass_mips_machine_reorg2): ...new function.
6340 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
6342         * passes.c (pass_manager::operator new): New.
6344 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
6346         Handwritten part of conversion of passes to C++ classes.
6348         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
6349         (toplev.o): Add dep on PASS_MANAGER_H.
6350         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
6351         of early local pases to reflect this moving from a global to a
6352         member of gcc::pass_manager.
6353         (cgraph_add_new_function): Likewise.
6354         * lto-cgraph.c (lto_output_node): Update for conversion of
6355         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
6356         * passes.c (opt_pass::clone): New.
6357         (opt_pass::gate): New.
6358         (opt_pass::execute): New.
6359         (opt_pass::opt_pass): New.
6360         (pass_manager::execute_early_local_passes): New.
6361         (pass_manager::execute_pass_mode_switching): new.
6362         (finish_optimization_passes): Convert to...
6363         (pass_manager::finish_optimization_passes): ...this.
6364         (finish_optimization_passes): Update for conversion of passes to
6365         C++ classes.
6366         (register_dump_files_1): Use has_gate since we cannot portably
6367         check a vtable entry against NULL.
6368         (dump_one_pass): Likewise.
6369         (ipa_write_summaries_2): Likewise.
6370         (ipa_write_optimization_summaries_1): Likewise.
6371         (ipa_read_summaries_1): Likewise.
6372         (ipa_read_optimization_summaries_1): Likewise.
6373         (execute_ipa_stmt_fixups): Likewise.
6374         (pass_manager::pass_manager): Rewrite pass-creation, invoking
6375         pass-creation functions rather than wiring up globals, and
6376         storing the results in fields of pass_manager generated using
6377         pass-instances.def.
6378         (pass_manager::dump_profile_report): Update for conversion of
6379         passes to C++ classes.
6380         (pass_manager::execute_ipa_summary_passes): Likewise.
6381         (execute_one_ipa_transform_pass): Likewise.
6382         (execute_one_pass): Use has_gate and has_execute since we cannot
6383         portably check a vtable entry against NULL.
6384         * pass_manager.h (pass_manager::finish_optimization_passes): New.
6385         (pass_manager): Use pass-instances.def to add fields for the
6386         various pass instances.
6387         * toplev.c (finalize): Update for move of
6388         finish_optimization_passes to a method of gcc::pass_manager.
6389         * toplev.h (finish_optimization_passes): Move to method of class
6390         pass_manager.
6391         * tree-pass.h (struct pass_data): New.
6392         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
6393         (opt_pass::gate): Convert to virtual function.
6394         (opt_pass::~opt_pass): New.
6395         (opt_pass::clone): New.
6396         (opt_pass::execute): Convert to virtual function.
6397         (opt_pass::opt_pass): New.
6398         (opt_pass::ctxt_): new.
6399         (gimple_opt_pass): Convert to subclass of opt_pass.
6400         (gimple_opt_pass::gimple_opt_pass): New.
6401         (rtl_opt_pass): Convert to subclass of opt_pass.
6402         (rtl_opt_pass::rtl_opt_pass): New.
6403         (ipa_opt_pass_d): Convert to subclass of opt_pass.
6404         (ipa_opt_pass_d::ipa_opt_pass_d): New.
6405         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
6406         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
6407         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
6408         invocation of pass_mode_switching to reflect this moving from a
6409         global to a member of gcc::pass_manager.
6410         (ix86_option_override): Rework how pass_insert_vzeroupper is
6411         added to the pass_manager to reflect autogenerated changes.
6412         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
6414 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
6416         PR rtl-optimization/57708
6417         * recog.c (peep2_find_free_register): Validate all regs in a
6418         multi-reg mode.
6420 2013-08-05  Jan Hubicka  <jh@suse.cz>
6422         PR lto/57602
6423         * cgraph.c (verify_cgraph_node): Accept local flags from other
6424         partitions.
6425         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
6426         (function_and_variable_visibility): Likewise.
6427         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
6429 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
6431         * graph.c (init_graph_slim_pretty_print): Remove.
6432         (print_graph_cfg): Do not call it.  Use local pretty printer.
6433         (start_graph_dump): Likewise.
6435 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
6437         * gimple-pretty-print.c (buffer): Remove.
6438         (initialized): Likewise.
6439         (maybe_init_pretty_print): Likewise.
6440         (print_gimple_stmt): Do not call it.  Use non-static local
6441         pretty_printer variable.
6442         (print_gimple_expr): Likewise.
6443         (print_gimple_seq): Likewise.
6444         (gimple_dump_bb): Likewise.
6446 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
6448         * asan.c (asan_pp): Remove.
6449         (asan_pp_initialized): Likewise.
6450         (asan_pp_initialize): Likewise.
6451         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
6452         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
6453         (asan_add_global): Likewise.
6455 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
6457         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
6458         * diagnostic.h (diagnostic_flush_buffer): Adjust.
6459         * pretty-print.c (pp_formatted_text_data): Likewise.
6460         (pp_indent): Rename from pp_base_indent.
6461         (pp_format): Rename from pp_base_format.
6462         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
6463         (pp_format_verbatim): Rename from pp_base_format_verbatim.
6464         (pp_flush): Rename from pp_base_flush.
6465         (pp_set_line_maximum_length): Rename from
6466         pp_base_set_line_maximum_length.
6467         (pp_clear_output_area): Rename from pp_base_clear_output_area.
6468         (pp_set_prefix): Rename from pp_base_set_prefix.
6469         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
6470         (pp_emit_prefix): Rename from pp_base_emit_prefix.
6471         (pp_append_text): Rename from pp_base_append_text.
6472         (pp_formatted_text): Rename from pp_base_formatted_text.
6473         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
6474         (pp_remaining_character_count_for_line): Rename from
6475         pp_base_remaining_character_count_for_line.
6476         (pp_newline): Rename from pp_base_newline.
6477         (pp_character): Rename from pp_base_character.
6478         (pp_string): Rename from pp_base_string.
6479         (pp_maybe_space): Rename from pp_base_maybe_space.
6480         * asan.c (asan_pp_string): Adjust.
6481         (asan_emit_stack_protection): Likewise.
6482         (asan_add_global): Likewise.
6483         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
6484         * tree-mudflap.c (mf_varname_tree): Likewise.
6485         * tree-pretty-print.c (pp_tree_identifier): Rename from
6486         pp_base_tree_identifier.
6487         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
6488         Declare as function.
6490 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
6492         * pretty-print.h (pp_bar_bar): New.
6493         (pp_ampersand_ampersand): Likewise.
6494         (pp_less_equal): Likewise.
6495         (pp_greater_equal): Likewise.
6496         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
6497         printer functions instead of pp_string or operators and punctuators.
6498         (dump_gimple_call): Likewise.
6499         (dump_gimple_omp_for): Likewise.
6500         (dump_gimple_transaction): Likewise.
6501         (dump_gimple_phi): Likewise.
6502         (pp_gimple_stmt_1): Likewise.
6503         * sched-vis.c (print_insn): Likewise.
6504         * tree-mudflap.c (mf_varname_tree): Likewise.
6505         * tree-pretty-print.c (dump_block_node): Likewise.
6506         (dump_generic_node): Likewise.
6508 2013-08-02  Jan Hubicka  <jh@suse.cz>
6510         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
6511         boundaries.
6512         * lto-streamer-out.c (tree_is_indexable): Results decls and
6513         parm decls are not indexable.
6514         (DFS_write_tree_body): Do not follow args and results.
6515         (hash_tree): Likewise.
6516         (output_functions): Rearrange so struct function is needed
6517         only when real body is output; be able to also ouptut abstract
6518         functions; output DECL_ARGUMENTS and DECL_RESULT.
6519         (lto_output): When not in WPA, ale store abstract functions.
6520         (write_symbol): Do not care about RESULT_DECL.
6521         (output_symbol_p): Handle correctly sbtract decls.
6522         * lto-streamer-in.c (input_function): Rearrange so struct
6523         function can be NULL at entry; allow streaming of
6524         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
6525         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
6526         sanity check during LTO.
6527         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
6528         RESULT_DECl and DECL_ARGUMENTS.
6529         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
6530         Likewise.
6532 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
6534         * pretty-print.h (pp_underscore): New.
6535         (pp_comma): Tidy.
6536         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
6537         printer functions instead of pp_character.
6538         (dump_binary_rhs): Likewise.
6539         (dump_ternary_rhs): Likewise.
6540         (dump_gimple_call_args): Likewise.
6541         (pp_points_to_solution): Likewise.
6542         (dump_gimple_call): Likewise.
6543         (dump_gimple_switch): Likewise.
6544         (dump_gimple_cond): Likewise.
6545         (dump_gimple_bind): Likewise.
6546         (dump_gimple_try): Likewise.
6547         (dump_gimple_omp_for): Likewise.
6548         (dump_gimple_omp_continue): Likewise.
6549         (dump_gimple_omp_single): Likewise.
6550         (dump_gimple_omp_sections): Likewise.
6551         (dump_gimple_omp_block): Likewise.
6552         (dump_gimple_omp_critical): Likewise.
6553         (dump_gimple_transaction): Likewise.
6554         (dump_gimple_asm): Likewise.
6555         (dump_gimple_phi): Likewise.
6556         (dump_gimple_omp_parallel): Likewise.
6557         (dump_gimple_omp_task): Likewise.
6558         (dump_gimple_omp_atomic_load): Likewise.
6559         (dump_gimple_omp_atomic_store): Likewise.
6560         (dump_gimple_mem_ops): Likewise.
6561         (pp_gimple_stmt_1): Likewise.
6562         (pp_cfg_jump): Likewise.
6563         (dump_implicit_edges): Likewise.
6564         (gimple_dump_bb_for_graph): Likewise.
6565         * graph.c (draw_cfg_node): Likewise.
6566         * langhooks.c (lhd_print_error_function): Likewise.
6567         * sched-vis.c (print_exp): Likewise.
6568         (print_value): Likewise.
6569         (print_pattern): Likewise.
6570         (print_insn): Likewise.
6571         (rtl_dump_bb_for_graph): Likewise.
6572         * tree-pretty-print.c (dump_function_declaration): Likewise.
6573         (dump_array_domain): Likewise.
6574         (dump_omp_clause): Likewise.
6575         (dump_location): Likewise.
6576         (dump_generic_node): Likewise.
6577         (print_struct_decl): Likewise.
6578         * diagnostic.c (diagnostic_show_locus): Use pp_space.
6580 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
6582         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
6583         candidate table when replacing a candidate statement.
6584         (replace_rhs_if_not_dup): Likewise.
6585         (replace_one_candidate): Likewise.
6587 2013-08-02  Jan Hubicka  <jh@suse.cz>
6588             Martin Liska  <marxin.liska@gmail.com>
6590         * cgraphunit.c (add_new_function): Fix logic when adding from
6591         late IPA pass.
6592         (assemble_thunk): Rename to ...
6593         (expand_thunk); .. this one; export; get it working with
6594         general functions; make produced gimple valid.
6595         * cgraph.h (expand_thunk): Declare.
6597 2013-08-02  Jan Hubicka  <jh@suse.cz>
6599         * ipa-cp.c (gather_context_independent_values): Use
6600         ipa_get_param_move_cost.
6601         (get_replacement_map): Remove PARAM; move parameter folding
6602         into tree-inline.c
6603         (create_specialized_node): Update.
6604         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
6605         assert that we have gimple body; update move_cost.
6606         (count_formal_params): Assert that we have gimple body.
6607         (ipa_dump_param): New function.
6608         (ipa_alloc_node_params): Break out from ...
6609         (ipa_initialize_node_params): ... here.
6610         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
6611         (ipa_write_node_info): Stream move costs.
6612         (ipa_read_node_info): Read move costs.
6613         (ipa_update_after_lto_read): Do not recompute node params.
6614         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
6615         (ipa_get_param): Check we are not in WPA.
6616         (ipa_get_param_move_cost): New.
6617         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
6618         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
6619         parm numbers to be present.
6621 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
6623         PR rtl-optimization/58048
6624         * lra-constraints.c (process_alt_operands): Don't check asm
6625         operand on register.
6627 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
6629         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
6630         the implied StoreLoad barrier for atomic operations if before.
6632 2013-08-02  Jan Hubicka  <jh@suse.cz>
6633             Martin Liska  <marxin.liska@gmail.com>
6635         * cgraph.c (cgraph_function_body_availability): Do not check
6636         cgraph flags.
6637         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
6638         symtab_node_availability): Declare.
6639         * ipa.c (can_replace_by_local_alias): New.
6640         (function_and_variable_visibility): Use it.
6641         * symtab.c (symtab_for_node_and_aliases,
6642         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
6644 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
6646         PR rtl-optimization/57963
6647         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
6648         (lra_constraints): Use them.
6650 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
6652         * config/arm/types.md (define_attr "type"): Add "load_acq"
6653         and "store_rel".
6654         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
6655         changes.
6656         (cortex_a53_store1): Likewise.
6658 2013-08-01  Jan Hubicka  <jh@suse.cz>
6660         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
6661         partitions are not needed.
6663 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
6665         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
6666         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
6667         MAYBE_NON_Q_CLASS_P where appropriate.
6669 2013-08-01  Jan Hubicka  <jh@suse.cz>
6671         * cgraph.h (release_function_body): Declare.
6672         * tree.c (free_lang_data_in_decl): Free, parameters and return values
6673         of unused delcarations.
6675 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6677         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
6678         RTL form when subtracting a constant.
6680 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6682         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
6683         Generate canonical plus rtx with negated immediate instead of minus
6684         where appropriate.
6685         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
6687 2013-08-01  Jan Hubicka  <jh@suse.cz>
6689         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
6690         (cgraph_release_function_body): Likewise.
6691         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
6692         * cgraph.h (cgrpah_node): Rename abstract_and_needed
6693         to used_as_abstract_origin.
6694         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
6695         symbols used as abstract origins.
6696         * cgraphunit.c (analyze_functions): Update.
6697         * ipa.c (symtab_remove_unreachable_nodes): Recompute
6698         used_as_abstract_origin.
6699         * tree-inline.c (tree_function_versioning): Update
6700         used_as_abstract_origin; be ready for DECL_RESULT and
6701         DECL_ARGUMENTS to be NULL.
6703         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
6704         for abstract functions.
6705         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
6706         real symbols.
6708 2013-08-01  Jan Hubicka  <jh@suse.cz>
6710         * profile.c (compute_value_histograms): Fix thinko.
6712 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
6714         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
6715         aarch-common-protos.h to extra_headers.
6716         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
6717         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
6718         * config/aarch64/t-aarch64 (aarch-common.o): Define.
6720 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
6722         * config/aarch64/aarch64.md (define_attr "type"): Delete.
6723         Include "../arm/types.md".  Define "type" attribute for all patterns.
6724         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
6725         attribute changes.
6727 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
6729         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
6730         to support power8 load fusion.
6731         (fusion_gpr_mem_load): Likewise.
6733         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
6735         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
6736         declarations for power8 load fusion.
6737         (emit_fusion_gpr_load): Likewise.
6739         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
6740         tuning for power8, turn on fusion mode by default.  Turn on sign
6741         extending fusion mode if normal fusion mode is on, and we are at
6742         -O2 or -O3.
6743         (fusion_gpr_load_p): New function, return true if we can fuse an
6744         addis instruction with a dependent load to a GPR.
6745         (emit_fusion_gpr_load): Emit the instructions for power8 load
6746         fusion to GPRs.
6748         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
6749         (VSX load fusion peepholes): New peepholes to fuse together an
6750         addi instruction with a VSX load instruction.
6752         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
6753         peepholes to fuse an addis instruction with a load to a GPR base
6754         register.  If we are supporting sign extending fusions, convert
6755         sign extending loads to zero extending loads and add an explicit
6756         sign extension.
6758 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
6760         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
6761         aarch-common-protos.h to extra_headers.
6762         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
6763         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
6764         (arm_early_store_addr_dep): Likewise.
6765         (arm_no_early_alu_shift_dep): Likewise.
6766         (arm_no_early_alu_shift_value_dep): Likewise.
6767         (arm_no_early_mul_dep): Likewise.
6768         (arm_no_early_store_addr_dep): Likewise.
6769         (arm_mac_accumulator_is_mul_result): Likewise.
6770         (arm_mac_accumulator_is_result): Likewise.
6771         * config/arm/aarch-common.c: ... here.  New file.
6772         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
6773         here to ...
6774         (arm_early_store_addr_dep): Likewise.
6775         (arm_no_early_alu_shift_dep): Likewise.
6776         (arm_no_early_alu_shift_value_dep): Likewise.
6777         (arm_no_early_mul_dep): Likewise.
6778         (arm_no_early_store_addr_dep): Likewise.
6779         (arm_mac_accumulator_is_mul_result): Likewise.
6780         (arm_mac_accumulator_is_result): Likewise.
6781         * config/arm/aarch-common-protos.h: ... here.  New file.
6782         * config/arm/t-arm (aarch-common.o): Define.
6784 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
6786         * config/arm/arm.md: Include new file "types.md".
6787         (define_attr "type"): Move from here to ...
6788         (define_attr "mul32"): Likewise.
6789         (define_attr "mul64"): Likewise.
6790         * config/arm/types.md: ... here.  New file.
6792 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
6794         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
6795         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
6797 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
6799         * gen-pass-instances.awk: Fix offset of substr().
6801 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
6803         * Makefile.in (pass-instances.def): New.
6804         (passes.o): Replace dependency on passes.def with one on
6805         pass-instances.def
6807         * gen-pass-instances.awk: New.
6809         * passes.c (pass_manager::pass_manager): Use pass-instances.def
6810         rather than passes.def, updating local definition of NEXT_PASS
6811         macro to add an extra NUM parameter (currently unused).
6813 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
6815         * Makefile.in (PASS_MANAGER_H): New.
6816         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
6817         (passes.o): Likewise.
6818         (statistics.o): Likewise.
6819         (cgraphunit.o): Likewise.
6820         (context.o): Depend on PASS_MANAGER_H.
6822         * pass_manager.h: New.
6824         * cgraphunit.c (cgraph_add_new_function): Update for moves
6825         of globals to fields of pass_manager.
6826         (analyze_function): Likewise.
6827         (expand_function): Likewise.
6828         (ipa_passes): Likewise.
6829         (compile): Likewise.
6831         * context.c (context::context): New.
6832         * context.h  (context::context): New.
6833         (context::get_passes): New.
6834         (context::passes_): New.
6836         * lto-cgraph.c (input_node): Update for moves of globals to
6837         fields of pass_manager.
6839         * passes.c (all_passes): Remove, in favor of a field of the
6840         same name within the new class pass_manager.
6841         (all_small_ipa_passes): Likewise.
6842         (all_lowering_passes): Likewise.
6843         (all_regular_ipa_passes): Likewise.
6844         (all_late_ipa_passes): Likewise.
6845         (all_lto_gen_passes): Likewise.
6846         (passes_by_id): Likewise.
6847         (passes_by_id_size): Likewise.
6848         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
6849         the new class pass_manager.
6850         (set_pass_for_id): Convert to...
6851         (pass_manager::set_pass_for_id): ...method.
6852         (get_pass_for_id): Convert to...
6853         (pass_manager::get_pass_for_id): ...method.
6854         (register_one_dump_file): Move body of implementation into...
6855         (pass_manager::register_one_dump_file): ...here.
6856         (register_dump_files_1): Convert to...
6857         (pass_manager::register_dump_files_1): ...method.
6858         (register_dump_files): Convert to...
6859         (pass_manager::register_dump_files): ...method.
6860         (create_pass_tab): Update for moves of globals to fields of
6861         pass_manager.
6862         (dump_passes): Move body of implementation into...
6863         (pass_manager::dump_passes): ...here.
6864         (register_pass): Move body of implementation into...
6865         (pass_manager::register_pass): ...here.
6866         (init_optimization_passes): Convert into...
6867         (pass_manager::pass_manager): ...constructor for new
6868         pass_manager class, and initialize the pass_lists array.
6869         (check_profile_consistency): Update for moves of globals to
6870         fields of pass_manager.
6871         (dump_profile_report): Move body of implementation into...
6872         (pass_manager::dump_profile_report): ...here.
6873         (ipa_write_summaries_1): Update for moves of pass lists from
6874         being globals to fields of pass_manager.
6875         (ipa_write_optimization_summaries): Likewise.
6876         (ipa_read_summaries):  Likewise.
6877         (ipa_read_optimization_summaries): Likewise.
6878         (execute_all_ipa_stmt_fixups): Likewise.
6880         * statistics.c (statistics_fini): Update for moves of globals to
6881         fields of pass_manager.
6883         * toplev.c (general_init): Replace call to
6884         init_optimization_passes with construction of the pass_manager
6885         instance.
6887         * tree-pass.h (all_passes): Remove, in favor of a field of the
6888         same name within the new class pass_manager.
6889         (all_small_ipa_passes): Likewise.
6890         (all_lowering_passes): Likewise.
6891         (all_regular_ipa_passes): Likewise.
6892         (all_lto_gen_passes): Likewise.
6893         (all_late_ipa_passes): Likewise.
6894         (passes_by_id): Likewise.
6895         (passes_by_id_size): Likewise.
6896         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
6897         the new class pass_manager.
6898         (get_pass_for_id): Remove.
6900 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
6902         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
6903         configs.
6905 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
6907         * arm.md (mulhi3): New expand pattern.
6909 2013-07-30  Jan Hubicka  <jh@suse.cz>
6910             Martin Liska  <marxin.liska@gmail.com>
6912         * profile.c (compute_value_histograms): Do not ICE when
6913         there is mismatch only on some counters.
6915 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
6917         PR rtl-optimization/57637
6918         * function.c (move_insn_for_shrink_wrap): Also check the
6919         GEN set of the LIVE problem for the liveness analysis
6920         if it exists, otherwise give up.
6922 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
6924         PR tree-optimization/57993
6925         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
6926         replaced statement in the candidate table.
6927         (phi_add_costs): Return infinite cost when the hidden basis does
6928         not dominate all phis on which the candidate is dependent.
6929         (replace_one_candidate): Record replaced statement in the
6930         candidate table.
6932 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
6934         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
6935         (ashlv2si3): New expander.
6936         (*ashlv2si3_i): New define_insn_and_split.
6937         * predicates.md (float_operation): Allow patterns with three
6938         basic sub-patterns.
6940         PR rtl-optimization/58021
6941         * mode-switching.c (create_pre_exit): Always split off preceding
6942         insns if we are not at the basic block head.
6944 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
6946         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
6947         (UCLIBC_DYNAMIC_LINKER): New macro.
6948         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
6949         `-mnan=2008'.
6950         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
6951         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
6952         `-mnan=2008'.
6953         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
6954         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
6955         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
6956         for SF and DF modes.  Use ieee_quad_format for TF mode.
6957         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
6958         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
6959         (mips_option_override): Handle `-mnan=legacy'.
6960         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
6961         `-mabs=2008' and `-mnan=2008'.
6962         (OPTION_DEFAULT_SPECS): Add "nan" default.
6963         (ASM_SPEC): Handle `-mnan='.
6964         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
6965         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
6966         comment accordingly.
6967         (neg<mode>2): Likewise.
6968         * config/mips/mips.opt (mabs, mnan): New options.
6969         * doc/install.texi (Configuration): Document `--with-nan=' option.
6970         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
6971         `-mnan=' options.
6972         (MIPS Options): Document them.
6973         * config.gcc <mips*-*-*>: Handle `--with-nan='.
6974         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
6975         * configure: Regenerate.
6976         * config.in: Regenerate.
6978 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
6980         * config/i386/i386.md (float post-reload splitters): Do not check
6981         for subregs of SSE registers.
6983 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
6984             H.J. Lu  <hongjiu.lu@intel.com>
6986         PR target/57954
6987         PR target/57988
6988         * config/i386/i386.md (post-reload splitter
6989         to avoid partial SSE reg dependency stalls): New pattern.
6991 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
6993         * config/s390/s390.md ("movcc"): Swap load and store instructions.
6995 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
6997         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
6998         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
7000 2013-07-26  Cary Coutant  <ccoutant@google.com>
7002         * dwarf2out.c (die_checksum_ordered): Don't include template
7003         instantiations in signature.
7004         (is_template_parameter): New function.
7005         (is_template_instantiation): New function.
7006         (generate_skeleton_bottom_up): Don't include template instantiations
7007         in type unit DIE.
7008         (generate_skeleton): Likewise.
7009         (break_out_comdat_types): Move recursive call to break out nested
7010         types earlier.
7011         (prune_unused_types_mark_generic_parms_dies): Call
7012         is_template_parameter.
7014 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
7016         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
7017         uses vector registers.
7018         * config/aarch64/iterators.md: Add attributes rtn and vas.
7020 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7021             Richard Earnshaw  <richard.earnshaw@arm.com>
7023         * combine.c (simplify_comparison): Re-canonicalize operands
7024         where appropriate.
7025         * config/arm/arm.md (movcond_addsi): New splitter.
7027 2013-07-25  Sterling Augustine  <saugustine@google.com>
7029         * dwarf2out.c (size_of_pubnames): Move code to...
7030         (include_pubname_in_output): ...here.  New.
7031         (want_pubnames): Rearrange.
7032         (output_pubnames): Call include_pubname_in_output.  Move assertion.
7034 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
7036         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
7038 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
7040         PR target/38836
7041         * doc/extend.texi: Remove obsolete builtins. Fix
7042         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
7044 2013-07-25  Jan Hubicka  <jh@suse.cz>
7046         * cgraph.c (release_function_body): Break out from ...
7047         (cgraph_release_function_body): ... this one; also release DECL_RESULT
7048         and DECL_ARGUMENTS.
7049         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
7050         old_tree in the map.
7051         (create_specialized_node): Update.
7052         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
7053         into index.
7054         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
7055         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
7056         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
7057         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
7058         DECL_RESULT.
7060 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7062         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
7063         addsi3_carryin_alt2_<optab>): Correct output template.
7065 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7067         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
7068         Adjust for arm_restrict_it.
7069         Remove trailing whitespace.
7071 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
7073         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
7074         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
7076         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
7078 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
7080         PR rtl-optimization/57960
7081         * lra-constraints.c (process_alt_operands): Use the right mode
7082         when checking strict_low.
7084 2013-07-25  Jan Hubicka  <jh@suse.cz>
7086         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
7087         * cgraph.c (cgraph_remove_node): Do not release function body
7088         when in cgraph streaming.
7089         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
7090         in other partitions are not considered reachable; fix handling of
7091         clones.
7093 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7095         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
7097 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7099         PR target/19599
7100         PR target/57731
7101         PR target/57837
7102         * config/arm/arm.md ("*sibcall_insn): Replace use of
7103         Ss with US. Adjust output for v5 and v4t.
7104         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
7106         * config/arm/constraints.md ("Ss"): Rename to US.
7108 2013-07-25  Terry Guo  <terry.guo@arm.com>
7110         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
7111         shift_add/shift_sub0/shift_sub1 RTXs.
7113 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
7114             Anton Blanchard  <anton@au1.ibm.com>
7116         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
7117         (altivec_vpks<VI_char>ss): Likewise.
7118         (altivec_vpks<VI_char>us): Likewise.
7119         (altivec_vpku<VI_char>us): Likewise.
7120         (altivec_vpku<VI_char>um): Likewise.
7122 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
7124         Introduce context class.
7126         * Makefile.in (CONTEXT_H): New.
7127         (OBJS): Add context.o.
7128         (toplev.o): Add CONTEXT_H to dependencies.
7129         (context.o): New.
7131         * toplev.c (general_init): Create the singleton gcc::context instance.
7133         * context.c: New.
7135         * context.h: New.
7137 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
7139         PR rtl-optimization/57968
7140         * mode-switching.c (create_pre_exit): Allow instructions that
7141         don't set a return register to need a non-exit mode.
7143 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
7144             Anton Blanchard  <anton@au1.ibm.com>
7146         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
7147         operands to vperm for little endian.
7148         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
7149         of lvsl to create the control mask for a vperm for little endian.
7151 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7152             Anton Blanchard  <anton@au1.ibm.com>
7154         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
7155         two operands for little-endian.
7157 2013-07-23  Steve Ellcey  <sellcey@mips.com>
7159         * config/mips/mips.c (mips_case_values_threshold): New.
7160         (TARGET_CASE_VALUES_THRESHOLD): Define.
7162 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7163             Anton Blanchard  <anton@au1.ibm.com>
7165         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
7166         selection of field for vector splat in little endian mode.
7168 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
7170         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
7171         expanders to rs6000.md.
7172         (ior<mode>3): Likewise.
7173         (and<mode>3): Likewise.
7174         (one_cmpl<mode>2): Likewise.
7175         (nor<mode>3): Likewise.
7176         (andc<mode>3): Likewise.
7177         (eqv<mode>3): Likewise.
7178         (nand<mode>3): Likewise.
7179         (orc<mode>3): Likewise.
7181         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
7182         declaration.
7184         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
7185         to split multi-word logical operations.
7186         (rs6000_split_logical_di): Likewise.
7187         (rs6000_split_logical): Likewise.
7189         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
7190         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
7191         and allow TImode operations in 32-bit.
7192         (vsx_and<mode>3_64bit): Likewise.
7193         (vsx_ior<mode>3_32bit): Likewise.
7194         (vsx_ior<mode>3_64bit): Likewise.
7195         (vsx_xor<mode>3_32bit): Likewise.
7196         (vsx_xor<mode>3_64bit): Likewise.
7197         (vsx_one_cmpl<mode>2_32bit): Likewise.
7198         (vsx_one_cmpl<mode>2_64bit): Likewise.
7199         (vsx_nor<mode>3_32bit): Likewise.
7200         (vsx_nor<mode>3_64bit): Likewise.
7201         (vsx_andc<mode>3_32bit): Likewise.
7202         (vsx_andc<mode>3_64bit): Likewise.
7203         (vsx_eqv<mode>3_32bit): Likewise.
7204         (vsx_eqv<mode>3_64bit): Likewise.
7205         (vsx_nand<mode>3_32bit): Likewise.
7206         (vsx_nand<mode>3_64bit): Likewise.
7207         (vsx_orc<mode>3_32bit): Likewise.
7208         (vsx_orc<mode>3_64bit): Likewise.
7210         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
7211         logical types in GPRs.
7213         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
7214         logical insns to rs6000.md, and allow TImode operations in
7215         32-bit.
7216         (altivec_ior<mode>3): Likewise.
7217         (altivec_xor<mode>3): Likewise.
7218         (altivec_one_cmpl<mode>2): Likewise.
7219         (altivec_nor<mode>3): Likewise.
7220         (altivec_andc<mode>3): Likewise.
7222         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
7223         attributes for moving the 128-bit logical operations into
7224         rs6000.md.
7225         (BOOL_REGS_OUTPUT): Likewise.
7226         (BOOL_REGS_OP1): Likewise.
7227         (BOOL_REGS_OP2): Likewise.
7228         (BOOL_REGS_UNARY): Likewise.
7229         (BOOL_REGS_AND_CR0): Likewise.
7230         (one_cmpl<mode>2): Add support for DI logical operations on
7231         32-bit, splitting the operations to 32-bit.
7232         (anddi3): Likewise.
7233         (iordi3): Likewise.
7234         (xordi3): Likewise.
7235         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
7236         changes to combine the 32/64-bit code, allow logical operations on
7237         TI mode in 32-bit, and to use similar match_operator patterns like
7238         scalar mode uses.  Combine the Altivec and VSX code for logical
7239         operations, and move it here.
7240         (ior<mode>3, 128-bit types): Likewise.
7241         (xor<mode>3, 128-bit types): Likewise.
7242         (one_cmpl<mode>3, 128-bit types): Likewise.
7243         (nor<mode>3, 128-bit types): Likewise.
7244         (andc<mode>3, 128-bit types): Likewise.
7245         (eqv<mode>3, 128-bit types): Likewise.
7246         (nand<mode>3, 128-bit types): Likewise.
7247         (orc<mode>3, 128-bit types): Likewise.
7248         (and<mode>3_internal): Likewise.
7249         (bool<mode>3_internal): Likewise.
7250         (boolc<mode>3_internal1): Likewise.
7251         (boolc<mode>3_internal2): Likewise.
7252         (boolcc<mode>3_internal1): Likewise.
7253         (boolcc<mode>3_internal2): Likewise.
7254         (eqv<mode>3_internal1): Likewise.
7255         (eqv<mode>3_internal2): Likewise.
7256         (one_cmpl1<mode>3_internal): Likewise.
7258 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
7260         * config/microblaze/microblaze.c (microblaze_expand_prologue):
7261         Rename flag_stack_usage to flag_stack_usage_info.
7263 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
7265         * config/microblaze/sync.md: New file.
7266         * config/microblaze/microblaze.md: Include sync.md
7267         * config/microblaze/microblaze.c: Add print_operand 'y'.
7268         * config/microblaze/constraints.md: Add memory_contraint
7269         'Q' which is a single register.
7271 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
7273         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
7275 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
7277         * reload.c (find_reloads): Exit loop once we find this operand
7278         cannot be reloaded somehow for this alternative.
7280         * reload.c (find_reloads): Exit loop once we find a hard register.
7282         * rtlanal.c (computed_jump_p): Exit loop once we find label
7283         reference is used.
7285         * i386.c (ix86_pad_returns): Exit loop after setting replace.
7287         * cfgloopmanip.c (remove_path): Exit loop after setting
7288         irred_invalidated.
7290         * gensupport.c (subst_dup): Avoid loop if code is not
7291         MATCH_DUP nor MATCH_OP_DUP.
7293 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
7295         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
7297 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
7299         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
7300         true for SP_REGNUM if mode == ptr_mode.
7301         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
7302         with value R0_REGNUM + 31.
7304 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
7306         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
7307         pad pointer-typed argument downward.
7309 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
7311         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
7312         and __ILP32__ when the ILP32 model is in use.
7314 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
7316         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
7317         (aarch64_load_symref_appropriately): In the case of
7318         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
7319         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
7320         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
7321         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
7322         if 'mode' doesn't equal to 'ptr_mode'.
7323         (aarch64_output_mi_thunk): Add an assertion on the alignment of
7324         'vcall_offset'; change to call aarch64_emit_move differently depending
7325         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
7326         to calculate the upper bound of 'vcall_offset'.
7327         (aarch64_cannot_force_const_mem): Change to also return true if
7328         mode != ptr_mode.
7329         (aarch64_legitimize_reload_address): In the case of large
7330         displacements, add new local variable 'xmode' and an assertion
7331         based on it; change to use 'xmode' to generate the new rtx and
7332         reload.
7333         (aarch64_asm_trampoline_template): Change to generate the template
7334         differently depending on TARGET_ILP32 or not; change to use
7335         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
7336         (aarch64_trampoline_size): Removed.
7337         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
7338         and replace immediate literals with it.  Change to use 'ptr_mode'
7339         instead of 'DImode' and call convert_memory_address if the mode
7340         of 'fnaddr' doesn't equal to 'ptr_mode'.
7341         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
7342         to output symbol.
7343         (aarch64_elf_asm_destructor): Likewise.
7344         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
7345         on TARGET_ILP32 instead of aarch64_trampoline_size.
7346         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
7347         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
7348         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
7349         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
7350         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
7351         (storewb_pair<GPI:mode>_<P:mode>): ... this.
7352         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
7353         depending on the value of 'mode'.
7354         (add_losym_<mode>): New.
7355         (ldr_got_small_<mode>): New, based on ldr_got_small.
7356         (ldr_got_small): Remove.
7357         (ldr_got_small_sidi): New.
7358         * config/aarch64/iterators.md (P): New.
7359         (PTR): Change to 'ptr_mode' in the condition.
7361 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
7363         * config.gcc (aarch64*-*-*): Support --with-abi.
7364         (aarch64*-*-elf): Support --with-multilib-list.
7365         (aarch64*-*-linux*): Likewise.
7366         (supported_defaults): Add abi to aarch64*-*-*.
7367         * configure.ac: Mention AArch64 for --with-multilib-list.
7368         * configure: Re-generated.
7369         * config/aarch64/biarchilp32.h: New file.
7370         * config/aarch64/biarchlp64.h: New file.
7371         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
7372         (ABI_SPEC): Ditto.
7373         (MULTILIB_DEFAULTS): Ditto.
7374         (DRIVER_SELF_SPECS): Ditto.
7375         (ASM_SPEC): Update to also substitute -mabi.
7376         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
7377         file whose name depends on -mabi= and -mbig-endian.
7378         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
7379         TARGET_ILP32.
7380         (POINTER_SIZE): New define.
7381         (POINTERS_EXTEND_UNSIGNED): Ditto.
7382         (enum aarch64_abi_type): New enumeration tag.
7383         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
7384         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
7385         (TARGET_ILP32): New define.
7386         * config/aarch64/aarch64.opt (mabi): New.
7387         (aarch64_abi): New.
7388         (ilp32, lp64): New values for -mabi.
7389         * config/aarch64/t-aarch64 (comma): New define.
7390         (MULTILIB_OPTIONS): Ditto.
7391         (MULTILIB_DIRNAMES): Ditto.
7392         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
7393         * doc/invoke.texi: Document -mabi for AArch64.
7395 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
7397         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
7399 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
7400             Anton Blanchard  <anton@au1.ibm.com>
7402         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
7403         endianness when selecting field to splat.
7405 2013-07-22  Eric Christopher  <echristo@gmail.com>
7407         * dwarf2out.c (die_odr_checksum): New function to use
7408         CHECKSUM_ macros and ULEB128 for DIE tag.
7409         (generate_type_signature): Use.
7411 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
7413         * config.gcc (sparc*-*-*): Accept leon3 processor.
7414         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
7415         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
7416         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
7417         * config/sparc/sparc.opt (enum processor_type): Add leon3.
7418         (mfix-ut699): Adjust comment.
7419         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
7420         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
7421         (CPP_CPU_SPEC): Likewise.
7422         (ASM_CPU_SPEC): Likewise.
7423         * config/sparc/sparc.c (leon3_cost): New constant.
7424         (sparc_option_override): Add leon3 support.
7425         (mem_ref): New function.
7426         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
7427         (sparc_do_work_around_errata): Look into the instruction in the delay
7428         slot and adjust accordingly.  Add fix for the data cache nullify issues
7429         of the UT699.  Change insertion position for the NOP.
7430         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
7431         (leon3_load): New reservation.
7432         (leon_store): Bump latency to 2.
7433         (grfpu): New automaton.
7434         (grfpu_alu): New unit.
7435         (grfpu_ds): Likewise.
7436         (leon_fp_alu): Adjust.
7437         (leon_fp_mult): Delete.
7438         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
7439         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
7440         * config/sparc/sparc.md (cpu): Add leon3.
7441         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
7442         (swapsi): Likewise.
7443         (atomic_test_and_set): Likewise.
7444         (ldstub): Likewise.
7446 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
7448         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
7449         default for R5900 targets.
7450         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
7451         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
7452         * config/mips/mips.c (mips_option_override): Report an error for
7453         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
7454         for -march=r5900 -mhard-float.
7456 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
7458         * df-problems.c (can_move_insns_across): Exit loop once we
7459         find a non-fixed, non-global register.
7461         * ipa-pure-const.c (propagate_nothrow): Exit loop after
7462         setting can_throw.
7464         * omega.c (omega_eliminate_red): Break after setting red_found.
7465         (omega_problem_has_red_equations): Similarly after setting found.
7466         (omega_query_variable): Similarly after setting coupled.
7468 2013-07-22  Marek Polacek  <polacek@redhat.com>
7470         * gimplify.c: Don't include gimple.h twice.
7472 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7474         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
7475         instead of TARGET_THUMB1.
7476         (Pz): New constraint.
7477         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
7478         encodings.
7479         (compare_negsi_si): Likewise.
7480         (compare_addsi2_op0): Likewise.
7481         (compare_addsi2_op1): Likewise.
7482         (addsi3_carryin_<optab>): Likewise.
7483         (addsi3_carryin_alt2_<optab>): Likewise.
7484         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
7485         for arm_restrict_it.
7486         (subsi3_carryin): Likewise.
7487         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
7488         (minmax_arithsi): Disable for arm_restrict_it.
7489         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
7490         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
7491         (satsi_<SAT:code>_shift): Likewise.
7492         (arm_shiftsi3): Add alternative for 16-bit encoding.
7493         (arm32_movhf): Disable for arm_restrict_it.
7494         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
7495         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
7497 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
7499         * config/arm/arm.md (attribute "insn"): Delete.
7500         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
7501         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
7502         (not_shiftsi): Update for attribute change.
7503         (not_shiftsi_compare0): Likewise.
7504         (not_shiftsi_compare0_scratch): Likewise.
7505         (arm_one_cmplsi2): Likewise.
7506         (thumb1_one_cmplsi2): Likewise.
7507         (notsi_compare0): Likewise.
7508         (notsi_compare0_scratch): Likewise.
7509         (thumb1_movdi_insn): Likewise.
7510         (arm_movsi_insn): Likewise.
7511         (movhi_insn_arch4): Likewise.
7512         (movhi_bytes): Likewise.
7513         (arm_movqi_insn): Likewise.
7514         (thumb1_movqi_insn): Likewise.
7515         (arm32_movhf): Likewise.
7516         (thumb1_movhf): Likewise.
7517         (arm_movsf_soft_insn): Likewise.
7518         (thumb1_movsf_insn): Likewise.
7519         (thumb_movdf_insn): Likewise.
7520         (movsicc_insn): Likewise.
7521         (movsfcc_soft_insn): Likewise.
7522         (and_scc): Likewise.
7523         (cond_move): Likewise.
7524         (if_move_not): Likewise.
7525         (if_not_move): Likewise.
7526         (if_shift_move): Likewise.
7527         (if_move_shift): Likewise.
7528         (if_shift_shift): Likewise.
7529         (if_not_arith): Likewise.
7530         (if_arith_not): Likewise.
7531         (cond_move_not): Likewise.
7532         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
7533         (neon_mov<mode>): Likewise.
7534         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
7535         (thumb2_movsi_vfp): Likewise.
7536         (movsf_vfp): Likewise.
7537         (thumb2_movsf_vfp): Likewise.
7538         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
7539         change.
7540         (cortexa7_older_only): Likewise.
7541         (cortexa7_younger): Likewise.
7542         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
7543         (1020alu_shift_op): Likewise.
7544         (1020alu_shift_reg_op): Likewise.
7545         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
7546         (alu_shift_op): Likewise.
7547         (alu_shift_reg_op): Likewise.
7548         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
7549         (11_alu_shift_op): Likewise.
7550         (11_alu_shift_reg_op): Likewise.
7551         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
7552         (9_alu_shift_reg_op): Likewise.
7553         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7554         change.
7555         (cortex_a15_alu_shift): Likewise.
7556         (cortex_a15_alu_shift_reg): Likewise.
7557         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
7558         (cortex_a5_alu_shift): Likewise.
7559         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
7560         change.
7561         (cortex_a53_alu_shift): Likewise.
7562         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
7563         change.
7564         (cortex_a7_alu_reg): Likewise.
7565         (cortex_a7_alu_shift): Likewise.
7566         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
7567         (cortex_a8_alu_shift): Likewise.
7568         (cortex_a8_alu_shift_reg): Likewise.
7569         (cortex_a8_mov): Likewise.
7570         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
7571         (cortex_a9_dp_shift): Likewise.
7572         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
7573         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
7574         (cortex_r4_mov): Likewise.
7575         (cortex_r4_alu_shift): Likewise.
7576         (cortex_r4_alu_shift_reg): Likewise.
7577         * config/arm/fa526.md (526_alu_op): Update for attribute change.
7578         (526_alu_shift_op): Likewise.
7579         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
7580         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
7581         (626te_alu_shift_op): Likewise.
7582         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
7583         (726te_alu_op): Likewise.
7584         (726te_alu_shift_op): Likewise.
7585         (726te_alu_shift_reg_op): Likewise.
7586         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
7587         (mp626_alu_shift_op): Likewise.
7588         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
7589         (pj4_alu_e1_conds): Likewise.
7590         (pj4_alu): Likewise.
7591         (pj4_alu_conds): Likewise.
7592         (pj4_shift): Likewise.
7593         (pj4_shift_conds): Likewise.
7594         (pj4_alu_shift): Likewise.
7595         (pj4_alu_shift_conds): Likewise.
7597 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7599         * config/arm/predicates.md (shiftable_operator_strict_it):
7600         New predicate.
7601         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
7602         Disable cond_exec version for arm_restrict_it.
7603         (thumb2_smaxsi3): Convert to generate cond_exec.
7604         (thumb2_sminsi3): Likewise.
7605         (thumb32_umaxsi3): Likewise.
7606         (thumb2_uminsi3): Likewise.
7607         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
7608         (thumb2_neg_abssi2): Likewise.
7609         (thumb2_mov_scc): Add alternative for 16-bit encoding.
7610         (thumb2_movsicc_insn): Adjust alternatives.
7611         (thumb2_mov_negscc): Disable for arm_restrict_it.
7612         (thumb2_mov_negscc_strict_it): New pattern.
7613         (thumb2_mov_notscc_strict_it): New pattern.
7614         (thumb2_mov_notscc): Disable for arm_restrict_it.
7615         (thumb2_ior_scc): Likewise.
7616         (thumb2_ior_scc_strict_it): New pattern.
7617         (thumb2_cond_move): Adjust for arm_restrict_it.
7618         (thumb2_cond_arith): Disable for arm_restrict_it.
7619         (thumb2_cond_arith_strict_it): New pattern.
7620         (thumb2_cond_sub): Adjust for arm_restrict_it.
7621         (thumb2_movcond): Likewise.
7622         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
7623         (thumb2_zero_extendhisi2_v6): Likewise.
7624         (thumb2_zero_extendqisi2_v6): Likewise.
7625         (orsi_notsi_si): Likewise.
7626         (orsi_not_shiftsi_si): Likewise.
7628 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
7630         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
7631         instruction sequence is 1 byte shorter.
7633 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
7635         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
7636         it is not needed after split.
7638 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
7640         PR target/51784
7641         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
7642         second label for nonlocal goto receivers. Don't output pic base labels
7643         unless we're producing PIC; mark that action unreachable().
7644         (ix86_save_reg): If the function contains a nonlocal label, save the
7645         PIC base reg.
7646         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
7647         * config/darwin.c (emitted_pic_label_num): New GTY.
7648         (update_pic_label_number_if_needed): New.
7649         (machopic_output_function_base_name): Adjust for nonlocal receiver
7650         case.
7651         (machopic_should_output_picbase_label): New.
7652         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
7653         (nonlocal_goto_receiver): New insn and split.
7655 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
7657         * config/aarch64/aarch64-builtins.c
7658         (aarch64_fold_builtin): Fold abs in all modes.
7659         * config/aarch64/aarch64-simd-builtins.def
7660         (abs): Enable for all modes.
7661         * config/aarch64/arm_neon.h
7662         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
7663         (vabs_f64): Add missing intrinsic.
7665 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
7667         * config/aarch64/arm_neon.h (vabs_s64): New function
7669 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
7671         PR target/57516
7672         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
7673         * config/avr/avr.md (adjust_len): Add `round'.
7674         * config/avr/avr-protos.h (avr_out_round): New prototype.
7675         (avr_out_plus): Add `out_label' argument.
7676         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
7677         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
7678         Handle the case where `insn' is just a pattern.
7679         (avr_out_bitop): Handle the case where `insn' is just a pattern.
7680         (avr_out_round): New function.
7681         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
7683 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
7685         * config/microblaze/microblaze.c (microblaze_expand_prologue):
7686         Add check for flag_stack_usage to handle -fstack-usage support
7688 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
7690         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
7691         interaction for new Power8 flags and VSX.
7693 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
7695         PR middle-end/57698
7696         * tree-inline.c (expand_call_inline): Emit errors during
7697         early_inlining only if optimization is not turned on.
7699 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
7701         * passes.def: New.
7703         * passes.c (init_optimization_passes): Move the construction of
7704         the pass hierarchy into a new passes.def file.
7706         * Makefile.in (passes.o): Add dependency on passes.def.
7708 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
7710         * passes.c (init_optimization_passes): Introduce macros for
7711         constructing the tree of passes (INSERT_PASSES_AFTER,
7712         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
7713         TERMINATE_PASS_LIST).
7715 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
7716             Wei Mi  <wmi@google.com>
7718         PR rtl-optimization/57878
7719         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
7720         top.
7721         (reload_pseudo_compare_func): Check nregs first for reload
7722         pseudos.
7724 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
7726         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
7728 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
7730         * read-rtl.c (validate_const_int): Once an invalid character is
7731         seen, quit the loop.
7733         * gengtype.c (write_roots): Similarly once we find the "deletable"
7734         or "if_marked" option.
7736 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
7738         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
7739         "xtab" and "sat".  Move value "clz" from here to ...
7740         (attriubte "type"): ... here.
7741         (satsi_<SAT:code>): Delete "insn" attribute.
7742         (satsi_<SAT:code>_shift): Likewise.
7743         (arm_zero_extendqisi2addsi): Likewise.
7744         (arm_extendqisi2addsi): Likewise.
7745         (clzsi2): Update for attribute changes.
7746         (rbitsi2): Likewise.
7747         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
7748         attribute.
7749         (arm_usatsihi): Likewise.
7750         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
7752 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
7754         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
7755         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
7756         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
7757         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
7758         in alphabetical order.
7759         (attribute "core_cycles"): Update for attribute changes.
7760         (arm_addsi3): Likewise.
7761         (addsi3_compare0): Likewise.
7762         (addsi3_compare0_scratch): Likewise.
7763         (addsi3_compare_op1): Likewise.
7764         (addsi3_compare_op2): Likewise.
7765         (compare_addsi2_op0): Likewise.
7766         (compare_addsi2_op1): Likewise.
7767         (addsi3_carryin_shift_<optab>): Likewise.
7768         (subsi3_carryin_shift): Likewise.
7769         (rsbsi3_carryin_shift): Likewise.
7770         (arm_subsi3_insn): Likewise.
7771         (subsi3_compare0): Likewise.
7772         (subsi3_compare): Likewise.
7773         (arm_andsi3_insn): Likewise.
7774         (thumb1_andsi3_insn): Likewise.
7775         (andsi3_compare0): Likewise.
7776         (andsi3_compare0_scratch): Likewise.
7777         (zeroextractsi_compare0_scratch
7778         (andsi_not_shiftsi_si): Likewise.
7779         (iorsi3_insn): Likewise.
7780         (iorsi3_compare0): Likewise.
7781         (iorsi3_compare0_scratch): Likewise.
7782         (arm_xorsi3): Likewise.
7783         (thumb1_xorsi3_insn): Likewise.
7784         (xorsi3_compare0): Likewise.
7785         (xorsi3_compare0_scratch): Likewise.
7786         (satsi_<SAT:code>_shift): Likewise.
7787         (rrx): Likewise.
7788         (arm_shiftsi3): Likewise.
7789         (shiftsi3_compare0): Likewise.
7790         (not_shiftsi): Likewise.
7791         (not_shiftsi_compare0): Likewise.
7792         (not_shiftsi_compare0_scratch): Likewise.
7793         (arm_one_cmplsi2): Likewise.
7794         (thumb_one_complsi2): Likewise.
7795         (notsi_compare0): Likewise.
7796         (notsi_compare0_scratch): Likewise.
7797         (thumb1_zero_extendhisi2): Likewise.
7798         (arm_zero_extendhisi2): Likewise.
7799         (arm_zero_extendhisi2_v6): Likewise.
7800         (arm_zero_extendhisi2addsi): Likewise.
7801         (thumb1_zero_extendqisi2): Likewise.
7802         (thumb1_zero_extendqisi2_v6): Likewise.
7803         (arm_zero_extendqisi2): Likewise.
7804         (arm_zero_extendqisi2_v6): Likewise.
7805         (arm_zero_extendqisi2addsi): Likewise.
7806         (thumb1_extendhisi2): Likewise.
7807         (arm_extendhisi2): Likewise.
7808         (arm_extendhisi2_v6): Likewise.
7809         (arm_extendqisi): Likewise.
7810         (arm_extendqisi_v6): Likewise.
7811         (arm_extendqisi2addsi): Likewise.
7812         (thumb1_extendqisi2): Likewise.
7813         (thumb1_movdi_insn): Likewise.
7814         (arm_movsi_insn): Likewise.
7815         (movsi_compare0): Likewise.
7816         (movhi_insn_arch4): Likewise.
7817         (movhi_bytes): Likewise.
7818         (arm_movqi_insn): Likewise.
7819         (thumb1_movqi_insn): Likewise.
7820         (arm32_movhf): Likewise.
7821         (thumb1_movhf): Likewise.
7822         (arm_movsf_soft_insn): Likewise.
7823         (thumb1_movsf_insn): Likewise.
7824         (movdf_soft_insn): Likewise.
7825         (thumb_movdf_insn): Likewise.
7826         (arm_cmpsi_insn): Likewise.
7827         (cmpsi_shiftsi): Likewise.
7828         (cmpsi_shiftsi_swp): Likewise.
7829         (arm_cmpsi_negshiftsi_si): Likewise.
7830         (movsicc_insn): Likewise.
7831         (movsfcc_soft_insn): Likewise.
7832         (arith_shiftsi): Likewise.
7833         (arith_shiftsi_compare0
7834         (arith_shiftsi_compare0_scratch
7835         (sub_shiftsi): Likewise.
7836         (sub_shiftsi_compare0
7837         (sub_shiftsi_compare0_scratch
7838         (and_scc): Likewise.
7839         (cond_move): Likewise.
7840         (if_plus_move): Likewise.
7841         (if_move_plus): Likewise.
7842         (if_move_not): Likewise.
7843         (if_not_move): Likewise.
7844         (if_shift_move): Likewise.
7845         (if_move_shift): Likewise.
7846         (if_shift_shift): Likewise.
7847         (if_not_arith): Likewise.
7848         (if_arith_not): Likewise.
7849         (cond_move_not): Likewise.
7850         (thumb1_ashlsi3): Set type attribute.
7851         (thumb1_ashrsi3): Likewise.
7852         (thumb1_lshrsi3): Likewise.
7853         (thumb1_rotrsi3): Likewise.
7854         (shiftsi3_compare0_scratch): Likewise.
7855         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
7856         (neon_mov<mode>): Likewise.
7857         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
7858         attribute changes.
7859         (thumb2_movsi_insn): Likewise.
7860         (thumb2_cmpsi_neg_shiftsi): Likewise.
7861         (thumb2_extendqisi_v6): Likewise.
7862         (thumb2_zero_extendhisi2_v6): Likewise.
7863         (thumb2_zero_extendqisi2_v6): Likewise.
7864         (thumb2_shiftsi3_short): Likewise.
7865         (thumb2_addsi3_compare0_scratch): Likewise.
7866         (orsi_not_shiftsi_si): Likewise.
7867         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
7868         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
7869         changes.
7870         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
7871         (1020alu_shift_op): Likewise.
7872         (1020alu_shift_reg_op): Likewise.
7873         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
7874         (alu_shift_op): Likewise.
7875         (alu_shift_reg_op): Likewise.
7876         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
7877         (11_alu_shift_op): Likewise.
7878         (11_alu_shift_reg_op): Likewise.
7879         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
7880         (9_alu_shift_reg_op): Likewise.
7881         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7882         changes.
7883         (cortex_a15_alu_shift): Likewise.
7884         (cortex_a15_alu_shift_reg): Likewise.
7885         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
7886         changes.
7887         (cortex_a5_alu_shift): Likewise.
7888         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
7889         changes.
7890         (cortex_a53_alu_shift): Likewise.
7891         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
7892         changes.
7893         (cortex_a7_alu_reg): Likewise.
7894         (cortex_a7_alu_shift): Likewise.
7895         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
7896         changes.
7897         (cortex_a8_alu_shift): Likewise.
7898         (cortex_a8_alu_shift_reg): Likewise.
7899         (cortex_a8_mov): Likewise.
7900         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
7901         (cortex_a9_dp_shift): Likewise.
7902         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
7903         changes.
7904         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
7905         changes.
7906         (cortex_r4_mov): Likewise.
7907         (cortex_r4_alu_shift): Likewise.
7908         (cortex_r4_alu_shift_reg): Likewise.
7909         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
7910         (526_alu_shift_op): Likewise.
7911         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
7912         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
7913         (626te_alu_shift_op): Likewise.
7914         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
7915         (726te_alu_op): Likewise.
7916         (726te_alu_shift_op): Likewise.
7917         (726te_alu_shift_reg_op): Likewise.
7918         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
7919         (mp626_alu_shift_op): Likewise.
7920         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
7921         (pj4_alu_e1_conds): Likewise.
7922         (pj4_alu): Likewise.
7923         (pj4_alu_conds): Likewise.
7924         (pj4_shift): Likewise.
7925         (pj4_shift_conds): Likewise.
7926         (pj4_alu_shift): Likewise.
7927         (pj4_alu_shift_conds): Likewise.
7928         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
7929         changes.
7930         (cortexa7_older_only): Likewise.
7931         (cortexa7_younger): Likewise.
7933 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
7935         * ipa-pure-const.c (generate_summary): Rename to...
7936         (pure_const_generate_summary): ... this.
7938 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
7940         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
7942 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
7944         PR target/57909
7945         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
7946         usage in HI mode.
7948 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7950         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
7951         enabled without -march=zEC12.
7952         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
7953         flags to be set.
7955 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
7957         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
7958         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
7959         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
7960         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
7961         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
7962         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
7963         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
7964         ISA_HAS_FP4.
7965         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
7966         and ISA_HAS_NMADD3_NMSUB3.
7967         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
7968         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
7969         (nmsub4<mode>, nmsub3<mode>): Likewise.
7970         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
7972 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
7974         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
7975         TARGET_MIPS5400 checking.
7977 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
7978             Peter Bergner  <bergner@vnet.ibm.com>
7980         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
7981         registers in the comment.
7982         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
7983         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
7984         rather than FIRST_PSEUDO_REGISTERS.
7986 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
7988         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
7989         enable extra ISA flags with TARGET_HTM.
7991 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
7993         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
7994         Fix comment typos.
7996 2013-07-15  Cong Hou  <congh@google.com>
7998         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
7999         in compare function for sorting.
8001 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
8003         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
8004         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
8005         * config/rs6000/rs6000.opt: Add -mhtm option.
8006         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
8007         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
8008         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
8009         __HTM__ if the HTM instructions are available.
8010         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
8011         htm_spr_reg_operand): New define_predicates.
8012         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
8013         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
8014         Include htm.md.
8015         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
8016         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
8017         HTM builtin functions.
8018         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
8019         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
8020         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
8021         (rs6000_builtin_mask_calculate): Likewise.
8022         (rs6000_option_override_internal): Likewise.
8023         (bdesc_htm): Add new HTM builtin support.
8024         (htm_spr_num): New function.
8025         (htm_spr_regno): Likewise.
8026         (rs6000_htm_spr_icode): Likewise.
8027         (htm_expand_builtin): Likewise.
8028         (htm_init_builtins): Likewise.
8029         (rs6000_expand_builtin): Add support for HTM builtin functions.
8030         (rs6000_init_builtins): Likewise.
8031         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
8032         option.
8033         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
8034         (TARGET_HTM, MASK_HTM): Define macros.
8035         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
8036         (FIXED_REGISTERS): Likewise.
8037         (CALL_USED_REGISTERS): Likewise.
8038         (CALL_REALLY_USED_REGISTERS): Likewise.
8039         (REG_ALLOC_ORDER): Likewise.
8040         (enum reg_class): Likewise.
8041         (REG_CLASS_NAMES): Likewise.
8042         (REG_CLASS_CONTENTS): Likewise.
8043         (REGISTER_NAMES): Likewise.
8044         (ADDITIONAL_REGISTER_NAMES): Likewise.
8045         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
8046         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
8047         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
8048         * config/rs6000/htm.md: New file.
8049         * config/rs6000/htmintrin.h: New file.
8050         * config/rs6000/htmxlintrin.h: New file.
8052 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8054         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
8055         Define SYMBOL_TINY_GOT, update comment.
8056         * config/aarch64/aarch64.c
8057         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
8058         (aarch64_expand_mov_immediate): Likewise.
8059         (aarch64_print_operand): Likewise.
8060         (aarch64_classify_symbol): Likewise.
8061         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
8062         (ldr_got_tiny): Define.
8064 2013-07-13  Tobias Grosser  <tobias@grosser.es>
8066         PR tree-optimization/54094
8067         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
8068           scheduling dimension for the parallelism check from the polyhedral
8069           information in the AST.
8070         * graphite-dependences.c (carries_deps): Do not assume the schedule is
8071           in 2D + 1 form.
8073 2013-07-13  Jason Merrill  <jason@redhat.com>
8075         * print-tree.c (debug_vec_tree): Use debug_raw.
8076         (debug_raw (vec<tree, va_gc> &)): New.
8077         (debug_raw (vec<tree, va_gc> *)): New.
8078         * tree.h: Declare them.
8080 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
8082         * ifcvt.c (ifcvt_after_combine): New static variable.
8083         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
8084         for size.
8085         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
8086         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
8087         rest_of_handle_if_after_reload): Pass new argument for if_convert.
8089 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
8091         * config/mips/mips.c (mips_expand_call): Remove empty statement.
8093 2013-07-12  Michael Matz  <matz@suse.de>
8095         PR middle-end/55771
8096         * convert.c (convert_to_real): Reject non-float inner types.
8098 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
8100         * config/aarch64/aarch64-protos.h
8101         (aarch64_simd_immediate_valid_for_move): Remove.
8102         * config/aarch64/aarch64.c (simd_immediate_info): New member.
8103         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
8104         cases.
8105         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
8107 2013-07-11  Steve Ellcey  <sellcey@mips.com>
8109         * config/mips/mips.c (mips_conditional_register_usage): Do not
8110         use t[0-7] registers in MIPS16 mode when optimizing for size.
8112 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
8114         * config/i386/i386.c (dispatch_function_versions): Fix array
8115         indexing of function_version_info to match actual_versions.
8117 2013-07-11  Teresa Johnson  <tejohnson@google.com>
8119         * vec.h (struct va_gc): Move release out-of-line.
8120         (va_gc::release): Call ggc_free on released vec.
8122 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
8124         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
8125         Require GOT register as additional operand in UNSPEC.
8126         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
8127         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
8128         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
8129         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
8130         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
8131         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
8132         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
8134 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
8136         PR target/57631
8137         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
8138         name seen by assembler/linker rather if available.
8140 2013-07-11  Andreas Schwab  <schwab@suse.de>
8142         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
8144 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
8146         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
8148 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
8150         * doc/tm.texi.in: Move hook documentation to ....
8151         * target.def: ... here.
8153         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
8154         text on @hook line.
8155         * doc/tm.texi: Regenerate.
8157 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
8159         PR c++/57869
8160         * doc/invoke.texi: Document -Wconditionally-supported.
8162 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
8164         PR target/57844
8165         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
8166         of my_fp.
8168 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
8170         PR target/57506
8171         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
8172         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
8173         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
8174         Remove duplicate devices.
8175         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
8176         * config/avr/t-multilib: Regenerate.
8177         * config/avr/avr-tables.opt: Regenerate.
8178         * doc/avr-mmcu.texi: Regenerate.
8180 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
8182         PR target/56987
8183         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
8185 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
8187         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
8188         the cost of MULT when optimizing for size.
8190 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
8192         * config/cr16/cr16-protos.h: Don't include target.h.
8194 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
8196         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
8197         adjust register size for TDmode and TFmode for VSX registers.
8199 2013-07-08  Kai Tietz  <ktietz@redhat.com>
8201         PR target/56892
8202         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
8203         hook_bool_const_tree_true.
8205 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8207         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
8208         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
8209         * config/s390/s390.md: Define FPR*_REGNUM constants.
8210         Fix FPR2_REGNUM constant (18 -> 17).
8211         ("*trunc<BFP:mode><DFP_ALL:mode>2")
8212         ("*trunc<DFP_ALL:mode><BFP:mode>2")
8213         ("trunc<BFP:mode><DFP_ALL:mode>2")
8214         ("trunc<DFP_ALL:mode><BFP:mode>2")
8215         ("*extend<BFP:mode><DFP_ALL:mode>2")
8216         ("*extend<DFP_ALL:mode><BFP:mode>2")
8217         ("extend<BFP:mode><DFP_ALL:mode>2")
8218         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
8219         FPR4_REGNUM.
8221 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
8223         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
8225 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8227         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
8228         and cfun_fpr_bit_p to cfun_fpr_save_p.
8229         (s390_frame_area, s390_register_info, s390_frame_info)
8230         (s390_emit_prologue, s390_emit_epilogue)
8231         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
8232         register numbers.
8233         * config/s390/s390.h: Define *_REGNUM macros for floating point
8234         register numbers.
8236 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
8238         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
8240 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
8242         PR rtl-optimization/57786
8243         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
8244         and break out of the loop when it is set to false.
8246 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
8248         PR target/57819
8249         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
8250         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
8251         (const_int 63)) 0)).
8252         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
8253         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
8254         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
8256         PR rtl-optimization/57829
8257         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
8258         mask bits outside of mode are just sign-extension from mode to HWI.
8260 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
8262         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
8263         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
8264         adjust_address instead of change_address to keep info about alignment.
8265         (emit_strmov): Remove.
8266         (emit_memmov): New function.
8267         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
8268         (expand_movmem_epilogue): Likewise and return updated rtx for
8269         destination.
8270         (expand_constant_movmem_prologue): Likewise and return updated rtx for
8271         destination and source.
8272         (decide_alignment): Refactor, handle vector_loop.
8273         (ix86_expand_movmem): Likewise.
8274         (ix86_expand_setmem): Likewise.
8275         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
8277 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
8279         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
8280         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
8282 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
8284         * config/i386/sse.md (sse_movlhps): Change alternative 3
8285         of operand 2 to "m".
8287 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
8289         PR target/57807
8290         * config/i386/sse.md (iptr): New mode attribute.
8291         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
8292         (<sse>_vm<plusminus_insn><mode>3): Ditto.
8293         (<sse>_vmmul<mode>3): Ditto.
8294         (<sse>_vmdiv<mode>3): Ditto.
8295         (sse_vmrcpv4sf2): Ditto.
8296         (<sse>_vmsqrt<mode>2): Ditto.
8297         (sse_vmrsqrtv4sf2): Ditto.
8298         (<sse>_vm<code><mode>3): Ditto.
8299         (avx_vmcmp<mode>3): Ditto.
8300         (<sse>_vmmaskcmp<mode>3): Ditto.
8301         (<sse>_comi): Ditto.
8302         (<sse>_ucomi): Ditto.
8303         (*xop_vmfrcz_<mode>): Ditto.
8304         (*fmai_fmadd_<mode>): Ditto.
8305         (*fmai_fmsub_<mode>): Ditto.
8306         (*fmai_fnmadd_<mode>): Ditto.
8307         (*fmai_fnmsub_<mode>): Ditto.
8308         (*fma4i_vmfmadd_<mode>): Ditto.
8309         (*fma4i_vmfmsub_<mode>): Ditto.
8310         (*fma4i_vmfnmadd_<mode>): Ditto.
8311         (*fma4i_vmfnmsub_<mode>): Ditto.
8312         (*xop_vmfrcz_<mode>): Ditto.
8313         (sse_cvtps2pi): Ditto.
8314         (sse_cvttps2pi): Ditto.
8315         (sse_cvtss2si): Ditto.
8316         (sse_cvtss2si_2): Ditto.
8317         (sse_cvtss2siq_2): Ditto.
8318         (sse_cvttss2si): Ditto.
8319         (sse_cvttss2siq): Ditto.
8320         (sse_cvtsd2si): Ditto.
8321         (sse_cvtsd2si_2): Ditto.
8322         (sse_cvtsd2siq_2): Ditto.
8323         (sse_cvttsd2si): Ditto.
8324         (sse_cvttsd2siq): Ditto.
8325         (sse_cvtsd2ss): Ditto.
8326         (sse_cvtss2sd): Ditto.
8327         (avx2_pbroadcast<mode>): Ditto.
8328         (avx2_pbroadcast<mode>_1): Ditto.
8329         (*avx_vperm_broadcast_v4sf): Ditto.
8331         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
8332         (sse_movlhps): Ditto.
8333         (sse_storehps): Ditto.
8334         (sse_loadhps): Ditto.
8335         (sse_storelps): Ditto.
8336         (sse_loadlps): Ditto.
8337         (*vec_concatv4sf): Ditto.
8338         (*vec_interleave_highv2df): Ditto.
8339         (*vec_interleave_lowv2df): Ditto.
8340         (*vec_extractv2df_1_sse): Ditto.
8341         (*vec_extractv2df_0_sse): Ditto.
8342         (sse2_storelpd): Ditto.
8343         (sse2_loadlpd): Ditto.
8344         (sse2_movsd): Ditto.
8345         (*vec_concatv4si): Ditto.
8346         (vec_concatv2di): Ditto.
8348         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
8349         for Intel asm dialect.
8350         (mmx_punpcklwd): Ditto.
8351         (mmx_punpckldq): Ditto.
8353         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
8354         for intel assembler dialect.
8356 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
8358         PR target/29776
8359         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
8360         for BUILT_IN_C{LZ,LRSB}*.
8361         * tree.h (CASE_INT_FN): Add FN##IMAX case.
8362         * tree-vrp.c (extract_range_basic): Handle
8363         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
8364         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
8365         fall thru to code calling set_value*.
8366         * builtins.c (expand_builtin): Remove *IMAX cases.
8367         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
8368         if width is bigger than 2*HWI.
8370 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
8372         PR rtl-optimization/55342
8373         * lra-int.h (lra_subreg_reload_pseudos): New.
8374         * lra.c: Add undoing optional reloads to the block diagram.
8375         (lra_subreg_reload_pseudos): New.
8376         (lra_optional_reload_pseudos): Change comments.
8377         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
8378         lra_optional_reload_pseudos after undo transformations.
8379         * lra-assigns.c (pseudo_prefix_title): New.
8380         (lra_setup_reg_renumber): Use it.
8381         (spill_for): Ditto.  Check subreg reload pseudos too.
8382         (assign_by_spills): Consider subreg reload pseudos too.
8383         * lra-constraints.c (simplify_operand_subreg): Use
8384         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
8385         (curr_insn_transform): Recognize and do optional reloads.
8386         (undo_optional_reloads): New.
8387         (lra_undo_inheritance): Call undo_optional_reloads.
8389 2013-07-05  Thomas Quinot  <quinot@adacore.com>
8391         * tree-complex.c (expand_complex_operations_1): Fix typo.
8393 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
8395         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
8396         (tune_params): New member 'const vec_costs'.
8397         * config/aarch64/aarch64.c (generic_vector_cost): New.
8398         (generic_tunings): New member 'generic_vector_cost'.
8399         (aarch64_builtin_vectorization_cost): New.
8400         (aarch64_add_stmt_cost): New.
8401         (TARGET_VECTORIZE_ADD_STMT_COST): New.
8402         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
8404 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
8406         PR target/57777
8407         * config/i386/predicates.md (vsib_address_operand): Disallow
8408         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
8410 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
8412         PR middle-end/55030
8413         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
8414         expand_builtin_setjmp_receiver.
8415         (expand_label): Adjust, call expand_builtin_setjmp_receiver
8416         with NULL for the label parameter.
8417         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
8418         the frame-pointer.  Adjust comments.
8419         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
8420         only if LABEL is non-NULL.
8422 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
8424         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
8425         (ARM_ABI_AAPCS64): Ditto.
8426         (arm_abi): Ditto.
8427         (ARM_DEFAULT_ABI): Ditto.
8429 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
8431         * config/aarch64/aarch64-builtins.c
8432         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
8433         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
8434         (st1): Likewise.
8435         * config/aarch64/aarch64-simd.md
8436         (aarch64_ld1<VALL:mode>): New.
8437         (aarch64_st1<VALL:mode>): Likewise.
8438         * config/aarch64/arm_neon.h
8439         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
8441 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
8443         * config/i386/i386.c (gate_insert_vzeroupper): Check if
8444         target ISA is AVX.
8445         (ix86_option_override_internal):Turn on all -mavx target flags by
8446         default as they are dependent on AVX anyway.
8448 2013-07-02  Cary Coutant  <ccoutant@google.com>
8450         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
8451         deterministic hash.
8452         (loc_checksum_ordered): Likewise.
8453         (hash_loc_operands): Remove inline keyword.
8455 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
8457         PR tree-optimization/57741
8458         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
8459         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
8460         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
8461         Allow REAL_CST step_exprs if flag_associative_math.
8462         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
8464 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
8466         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
8468 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
8470         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
8472 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8474         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
8475         encoding.
8476         (iorsi3_insn): Likewise.
8477         (arm_xorsi3): Likewise.
8479 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
8481         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
8482         to ...
8483         (attribute "type"): ... here, and prefix with "wmmx_".
8484         (attribute "core_cycles"): Update for attribute changes.
8485         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
8486         (tbcstv4hi): Likewise.
8487         (tbcstv2si): Likewise.
8488         (iwmmxt_iordi3): Likewise.
8489         (iwmmxt_xordi3): Likewise.
8490         (iwmmxt_anddi3): Likewise.
8491         (iwmmxt_nanddi3): Likewise.
8492         (iwmmxt_arm_movdi): Likewise.
8493         (iwmmxt_movsi_insn): Likewise.
8494         (mov<mode>_internal): Likewise.
8495         (and<mode>3_iwmmxt): Likewise.
8496         (ior<mode>3_iwmmxt): Likewise.
8497         (xor<mode>3_iwmmxt): Likewise.
8498         (add<mode>3_iwmmxt): Likewise.
8499         (ssaddv8qi3): Likewise.
8500         (ssaddv4hi3): Likewise.
8501         (ssaddv2si3): Likewise.
8502         (usaddv8qi3): Likewise.
8503         (usaddv4hi3): Likewise.
8504         (usaddv2si3): Likewise.
8505         (sub<mode>3_iwmmxt): Likewise.
8506         (sssubv8qi3): Likewise.
8507         (sssubv4hi3): Likewise.
8508         (sssubv2si3): Likewise.
8509         (ussubv8qi3): Likewise.
8510         (ussubv4hi3): Likewise.
8511         (ussubv2si3): Likewise.
8512         (mulv4hi3_iwmmxt): Likewise.
8513         (smulv4hi3_highpart): Likewise.
8514         (umulv4hi3_highpart): Likewise.
8515         (iwmmxt_wmacs): Likewise.
8516         (iwmmxt_wmacsz): Likewise.
8517         (iwmmxt_wmacu): Likewise.
8518         (iwmmxt_wmacuz): Likewise.
8519         (iwmmxt_clrdi): Likewise.
8520         (iwmmxt_clrv8qi): Likewise.
8521         (iwmmxt_clr4hi): Likewise.
8522         (iwmmxt_clr2si): Likewise.
8523         (iwmmxt_uavgrndv8qi3): Likewise.
8524         (iwmmxt_uavgrndv4hi3): Likewise.
8525         (iwmmxt_uavgv8qi3): Likewise.
8526         (iwmmxt_uavgv4hi3): Likewise.
8527         (iwmmxt_tinsrb): Likewise.
8528         (iwmmxt_tinsrh): Likewise.
8529         (iwmmxt_tinsrw): Likewise.
8530         (iwmmxt_textrmub): Likewise.
8531         (iwmmxt_textrmsb): Likewise.
8532         (iwmmxt_textrmuh): Likewise.
8533         (iwmmxt_textrmsh): Likewise.
8534         (iwmmxt_textrmw): Likewise.
8535         (iwmxxt_wshufh): Likewise.
8536         (eqv8qi3): Likewise.
8537         (eqv4hi3): Likewise.
8538         (eqv2si3): Likewise.
8539         (gtuv8qi3): Likewise.
8540         (gtuv4hi3): Likewise.
8541         (gtuv2si3): Likewise.
8542         (gtv8qi3): Likewise.
8543         (gtv4hi3): Likewise.
8544         (gtv2si3): Likewise.
8545         (smax<mode>3_iwmmxt): Likewise.
8546         (umax<mode>3_iwmmxt): Likewise.
8547         (smin<mode>3_iwmmxt): Likewise.
8548         (umin<mode>3_iwmmxt): Likewise.
8549         (iwmmxt_wpackhss): Likewise.
8550         (iwmmxt_wpackwss): Likewise.
8551         (iwmmxt_wpackdss): Likewise.
8552         (iwmmxt_wpackhus): Likewise.
8553         (iwmmxt_wpackwus): Likewise.
8554         (iwmmxt_wpackdus): Likewise.
8555         (iwmmxt_wunpckihb): Likewise.
8556         (iwmmxt_wunpckihh): Likewise.
8557         (iwmmxt_wunpckihw): Likewise.
8558         (iwmmxt_wunpckilb): Likewise.
8559         (iwmmxt_wunpckilh): Likewise.
8560         (iwmmxt_wunpckilw): Likewise.
8561         (iwmmxt_wunpckehub): Likewise.
8562         (iwmmxt_wunpckehuh): Likewise.
8563         (iwmmxt_wunpckehuw): Likewise.
8564         (iwmmxt_wunpckehsb): Likewise.
8565         (iwmmxt_wunpckehsh): Likewise.
8566         (iwmmxt_wunpckehsw): Likewise.
8567         (iwmmxt_wunpckelub): Likewise.
8568         (iwmmxt_wunpckeluh): Likewise.
8569         (iwmmxt_wunpckeluw): Likewise.
8570         (iwmmxt_wunpckelsb): Likewise.
8571         (iwmmxt_wunpckelsh): Likewise.
8572         (iwmmxt_wunpckelsw): Likewise.
8573         (ror<mode>3): Likewise.
8574         (ashr<mode>3_iwmmxt): Likewise.
8575         (lshr<mode>3_iwmmxt): Likewise.
8576         (ashl<mode>3_iwmmxt): Likewise.
8577         (ror<mode>3_di): Likewise.
8578         (ashr<mode>3_di): Likewise.
8579         (lshr<mode>3_di): Likewise.
8580         (ashl<mode>3_di): Likewise.
8581         (iwmmxt_wmadds): Likewise.
8582         (iwmmxt_wmaddu): Likewise.
8583         (iwmmxt_tmia): Likewise.
8584         (iwmmxt_tmiaph): Likewise.
8585         (iwmmxt_tmiabb): Likewise.
8586         (iwmmxt_tmiatb): Likewise.
8587         (iwmmxt_tmiabt): Likewise.
8588         (iwmmxt_tmiatt): Likewise.
8589         (iwmmxt_tmovmskb): Likewise.
8590         (iwmmxt_tmovmskh): Likewise.
8591         (iwmmxt_tmovmskw): Likewise.
8592         (iwmmxt_waccb): Likewise.
8593         (iwmmxt_wacch): Likewise.
8594         (iwmmxt_waccw): Likewise.
8595         (iwmmxt_waligni): Likewise.
8596         (iwmmxt_walignr): Likewise.
8597         (iwmmxt_walignr0): Likewise.
8598         (iwmmxt_walignr1): Likewise.
8599         (iwmmxt_walignr2): Likewise.
8600         (iwmmxt_walignr3): Likewise.
8601         (iwmmxt_wsadb): Likewise.
8602         (iwmmxt_wsadh): Likewise.
8603         (iwmmxt_wsadbz): Likewise.
8604         (iwmmxt_wsadhz): Likewise.
8605         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
8606         (iwmmxt_wabsdiffb): Likewise.
8607         (iwmmxt_wabsdiffh): Likewise.
8608         (iwmmxt_wabsdiffw): Likewise.
8609         (iwmmxt_waddsubhx): Likewise
8610         (iwmmxt_wsubaddhx): Likewise.
8611         (addc<mode>3): Likewise.
8612         (iwmmxt_avg4): Likewise.
8613         (iwmmxt_avg4r): Likewise.
8614         (iwmmxt_wmaddsx): Likewise.
8615         (iwmmxt_wmaddux): Likewise.
8616         (iwmmxt_wmaddsn): Likewise.
8617         (iwmmxt_wmaddun): Likewise.
8618         (iwmmxt_wmulwsm): Likewise.
8619         (iwmmxt_wmulwum): Likewise.
8620         (iwmmxt_wmulsmr): Likewise.
8621         (iwmmxt_wmulumr): Likewise.
8622         (iwmmxt_wmulwsmr): Likewise.
8623         (iwmmxt_wmulwumr): Likewise.
8624         (iwmmxt_wmulwl): Likewise.
8625         (iwmmxt_wqmulm): Likewise.
8626         (iwmmxt_wqmulwm): Likewise.
8627         (iwmmxt_wqmulmr): Likewise.
8628         (iwmmxt_wqmulwmr): Likewise.
8629         (iwmmxt_waddbhusm): Likewise.
8630         (iwmmxt_waddbhusl): Likewise.
8631         (iwmmxt_wqmiabb): Likewise.
8632         (iwmmxt_wqmiabt): Likewise.
8633         (iwmmxt_wqmiatb): Likewise.
8634         (iwmmxt_wqmiatt): Likewise.
8635         (iwmmxt_wqmiabbn): Likewise.
8636         (iwmmxt_wqmiabtn): Likewise.
8637         (iwmmxt_wqmiatbn): Likewise.
8638         (iwmmxt_wqmiattn): Likewise.
8639         (iwmmxt_wmiabb): Likewise.
8640         (iwmmxt_wmiabt): Likewise.
8641         (iwmmxt_wmiatb): Likewise.
8642         (iwmmxt_wmiatt): Likewise.
8643         (iwmmxt_wmiabbn): Likewise.
8644         (iwmmxt_wmiabtn): Likewise.
8645         (iwmmxt_wmiatbn): Likewise.
8646         (iwmmxt_wmiattn): Likewise.
8647         (iwmmxt_wmiawbb): Likewise.
8648         (iwmmxt_wmiawbt): Likewise.
8649         (iwmmxt_wmiawtb): Likewise.
8650         (iwmmxt_wmiawtt): Likewise.
8651         (iwmmxt_wmiawbbn): Likewise.
8652         (iwmmxt_wmiawbtn): Likewise.
8653         (iwmmxt_wmiawtbn): Likewise.
8654         (iwmmxt_wmiawttn): Likewise.
8655         (iwmmxt_wmerge): Likewise.
8656         (iwmmxt_tandc<mode>3): Likewise.
8657         (iwmmxt_torc<mode>3): Likewise.
8658         (iwmmxt_torvsc<mode>3): Likewise.
8659         (iwmmxt_textrc<mode>3): Likewise.
8660         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
8661         (wmmxt_pack): Likewise.
8662         (wmmxt_mult_c1): Likewise.
8663         (wmmxt_mult_c2): Likewise.
8664         (wmmxt_alu_c1): Likewise.
8665         (wmmxt_alu_c2): Likewise.
8666         (wmmxt_alu_c3): Likewise.
8667         (wmmxt_transfer_c1): Likewise.
8668         (wmmxt_transfer_c2): Likewise.
8669         (wmmxt_transfer_c3): Likewise.
8670         (marvell_f_iwmmxt_wstr): Likewise.
8671         (marvell_f_iwmmxt_wldr): Likewise.
8673 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
8675         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
8677 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
8679         Revert:
8680         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
8681         * lra-constraints.c (need_for_split_p): Check call used hard regs
8682         living through calls.
8684         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
8685         call used regs for call insn.
8687 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
8689         PR target/57736
8690         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
8691         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
8693 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
8695         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
8697 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
8699         * lra-constraints.c (need_for_split_p): Check call used hard regs
8700         living through calls.
8702 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
8704         PR target/57744
8705         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
8706         to tie with any other modes.  Eliminate Altivec vector mode tests,
8707         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
8708         code, to return 0 if testing MODE2 for a condition, if we've
8709         already tested MODE1 for the same condition.
8711 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8713         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
8714         layout.
8716 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8718         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
8719         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
8721 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8723         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
8724         Define.
8725         (aarch64_symbolic_constant_p): Remove.
8726         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
8727         static.  Fix line length and white space.
8728         (aarch64_symbolic_constant_p): Remove.
8729         * config/aarch64/predicates.md (aarch64_valid_symref):
8730         Use aarch64_classify_symbol_expression.
8732 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8734         * config/arm/constraints.md (Ts): New constraint.
8735         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
8736         16-bit encodings.
8737         (compare_scc): Use "Ts" constraint for operand 0.
8738         (ior_scc_scc): Likewise.
8739         (and_scc_scc): Likewise.
8740         (and_scc_scc_nodom): Likewise.
8741         (ior_scc_scc_cmp): Likewise for operand 7.
8742         (and_scc_scc_cmp): Likewise.
8743         * config/arm/thumb2.md (thumb2_movsi_insn):
8744         Add alternatives for 16-bit encodings.
8745         (thumb2_movhi_insn): Likewise.
8746         (thumb2_movsicc_insn): Likewise.
8747         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
8748         (thumb2_negscc): Use "Ts" constraint.
8749         Move mvn instruction outside cond_exec block.
8750         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
8751         for 16-bit encodings.
8753 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8755         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
8756         encoding.
8757         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
8758         (mulsi3subsi): Likewise.
8759         (mulsidi3adddi): Likewise.
8760         (mulsidi3_v6): Likewise.
8761         (umulsidi3_v6): Likewise.
8762         (umulsidi3adddi_v6): Likewise.
8763         (smulsi3_highpart_v6): Likewise.
8764         (umulsi3_highpart_v6): Likewise.
8765         (mulhisi3tb): Likewise.
8766         (mulhisi3bt): Likewise.
8767         (mulhisi3tt): Likewise.
8768         (maddhisi4): Likewise.
8769         (maddhisi4tb): Likewise.
8770         (maddhisi4tt): Likewise.
8771         (maddhidi4): Likewise.
8772         (maddhidi4tb): Likewise.
8773         (maddhidi4tt): Likewise.
8774         (zeroextractsi_compare0_scratch): Likewise.
8775         (insv_zero): Likewise.
8776         (insv_t2): Likewise.
8777         (anddi_notzesidi_di): Likewise.
8778         (anddi_notsesidi_di): Likewise.
8779         (andsi_notsi_si): Likewise.
8780         (iordi_zesidi_di): Likewise.
8781         (xordi_zesidi_di): Likewise.
8782         (andsi_iorsi3_notsi): Likewise.
8783         (smax_0): Likewise.
8784         (smax_m1): Likewise.
8785         (smin_0): Likewise.
8786         (not_shiftsi): Likewise.
8787         (unaligned_loadsi): Likewise.
8788         (unaligned_loadhis): Likewise.
8789         (unaligned_loadhiu): Likewise.
8790         (unaligned_storesi): Likewise.
8791         (unaligned_storehi): Likewise.
8792         (extv_reg): Likewise.
8793         (extzv_t2): Likewise.
8794         (divsi3): Likewise.
8795         (udivsi3): Likewise.
8796         (arm_zero_extendhisi2addsi): Likewise.
8797         (arm_zero_extendqisi2addsi): Likewise.
8798         (compareqi_eq0): Likewise.
8799         (arm_extendhisi2_v6): Likewise.
8800         (arm_extendqisi2addsi): Likewise.
8801         (arm_movt): Likewise.
8802         (thumb2_ldrd): Likewise.
8803         (thumb2_ldrd_base): Likewise.
8804         (thumb2_ldrd_base_neg): Likewise.
8805         (thumb2_strd): Likewise.
8806         (thumb2_strd_base): Likewise.
8807         (thumb2_strd_base_neg): Likewise.
8808         (arm_negsi2): Add alternative for 16-bit encoding.
8809         (arm_one_cmplsi2): Likewise.
8811 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8813         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
8814         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
8815         (movdfcc): Likewise.
8816         * config/arm/vfp.md (*thumb2_movsf_vfp):
8817         Disable predication for arm_restrict_it.
8818         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
8819         (*thumb2_movdfcc_vfp): Likewise.
8820         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
8821         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
8822         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
8823         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
8824         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
8825         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
8826         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
8827         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
8828         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
8829         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
8830         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
8831         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
8832         Disable predication for arm_restrict_it.
8834 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
8836         * config/i386/bmiintrin.h (_bextr_u32): New.
8837         (_bextr_u64): Ditto.
8839 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
8841         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
8842         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
8843         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
8844         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
8845         * config/mips/n32-elf.h: ...this new file.
8847 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
8849         PR target/57224
8850         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
8851         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
8853 2013-06-27  Catherine Moore  <clm@codesourcery.com>
8855         * config/mips/mips-tables.opt: Regenerate.
8856         * config/mips/mips-cpus.def: Add m14ke and m14kec.
8857         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
8858         * doc/invoke.texi: Add -m14kc.
8860 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
8862         PR target/57623
8863         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
8864         constraints of operand 1 and 2.
8866         PR target/57623
8867         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
8868         to match RTL canonicalization.  Swap predicates and
8869         constraints of operand 1 and 2.
8871 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
8873         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
8874         Process OP_INOUT regs for splitting too.
8876 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
8878         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
8879         decl before the loop, initialize to NULL.
8880         (vectorizable_load): Initialize ptr_incr to NULL.
8882 2013-06-27  Martin Jambor  <mjambor@suse.cz>
8884         PR lto/57208
8885         * ipa-ref.h (ipa_maybe_record_reference): Declare.
8886         * ipa-ref.c (ipa_maybe_record_reference): New function.
8887         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
8888         * ipa-cp.c (create_specialized_node): Record potential references from
8889         aggvals.
8890         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
8892 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8894         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
8895         parameter 'mode' of type 'enum machine_mode mode'; change to pass
8896         'mode' to force_reg.
8897         (aarch64_add_offset): Update calls to aarch64_force_temporary.
8898         (aarch64_expand_mov_immediate): Likewise.
8900 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8902         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
8903         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
8905 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8907         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
8908         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
8909         (struct machine_function): Add tbegin_p.
8910         (s390_canonicalize_comparison): Fold CC mode compares to
8911         conditional jump if possible.
8912         (s390_emit_jump): Return the emitted jump.
8913         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
8914         Handle CCRAWmode compares.
8915         (s390_option_override): Default to -mhtm if available.
8916         (s390_reg_clobbered_rtx): Handle floating point regs as well.
8917         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
8918         FPRs instead of df_regs_ever_live_p.
8919         (s390_optimize_nonescaping_tx): New function.
8920         (s390_init_frame_layout): Extend clobbered_regs array to cover
8921         FPRs as well.
8922         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
8923         (s390_expand_tbegin): New function.
8924         (enum s390_builtin): New enum definition.
8925         (code_for_builtin): New array definition.
8926         (s390_init_builtins): New function.
8927         (s390_expand_builtin): New function.
8928         (TARGET_INIT_BUILTINS): Define.
8929         (TARGET_EXPAND_BUILTIN): Define.
8930         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
8931         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
8932         (s390_alc_comparison): Likewise.
8933         * config/s390/s390-modes.def: Add CCRAWmode.
8934         * config/s390/s390.h (processor_flags): Add PF_TX.
8935         (TARGET_CPU_HTM): Define macro.
8936         (TARGET_HTM): Define macro.
8937         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
8938         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
8939         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
8940         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
8941         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
8942         (TBEGIN_MASK, TBEGINC_MASK): New constants.
8943         ("*cc_to_int"): Move up.
8944         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
8945         constants other than 0.
8946         ("*ccraw_to_int"): New insn and splitter definition.
8947         ("tbegin", "tbegin_nofloat", "tbegin_retry")
8948         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
8949         ("tx_assist"): New expander.
8950         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
8951         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
8952         * config/s390/s390.opt: Add -mhtm option.
8953         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
8954         * config/s390/htmxlintrin.h: New file.
8955         * config/s390/htmintrin.h: New file.
8956         * config/s390/s390intrin.h: New file.
8957         * doc/extend.texi: Document htm builtins.
8958         * config.gcc: Add the new header files to extra_headers.
8960 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
8962         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
8963         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
8965 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
8966             Pat Haugen  <pthaugen@us.ibm.com>
8967             Peter Bergner  <bergner@vnet.ibm.com>
8969         * config/rs6000/power8.md: New.
8970         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
8971         setting for power8 entry.
8972         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
8973         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
8974         test for Power4/Power5 only.
8975         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
8976         support.
8977         (force_new_group): Adjust comment.
8978         * config/rs6000/rs6000.md: Include power8.md.
8980 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
8982         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
8983         * config/arm/arm-protos.h (arm_max_conditional_execute): New
8984         declaration.
8985         (tune_params): Update comment.
8986         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
8987         (arm_max_conditional_execute): New function.
8988         (thumb2_final_prescan_insn): Use max_insn_skipped and
8989         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
8991 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
8993         PR tree-optimization/57705
8994         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
8995         SSA_NAME step, provided that it is not defined inside the loop.
8996         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
8997         (get_initial_def_for_induction): Handle SSA_NAME IV step.
8999 2013-06-25  Martin Jambor  <mjambor@suse.cz>
9001         PR middle-end/57670
9002         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
9003         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
9004         calls in the dump.
9005         (ipa_note_param_call): Initialize member_ptr flag.
9006         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
9007         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
9008         (ipa_write_indirect_edge_info): Stream member_ptr flag.
9009         (ipa_read_indirect_edge_info): Likewise.
9011 2013-06-25  Richard Biener  <rguenther@suse.de>
9013         PR middle-end/56977
9014         * passes.c (init_optimization_passes): Move pass_fold_builtins
9015         and pass_dce earlier with -Og.
9017 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
9019         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
9020         <BIT_FIELD_REF>: Remove trailing TAB.
9021         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
9022         remove blank line.
9024 2013-06-24  Martin Jambor  <mjambor@suse.cz>
9026         PR tree-optimization/57358
9027         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
9028         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
9029         (ipa_analyze_params_uses): Generate pessimistic info when true.
9031 2013-06-24  Martin Jambor  <mjambor@suse.cz>
9033         PR tree-optimization/57539
9034         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
9035         global.inlined_to of the new node to it.  All callers changed.
9036         * ipa-inline-transform.c (clone_inlined_nodes): New variable
9037         inlining_into, pass it to cgraph_clone_node.
9038         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
9039         ipa_free_edge_args_substructures.
9040         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
9041         rdesc linked list.  Do not assert rdesc edges have inlined caller.
9042         Assert we have found an rdesc in the rdesc list.
9044 2013-06-24  Richard Biener  <rguenther@suse.de>
9046         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
9047         (pointer_set_lookup): Declare.
9048         (class pointer_map): New template class implementing a
9049         generic pointer to T map.
9050         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
9051         pointer_map<T>::contains, pointer_map<T>::insert,
9052         pointer_map<T>::traverse): New functions.
9053         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
9054         (pointer_set_lookup): New function.
9055         (pointer_set_contains): Use pointer_set_lookup.
9056         (pointer_set_insert): Likewise.
9057         (insert_aux): Remove.
9058         (struct pointer_map_t): Embed a pointer_set_t.
9059         (pointer_map_create): Adjust.
9060         (pointer_map_destroy): Likewise.
9061         (pointer_map_contains): Likewise.
9062         (pointer_map_insert): Likewise.
9063         (pointer_map_traverse): Likewise.
9064         * tree-streamer.h (struct streamer_tree_cache_d): Use a
9065         pointer_map<unsigned> instead of a pointer_map_t.
9066         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
9067         (streamer_tree_cache_lookup): Likewise.
9068         (streamer_tree_cache_create): Likewise.
9069         (streamer_tree_cache_delete): Likewise.
9070         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
9071         pointer_map<unsigned> instead of a pointer_map_t.
9072         (lto_init_tree_ref_encoder): Adjust.
9073         (lto_destroy_tree_ref_encoder): Likewise.
9074         * lto-section-out.c (lto_output_decl_index): Likewise.
9075         (lto_record_function_out_decl_state): Likewise.
9076         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
9078 2013-06-24  Richard Biener  <rguenther@suse.de>
9080         PR tree-optimization/57488
9081         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
9083 2013-06-24  Alan Modra  <amodra@gmail.com>
9085         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
9086         (gen_easy_altivec_constant): Likewise.
9087         * config/rs6000/predicates.md (easy_vector_constant_add_self,
9088         easy_vector_constant_msb): Likewise.
9090 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
9092         PR target/57688
9093         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
9094         add missing return true.
9096 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
9098         PR target/52483
9099         * config/sh/predicates.md (general_extend_operand): Invoke
9100         general_movsrc_operand for memory operands.
9101         (general_movsrc_operand): Allow reg+reg addressing, do not use
9102         general_operand for memory operands.
9104 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
9106         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
9107         when current target options does not apply.
9108         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
9109         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
9110         * config/i386/bmiintrin.h: Pass appropriate target
9111         attributes to header.
9112         * config/i386/mmintrin.h: Ditto.
9113         * config/i386/nmmintrin.h: Ditto.
9114         * config/i386/avx2intrin.h: Ditto.
9115         * config/i386/fxsrintrin.h: Ditto.
9116         * config/i386/tbmintrin.h: Ditto.
9117         * config/i386/xsaveintrin.h: Ditto.
9118         * config/i386/f16cintrin.h: Ditto.
9119         * config/i386/xtestintrin.h: Ditto.
9120         * config/i386/xsaveoptintrin.h: Ditto.
9121         * config/i386/bmi2intrin.h: Ditto.
9122         * config/i386/lzcntintrin.h: Ditto.
9123         * config/i386/smmintrin.h: Ditto.
9124         * config/i386/wmmintrin.h: Ditto.
9125         * config/i386/x86intrin.h: Remove all header include guards.
9126         * config/i386/prfchwintrin.h: Ditto.
9127         * config/i386/pmmintrin.h: Ditto.
9128         * config/i386/tmmintrin.h: Ditto.
9129         * config/i386/xmmintrin.h: Ditto.
9130         * config/i386/popcntintrin.h: Ditto.
9131         * config/i386/rdseedintrin.h: Ditto.
9132         * config/i386/ammintrin.h: Ditto.
9133         * config/i386/emmintrin.h: Ditto.
9134         * config/i386/immintrin.h: Remove all header include guards.
9135         * config/i386/fma4intrin.h: Ditto.
9136         * config/i386/lwpintrin.h: Ditto.
9137         * config/i386/xopintrin.h: Ditto.
9138         * config/i386/ia32intrin.h: Ditto.
9139         * config/i386/avxintrin.h: Ditto.
9140         * config/i386/rtmintrin.h: Ditto.
9141         * config/i386/fmaintrin.h: Ditto.
9142         * config/i386/mm3dnow.h: Ditto.
9144 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
9146         * common/config/i386/i386-common.c: Handle LZCNT.
9148 2013-06-22  Andi Kleen  <ak@linux.intel.com>
9150         * doc/extend.texi: Use __atomic_store_n instead of
9151         __atomic_store in HLE example.
9153 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
9155         * config/sh/sh.c: Remove <cstdlib> workaround.
9157 2013-06-21  Andi Kleen  <ak@linux.intel.com>
9159         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
9161 2013-06-21  Andi Kleen  <ak@linux.intel.com>
9163         * doc/extend.texi: Document that __atomic_clear and
9164           __atomic_test_and_set should only be used with bool.
9166 2013-06-20  Jan Hubicka  <jh@suse.cz>
9168         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
9169         types_same_for_odr.
9170         * tree.c (decls_same_for_odr): New function.
9171         (same_for_edr): New function.
9172         (types_same_for_odr): New function.
9173         (get_binfo_at_offset): Use it.
9174         * tree.h (types_same_for_odr): Declare.
9176 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
9177             Jason Merrill  <jason@redhat.com>
9179         * system.h: Include <cstdlib> as well as <stdlib.h>.
9181 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
9183         PR target/57655
9184         * config/i386/i386.c (construct_container): Report error if
9185         long double is used with disabled x87 float returns.
9187 2013-06-20  Jan Hubicka  <jh@suse.cz>
9189         * lto-cgraph.c (input_symtab): Do not set cgraph state.
9191 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
9193         PR rtl-optimization/57425
9194         PR rtl-optimization/57569
9195         * alias.c (write_dependence_p): Remove parameters mem_mode and
9196         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
9197         Changed all callers.
9198         (canon_anti_dependence): Get comments and semantics in sync.
9199         Add parameter mem_canonicalized.  Changed all callers.
9200         * rtl.h (canon_anti_dependence): Update prototype.
9202 2013-06-20  Richard Biener  <rguenther@suse.de>
9204         * data-streamer-in.c (streamer_read_uhwi): Optimize single
9205         byte case, inline streamer_read_uchar and defer section
9206         overrun check.
9208 2013-06-20  Richard Biener  <rguenther@suse.de>
9210         PR tree-optimization/57584
9211         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
9212         SSA names into the expanded expression that take part in
9213         abnormal coalescing.
9215 2013-06-19  Sharad Singhai  <singhai@google.com>
9217         * gcov.c (print_usage): Handle new option.
9218         (process_args): Ditto.
9219         (get_gcov_intermediate_filename): New function.
9220         (output_intermediate_file): New function.
9221         (output_gcov_file): New function
9222         (generate_results): Handle new option.
9223         (release_function): Relase demangled name.
9224         (read_graph_file): Handle demangled name.
9225         (output_lines): Ditto.
9226         * doc/gcov.texi: Document gcov intermediate format.
9228 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
9230         PR bootstrap/57604
9231         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
9232         (lra_emit_add): Use the functions.  Add comment about Y as an
9233         address segment.
9235 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
9237         PR driver/57652
9238         * collect2.c (collect_atexit): New.
9239         (collect_exit): Delete.
9240         (main): Register collect_atexit with atexit.
9241         (collect_wait): Change collect_exit to exit.
9242         (do_wait): Same.
9243         * collect2.h (collect_exit): Delete.
9244         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
9246 2013-06-19  Wei Mi  <wmi@google.com>
9248         PR rtl-optimization/57518
9249         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
9250         if regno is used in paradoxical subreg.
9251         (update_equiv_regs): Check pdx_subregs[regno] before
9252         set a reg to be equivalent with a mem.
9254 2013-06-19  Matthias Klose  <doko@ubuntu.com>
9256         PR driver/57651
9257         * file-find.h (find_a_file): Add a mode parameter.
9258         * file-find.c (find_a_file): Likewise.
9259         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
9260         with X_OK for the executables.
9261         * collect2.c (main): Call find_a_file with X_OK.
9263 2013-06-19  Steve Ellcey  <sellcey@mips.com>
9265         PR target/56942
9266         * config/mips/mips.md (casesi_internal_mips16_<mode>):
9267         Use NEXT_INSN instead of next_real_insn.
9269 2013-06-19  Jan Hubicka  <jh@suse.cz>
9271         * cgraph.h (const_value_known_p): Replace by ...
9272         (ctor_for_folding): .. this one.
9273         * cgraphunit.c (process_function_and_variable_attributes): Use it.
9274         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
9275         * expr.c (expand_expr_real_1): Likewise.
9276         (string_constant): Likewise.
9277         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
9278         * ipa.c (process_references): Likewise.
9279         (symtab_remove_unreachable_nodes): Likewise.
9280         * ipa-inline-analysis.c (param_change_prob): Likewise.
9281         * gimple-fold.c (canonicalize_constructor_val): Likewise.
9282         (get_base_constructor): Likwise.
9283         * varpool.c (varpool_remove_node): Likewise.
9284         (varpool_remove_initializer): LIkewise.
9285         (dump_varpool_node): LIkwise.
9286         (const_value_known_p): Rewrite to ...
9287         (ctor_for_folding): ... this one.
9289 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
9291         PR driver/57651
9292         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
9293         PERSONALITY in $PATH derived prefixes.
9295 2013-06-19  Jeff Law  <law@redhat.com>
9297         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
9298         in comment.
9300         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
9301         (simplify_bitwise_binary): Use it to simpify certain binary ops on
9302         booleans.
9304 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
9306         * config/arm/vfp.md: Move VFP instruction classification documentation
9307         to ...
9308         * config/arm/arm.md: ... here.  Update instruction classification
9309         documentation.
9311 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
9313         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
9314         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
9315         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
9317 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
9319         PR target/57609
9320         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
9321         with NEXT_INSN.  Use tablejump_p to check for jump table data
9322         insns.
9324 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
9326         PR c++/56544
9327         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
9328         that now in C++ the value is correct per the C++ standards.
9330 2013-06-19  Richard Biener  <rguenther@suse.de>
9332         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
9333         for global context.
9335 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9337         Revert:
9338         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9340         PR target/57609
9341         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
9342         with next_active_insn.
9344 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
9346         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
9347         functions are inlined during failures to flag an error.
9348         * tree-inline.c (expand_call_inline): Allow the error to be flagged
9349         in early inline pass.
9351 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
9353         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
9354         in comments.
9356 2013-06-18  Julian Brown  <julian@codesourcery.com>
9358         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
9359         Permit virtual register pre-reload if !strict.
9360         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
9361         change.
9362         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
9363         prototype.
9364         * config/arm/neon.md (movmisalign<mode>): Use
9365         neon_perm_struct_or_reg_operand instead of
9366         neon_struct_or_register_operand.
9367         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
9368         neon_permissive_struct_operand instead of neon_struct_operand.
9369         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
9370         neon_vector_mem_operand.
9371         * config/arm/predicates.md (neon_struct_operand): Adjust call to
9372         neon_vector_mem_operand.
9373         (neon_permissive_struct_operand): New.
9374         (neon_struct_or_register_operand): Rename to...
9375         (neon_perm_struct_or_reg_operand): This. Adjust call to
9376         neon_vector_mem_operand.
9378 2013-06-18  Richard Biener  <rguenther@suse.de>
9380         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
9381         * lto-streamer.h: Include pointer-set.h.
9382         (struct lto_decl_slot): Remove.
9383         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
9384         Remove next_index entry.
9385         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
9386         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
9387         (lto_init_tree_ref_encoder): Adjust.
9388         (lto_destroy_tree_ref_encoder): Likewise.
9389         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
9390         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
9391         (lto_output_decl_index): Adjust.
9392         (lto_new_out_decl_state): Likewise.
9393         (lto_record_function_out_decl_state): Likewise.
9394         * lto-streamer-out.c (copy_function): Likewise.
9396 2013-06-18  Richard Biener  <rguenther@suse.de>
9398         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
9399         * cgraphunit.c: Include cfgloop.h.
9400         (init_lowered_empty_function): Initialize the loop tree.
9401         (assemble_thunk): Insert new BBs into loops.
9403 2013-06-18  Richard Biener  <rguenther@suse.de>
9405         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
9406         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
9407         the map from cache entry to cache index optional.
9408         (streamer_tree_cache_replace_tree): Adjust accordingly.
9409         (streamer_tree_cache_append): Likewise.
9410         (streamer_tree_cache_delete): Likewise.
9411         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
9412         streamer cache map from cache entry to cache index.
9413         * lto-streamer-out.c (create_output_block): Adjust.
9415 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
9417         * config/arm/arm.md (attribute "insn"): Move multiplication and
9418         division attributes to...
9419         (attribute "type"): ... here.  Remove mult.
9420         (attribute "mul32"): New attribute.
9421         (attribute "mul64"): Add umaal.
9422         (*arm_mulsi3): Update attributes.
9423         (*arm_mulsi3_v6): Likewise.
9424         (*thumb_mulsi3): Likewise.
9425         (*thumb_mulsi3_v6): Likewise.
9426         (*mulsi3_compare0): Likewise.
9427         (*mulsi3_compare0_v6): Likewise.
9428         (*mulsi_compare0_scratch): Likewise.
9429         (*mulsi_compare0_scratch_v6): Likewise.
9430         (*mulsi3addsi): Likewise.
9431         (*mulsi3addsi_v6): Likewise.
9432         (*mulsi3addsi_compare0): Likewise.
9433         (*mulsi3addsi_compare0_v6): Likewise.
9434         (*mulsi3addsi_compare0_scratch): Likewise.
9435         (*mulsi3addsi_compare0_scratch_v6): Likewise.
9436         (*mulsi3subsi): Likewise.
9437         (*mulsidi3adddi): Likewise.
9438         (*mulsi3addsi_v6): Likewise.
9439         (*mulsidi3adddi_v6): Likewise.
9440         (*mulsidi3_nov6): Likewise.
9441         (*mulsidi3_v6): Likewise.
9442         (*umulsidi3_nov6): Likewise.
9443         (*umulsidi3_v6): Likewise.
9444         (*umulsidi3adddi): Likewise.
9445         (*umulsidi3adddi_v6): Likewise.
9446         (*smulsi3_highpart_nov6): Likewise.
9447         (*smulsi3_highpart_v6): Likewise.
9448         (*umulsi3_highpart_nov6): Likewise.
9449         (*umulsi3_highpart_v6): Likewise.
9450         (mulhisi3): Likewise.
9451         (*mulhisi3tb): Likewise.
9452         (*mulhisi3bt): Likewise.
9453         (*mulhisi3tt): Likewise.
9454         (maddhisi4): Likewise.
9455         (*maddhisi4tb): Likewise.
9456         (*maddhisi4tt): Likewise.
9457         (maddhidi4): Likewise.
9458         (*maddhidi4tb): Likewise.
9459         (*maddhidi4tt): Likewise.
9460         (divsi3): Likewise.
9461         (udivsi3): Likewise.
9462         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
9463         (thumb2_mulsi_short_compare0): Likewise.
9464         (thumb2_mulsi_short_compare0_scratch): Likewise.
9465         * config/arm/arm1020e.md (1020mult1): Update attribute change.
9466         (1020mult2): Likewise.
9467         (1020mult3): Likewise.
9468         (1020mult4): Likewise.
9469         (1020mult5): Likewise.
9470         (1020mult6): Likewise.
9471         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
9472         change.
9473         (cortex_a15_mult64): Likewise.
9474         (cortex_a15_sdiv): Likewise.
9475         (cortex_a15_udiv): Likewise.
9476         * config/arm/arm1026ejs.md (mult1): Update attribute change.
9477         (mult2): Likewise.
9478         (mult3): Likewise.
9479         (mult4): Likewise.
9480         (mult5): Likewise.
9481         (mult6): Likewise.
9482         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
9483         (pj4_ir_div): Likewise.
9484         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
9485         (11_mult2): Likewise.
9486         (11_mult3): Likewise.
9487         (11_mult4): Likewise.
9488         (11_mult5): Likewise.
9489         (11_mult6): Likewise.
9490         (11_mult7): Likewise.
9491         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
9492         (cortex_a8_mla): Likewise.
9493         (cortex_a8_mull): Likewise.
9494         (cortex_a8_smulwy): Likewise.
9495         (cortex_a8_smlald): Likewise.
9496         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
9497         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
9498         (cortex_r4_mul_3): Likewise.
9499         (cortex_r4_mla_4): Likewise.
9500         (cortex_r4_mla_3): Likewise.
9501         (cortex_r4_smlald): Likewise.
9502         (cortex_r4_mull): Likewise.
9503         (cortex_r4_sdiv): Likewise.
9504         (cortex_r4_udiv): Likewise.
9505         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
9506         (cortex_a7_idiv): Likewise.
9507         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
9508         (9_mult2): Likewise.
9509         (9_mult3): Likewise.
9510         (9_mult4): Likewise.
9511         (9_mult5): Likewise.
9512         (9_mult6): Likewise.
9513         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
9514         (cortex_a53_sdiv): Likewise.
9515         (cortex_a53_udiv): Likewise.
9516         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
9517         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
9518         (mp626_mult2): Likewise.
9519         (mp626_mult3): Likewise.
9520         (mp626_mult4): Likewise.
9521         * config/arm/fa526.md (526_mult1): Update attribute change.
9522         (526_mult2): Likewise.
9523         * config/arm/arm-generic.md (mult): Update attribute change.
9524         (mult_ldsched_strongarm): Likewise.
9525         (mult_ldsched): Likewise.
9526         (multi_cycle): Likewise.
9527         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
9528         * config/arm/fa606te.md (606te_mult1): Update attribute change.
9529         (606te_mult2): Likewise.
9530         (606te_mult3): Likewise.
9531         (606te_mult4): Likewise.
9532         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
9533         (cortex_a9_mac16): Likewise.
9534         (cortex_a9_multiply): Likewise.
9535         (cortex_a9_mac): Likewise.
9536         (cortex_a9_multiply_long): Likewise.
9537         * config/arm/fa626te.md (626te_mult1): Update attribute change.
9538         (626te_mult2): Likewise.
9539         (626te_mult3): Likewise.
9540         (626te_mult4): Likewise.
9542 2013-06-18  Richard Biener  <rguenther@suse.de>
9544         PR lto/57334
9545         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
9547 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9549         PR target/57609
9550         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
9551         with next_active_insn.
9553 2013-06-18  Alan Modra  <amodra@gmail.com>
9555         * config/rs6000/rs6000.h (enum data_align): New.
9556         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
9557         (DATA_ABI_ALIGNMENT): Define.
9558         (CONSTANT_ALIGNMENT): Correct comment.
9559         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
9560         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
9562 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
9564         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
9565         ATTRIBUTE_UNUSED marking.
9567 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
9569         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
9570         alternative and update.
9571         (aarch64_dup_lanedi): Delete.
9572         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
9573         * config/aarch64/aarch64-simd-builtins.def: Update.
9575 2013-06-17  Richard Biener  <rguenther@suse.de>
9577         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
9578         (lto_input_scc): Declare.
9579         (lto_input_tree_1): Likewise.
9580         (struct lto_stats_d): Add num_tree_bodies_output and
9581         num_pickle_refs_output.
9582         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
9583         (lto_read_tree_1): Split out from ...
9584         (lto_read_tree): ... this.
9585         (lto_input_scc): New function.
9586         (lto_input_tree_1): Split out from ...
9587         (lto_input_tree): ... this.  Handle LTO_tree_scc.
9588         (lto_data_in_create): Create the streamer cache without hashes.
9589         * lto-streamer-out.c (create_output_block): Create the streamer
9590         cache with hashes when not doing WPA.
9591         (lto_write_tree_1): Split out from ...
9592         (lto_write_tree): ... this.
9593         (get_symbol_initial_value): New function.
9594         (lto_output_tree_1): Split out from ...
9595         (lto_output_tree): ... this.  Write trees as series of SCCs
9596         using a DFS walk via DFS_write_tree.
9597         (struct sccs, struct scc_entry): New types.
9598         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
9599         (DFS_write_tree_body): New function.
9600         (DFS_write_tree): Likewise.
9601         (hash_tree): Likewise.
9602         (scc_entry_compare): Likewise.
9603         (hash_scc): Likewise.
9604         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
9605         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
9606         TREE_CHAIN as regular reference.
9607         (streamer_read_integer_cst): Remove.
9608         (streamer_get_pickled_tree): Adjust.
9609         * tree-streamer-out.c (streamer_write_chain): Disable streaming
9610         of DECL_EXTERNALs in BLOCK_VARS for now.
9611         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
9612         reference.
9613         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
9614         Add hash value argument and record that if hashes are recorded
9615         in the cache.
9616         (streamer_tree_cache_insert_1): Adjust.
9617         (streamer_tree_cache_insert): Likewise.
9618         (streamer_tree_cache_insert_at): Rename to ...
9619         (streamer_tree_cache_replace_tree): ... this and adjust.
9620         (streamer_tree_cache_append): Adjust.
9621         (record_common_node): Likewise.
9622         (streamer_tree_cache_create): Add argument whether to
9623         record hash values together with trees.
9624         (streamer_tree_cache_delete): Adjust.
9625         * tree-streamer.h (struct streamer_tree_cache_d): Add
9626         vector of hashes.
9627         (streamer_read_integer_cst): Remove.
9628         (streamer_tree_cache_insert): Adjust.
9629         (streamer_tree_cache_append): Likewise.
9630         (streamer_tree_cache_insert_at): Rename to ...
9631         (streamer_tree_cache_replace_tree): ... this and adjust.
9632         (streamer_tree_cache_create): Add argument whether to record hashes.
9633         (streamer_tree_cache_get): Rename to ...
9634         (streamer_tree_cache_get_tree): ... this.
9635         (streamer_tree_cache_get_hash): New function.
9636         * tree.c (cache_integer_cst): New function.
9637         * tree.h (cache_integer_cst): Declare.
9638         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
9639         * lto-symtab.c (lto_varpool_replace_node): Only release
9640         DECL_INITIAL of non-prevailing decls.
9641         * varpool.c (varpool_remove_initializer): Do not release
9642         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
9644 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
9646         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
9647         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
9648         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
9649         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
9650         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
9651         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
9652         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
9653         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
9654         instead of TARGET_64BIT.
9655         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
9656         Require ISA_HAS_<D>DIV.
9658 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
9660         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
9661         (mips*-*-linux*): Move default with_llsc setting to where other
9662         defaults are set.
9663         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
9664         with_arch block.
9665         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
9666         Likewise.  Remove default with_tune setting.  Move default float
9667         setting to its own block.  Handle with_llsc in the same block as above.
9669 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
9671         PR rtl-optimization/57425
9672         PR rtl-optimization/57569
9673         * alias.c (write_dependence_p): Add new parameters mem_mode,
9674         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
9675         Changed all callers.
9676         (canon_anti_dependence): New function.
9677         * cse.c (check_dependence): Use canon_anti_dependence.
9678         * cselib.c (cselib_invalidate_mem): Likewise.
9679         * rtl.h (canon_anti_dependence): Declare.
9681 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
9683         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
9684         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
9685         ".set mips3" for 64-bit targets.
9687 2013-06-15  Dehao Chen  <dehao@google.com>
9689         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
9690         * gimple-low.c (gimple_check_call_matching_types): Likewise.
9691         (gimple_check_call_args): Likewise.
9692         * value-prof.c (check_ic_target): Likewise.
9693         * ipa-inline.c (early_inliner): Likewise.
9694         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
9695         * cgraph.c (cgraph_create_edge_1): Likewise.
9696         (cgraph_make_edge_direct): Likewise.
9698 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
9700         PR target/57615
9701         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
9702         rs6000_output_move_128bit to handle emitting quad memory
9703         operations.  Set attribute length to 8 bytes.
9705 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
9707         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
9708         New pattern.
9709         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
9710         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
9711         (aarch64_<su>mlsl<mode>): Likewise.
9713 2013-06-14  Mike Stump  <mikestump@comcast.net>
9715         * Makefile.in (TARGET_H): Add insn-codes.h.
9717 2013-06-14  Alan Modra  <amodra@gmail.com>
9719         PR middle-end/57134
9720         PR middle-end/57586
9721         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
9722         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
9723         bitfield expansion when EXPAND_MEMORY.
9724         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
9726 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
9728         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
9729         test for clearing quad memory on 32-bit later.
9731 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
9733         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
9734         (fold_negate_expr): Likewise.
9735         (fold_real_zero_addition_p): Handle vectors.
9736         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
9738 2013-06-14  Alan Modra  <amodra@gmail.com>
9740         * varasm.c (force_const_mem): Revert 2013-06-07 change.
9742 2013-06-13  Jan Hubicka  <jh@suse.cz>
9744         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
9745         Local comdats are not externally visible.
9746         * symtab.c (dump_symtab_base): Dump externally visible.
9747         (verify_symtab_base): Verify back links in the symtab hash.
9749 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
9751         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
9752         CONVERT_EXPR as equal nodes.
9754 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
9756         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
9758 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
9760         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
9761         Generalize to complex and vector.
9762         * tree.c (build_all_ones_cst): New function.
9763         * tree.h (build_all_ones_cst): Declare it.
9765 2013-06-13  Alan Modra  <amodra@gmail.com>
9767         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
9768         * config/rs6000/rs6000.md (signbittf2): New insn.
9769         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
9770         (abstf2_internal, cmptf_internal2): Likewise.
9771         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
9773 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
9774             Pat Haugen  <pthaugen@us.ibm.com>
9775             Peter Bergner  <bergner@vnet.ibm.com>
9777         * config/rs6000/rs6000.c (emit_load_locked): Add support for
9778         power8 byte, half-word, and quad-word atomic instructions.
9779         (emit_store_conditional): Likewise.
9780         (rs6000_expand_atomic_compare_and_swap): Likewise.
9781         (rs6000_expand_atomic_op): Likewise.
9783         * config/rs6000/sync.md (larx): Add new modes for power8.
9784         (stcx): Likewise.
9785         (AINT): New mode iterator to include TImode as well as normal
9786         integer modes on power8.
9787         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
9788         that VSX registers are not considered.  Use AINT mode iterator
9789         instead of INT1 to allow inclusion of quad word atomic operations
9790         on power8.
9791         (load_locked<mode>): Likewise.
9792         (store_conditional<mode>): Likewise.
9793         (atomic_compare_and_swap<mode>): Likewise.
9794         (atomic_exchange<mode>): Likewise.
9795         (atomic_nand<mode>): Likewise.
9796         (atomic_fetch_<fetchop_name><mode>): Likewise.
9797         (atomic_nand_fetch<mode>): Likewise.
9798         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
9799         each type.
9800         (ATOMIC): On power8, add QImode, HImode modes.
9801         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
9802         modes that promote to SImode.
9803         (load_lockedti): Convert TImode arguments to PTImode, so that we
9804         get a guaranteed even/odd register pair.
9805         (load_lockedpti): Likewise.
9806         (store_conditionalti): Likewise.
9807         (store_conditionalpti): Likewise.
9809         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
9810         atomic load/store instructions.
9811         (HSI): Likewise.
9813 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
9815         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
9816         loads.
9817         (insn_count): New attribute, with most cases extracted from...
9818         (length): ...here.  Redefine most cases in terms of insn_count.
9819         (single_insn): Delete.
9820         (can_delay): Use insn_count to check for single instructions.
9821         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
9822         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
9823         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
9824         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
9825         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
9826         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
9827         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
9828         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
9829         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
9830         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
9831         rather than "length".
9832         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
9833         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
9834         Use "insn_count" rather than "length".
9835         * config/mips/mips-dsp.md
9836         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
9837         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
9838         length attributes.
9840 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
9842         PR tree-optimization/57361
9843         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
9845 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
9847         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
9848         to split.
9849         (aarch64_simd_combine<mode>): New instruction expansion.
9850         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
9851         function prototype.
9852         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
9853         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
9855 2013-06-12  Jan Hubicka  <jh@suse.cz>
9857         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
9858         decl has when in streaming stage.
9859         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
9860         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
9862 2013-06-12  Roland Stigge  <stigge@antcom.de>
9864         PR target/57578
9865         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
9867 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
9869         PR tree-optimization/57537
9870         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
9871         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
9873 2013-06-12  Richard Biener  <rguenther@suse.de>
9875         * data-streamer.h (streamer_write_char_stream): CSE
9876         obs->current_pointer.
9877         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
9878         streamer_write_char_stream manually and optimize the resulting loop.
9879         (streamer_write_hwi_stream): Likewise.
9881 2013-06-12  Jan Hubicka  <jh@suse.cz>
9883         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
9884         * cgraph.h (varpool_create_empty_node): Declare.
9885         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
9886         duplicated nodes.
9887         * symtab.c (symtab_unregister_node): Be lax about missin entries
9888         in node hash.
9889         (symtab_get_node): Update comment.
9890         * varpool.c (varpool_create_empty_node): Break out from ...
9891         (varpool_node_for_decl): ... here.
9892         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
9894 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
9896         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
9897         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
9898         part.  Use straight-line flow at the end.
9899         <COMPONENT_REF>: Remove superfluous else.
9900         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
9902 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
9904         PR target/56564
9905         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
9906         target hook even for !TREE_PUBLIC decls.  If no resolution info
9907         is available, return false for common and external decls.
9909 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
9911         * config/rl78/constraints.md (U): New constraint.
9912         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
9913         valloc attribute.
9915 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
9917         PR target/57589
9918         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
9919         to allow returning address to AT_PLATFORM name.
9921 2013-06-11  Jan Hubicka  <jh@suse.cz>
9923         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
9924         * cgraph.h (symtab_node_base): Add weakref flag.
9925         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
9926         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
9927         (output_weakrefs): Use weakref flag.
9928         * fold-const.c (simple_operand_p): Handle WEAK.
9929         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
9930         * ipa.c (varpool_externally_visible_p): Drop weakref.
9931         (function_and_variable_visibility): Update comment; fix weakref
9932         sanity checks; do not clear DECL_WEAK on them.
9933         * lto-cgraph.c (lto_output_node): update.
9934         (lto_output_varpool_node): Update.
9935         (input_overwrite_node): Update.
9936         (input_node): Update.
9937         (input_varpool_node): Update.
9938         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
9939         (lto_symtab_merge_symbols): Add sanity check.
9940         (lto_symtab_prevailing_decl): Do not special case weakrefs.
9941         * passes.c (rest_of_decl_compilation): Set static flag, too.
9942         * symtab.c (dump_symtab_base): Dump weakref.
9943         (verify_symtab_base): Sanity check weakrefs.
9944         (symtab_make_decl_local): Remove duplicated code.
9945         (symtab_alias_ultimate_target): Simplify.
9946         * varpool.c (varpool_create_variable_alias): Set weakref flag.
9948 2013-06-11  Tom de Vries  <tom@codesourcery.com>
9950         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
9951         sequence_vect == NULL.
9953 2013-06-11  DJ Delorie  <dj@redhat.com>
9955         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
9956         (rl78_unwind_word_mode): New.
9958 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
9960         * final.c (debug_prefix_maps): Make static.
9962 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
9964         * function.c (initial_trampoline): Remove stray copy.
9966 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
9968         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
9970 2013-06-11  Martin Jambor  <mjambor@suse.cz>
9972         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
9973         within bounds at the beginning of the function.
9975 2013-06-11  Alan Modra  <amodra@gmail.com>
9977         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
9978         reporting.
9979         (get_named_section): Don't NULL !DECL_P decl.
9981 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
9983         * doc/invoke.texi (core-avx2): Document.
9984         (slm): Likewise.
9985         (atom): Updated with MOVBE.
9987 2013-06-11  Richard Biener  <rguenther@suse.de>
9989         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
9991 2013-06-11  Anton Blanchard  <anton@samba.org>
9993         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
9994         correct shift value in little-endian mode.
9996 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
9998         PR target/56564
9999         * varasm.c (get_variable_align): Move #endif to the right place.
10001 2013-06-10  Cary Coutant  <ccoutant@google.com>
10003         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
10004         for hash so that hash table traversal order is deterministic.
10006 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
10007             Pat Haugen  <pthaugen@us.ibm.com>
10008             Peter Bergner  <bergner@vnet.ibm.com>
10010         * config/rs6000/vector.md (GPR move splitter): Do not split moves
10011         of vectors in GPRS if they are direct moves or quad word load or
10012         store moves.
10014         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
10015         declaration.
10016         (direct_move_p): Likewise.
10017         (quad_load_store_p): Likewise.
10019         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
10020         classes into bins based on the physical register type.
10021         (reg_class_to_reg_type): Likewise.
10022         (IS_STD_REG_TYPE): Likewise.
10023         (IS_FP_VECT_REG_TYPE): Likewise.
10024         (reload_fpr_gpr): Arrays to determine what insn to use if we can
10025         use direct move instructions.
10026         (reload_gpr_vsx): Likewise.
10027         (reload_vsx_gpr): Likewise.
10028         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
10029         information that is a simplification of register classes.  Also
10030         precalculate direct move reload helpers.
10031         (direct_move_p): New function to return true if the operation can
10032         be done as a direct move instruciton.
10033         (quad_load_store_p): New function to return true if the operation
10034         is a quad memory operation.
10035         (rs6000_legitimize_address): If quad memory, only allow register
10036         indirect for TImode addresses.
10037         (rs6000_legitimate_address_p): Likewise.
10038         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
10039         (rs6000_reload_register_type): Likewise.
10040         (register_to_reg_type): Return register type.
10041         (rs6000_secondary_reload_simple_move): New helper function for
10042         secondary reload and secondary memory needed to identify anything
10043         that is a simple move, and does not need reloading.
10044         (rs6000_secondary_reload_direct_move): New helper function for
10045         secondary reload to identify cases that can be done with several
10046         instructions via the direct move instructions.
10047         (rs6000_secondary_reload_move): New helper function for secondary
10048         reload to identify moves between register types that can be done.
10049         (rs6000_secondary_reload): Add support for quad memory operations
10050         and for direct move.
10051         (rs6000_secondary_memory_needed): Likewise.
10052         (rs6000_debug_secondary_memory_needed): Change argument names.
10053         (rs6000_output_move_128bit): New function to return the move to
10054         use for 128-bit moves, including knowing about the various
10055         limitations of quad memory operations.
10057         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
10058         memory operations.  call rs6000_output_move_128bit for the actual
10059         instruciton(s) to generate.
10060         (vsx_movti_64bit): Likewise.
10062         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
10063         (UNSPEC_P8V_MTVSRWZ): Likewise.
10064         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
10065         (UNSPEC_P8V_MTVSRD): Likewise.
10066         (UNSPEC_P8V_XXPERMDI): Likewise.
10067         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
10068         (UNSPEC_FUSION_GPR): Likewise.
10069         (FMOVE128_GPR): New iterator for direct move.
10070         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
10071         (f32_sv): Likewise.
10072         (f32_dm): Likewise.
10073         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
10074         loads and direct move instructions.
10075         (zero_extendsidi2_lfiwzx): Likewise.
10076         (extendsidi2_lfiwax): Likewise.
10077         (extendsidi2_nocell): Likewise.
10078         (floatsi<mode>2_lfiwax): Likewise.
10079         (lfiwax): Likewise.
10080         (floatunssi<mode>2_lfiwzx): Likewise.
10081         (lfiwzx): Likewise.
10082         (fix_trunc<mode>_stfiwx): Likewise.
10083         (fixuns_trunc<mode>_stfiwx): Likewise.
10084         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
10085         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
10086         (parity<mode>2_cmpb): Set length/type attr.
10087         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
10088         for 'mr.' to fast_compare.
10089         (bpermd_<mode>): Change type attr to popcnt.
10090         (p8_fmrgow_<mode>): New insns for power8 direct move support.
10091         (p8_mtvsrwz_1): Likewise.
10092         (p8_mtvsrwz_2): Likewise.
10093         (reload_fpr_from_gpr<mode>): Likewise.
10094         (p8_mtvsrd_1): Likewise.
10095         (p8_mtvsrd_2): Likewise.
10096         (p8_xxpermdi_<mode>): Likewise.
10097         (reload_vsx_from_gpr<mode>): Likewise.
10098         (reload_vsx_from_gprsf): Likewise.
10099         (p8_mfvsrd_3_<mode>): LIkewise.
10100         (reload_gpr_from_vsx<mode>): Likewise.
10101         (reload_gpr_from_vsxsf): Likewise.
10102         (p8_mfvsrd_4_disf): Likewise.
10103         (multi-word GPR splits): Do not split direct moves or quad memory
10104         operations.
10106 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
10108         * tree-into-ssa.c (interesting_blocks): Make static.
10110 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
10112         PR target/56564
10113         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
10114         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
10115         Use DATA_ABI_ALIGNMENT for that case instead if defined.
10116         (get_variable_align): New function.
10117         (get_variable_section, emit_bss, emit_common,
10118         assemble_variable_contents, place_block_symbol): Use
10119         get_variable_align instead of DECL_ALIGN.
10120         (assemble_noswitch_variable): Add align argument, use it
10121         instead of DECL_ALIGN.
10122         (assemble_variable): Adjust caller.  Use get_variable_align
10123         instead of DECL_ALIGN.
10124         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
10125         caller.
10126         (DATA_ABI_ALIGNMENT): Define.
10127         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
10128         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
10129         opt is false, only return the psABI mandated alignment increase.
10130         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
10131         (DATA_ABI_ALIGNMENT): ... this.
10132         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
10133         (DATA_ABI_ALIGNMENT): ... this.
10134         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
10135         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
10136         (DATA_ABI_ALIGNMENT): ... this.
10137         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
10138         * doc/tm.texi: Regenerated.
10140 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
10142         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
10143         cmp_code to construct REG_EQUAL note.
10145 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
10147         PR target/57568
10148         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
10149         that operands[2] doesn't overlap with operands[0].
10151 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
10152             Jan Hubicka  <jh@suse.cz>
10154         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
10155         hack to mark symbols as used.
10157 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
10159         PR rtl-optimization/57559
10160         * lra-constraints.c (process_alt_operands): Don't discourage
10161         memory with known offset for offsetable memory constraint.
10162         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
10164 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
10166         * varasm.c (struct oc_local_state): Reorder fields.
10167         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
10168         and adjust accordingly.
10169         (output_constructor): Reorder initialization code and adjust call to
10170         output_constructor_bitfield.
10172 2013-06-07  Jan Hubicka  <jh@suse.cz>
10174         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
10176 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
10178         * tree-object-size.c (unknown): Make const.
10180 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
10182         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
10183         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
10184         for last alternative in the cpu_facility attribute.
10186 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10188         PR target/56315
10189         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
10190         (xordi3): Change operand 2 constraint to arm_xordi_operand.
10191         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
10192         * config/arm/constraints.md (Dg): New constraint.
10193         * config/arm/neon.md (xordi3_neon): Remove.
10194         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
10195         * config/arm/predicates.md (arm_xordi_operand): New predicate.
10197 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10199         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
10200         Clean up alternatives.
10202 2013-06-07  Alan Modra  <amodra@gmail.com>
10204         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
10205         va_list_gpr_size.
10207 2013-06-07  Alan Modra  <amodra@gmail.com>
10209         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
10211 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10213         * config/arm/constraints.md (Df): New constraint.
10214         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
10215         Correct length attribute for last two alternatives.
10217 2013-06-07  Alan Modra  <amodra@gmail.com>
10219         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
10220         override user -mfp-in-toc.
10221         (offsettable_ok_by_alignment): Consider just the current access
10222         rather than the whole object, unless BLKmode.  Handle
10223         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
10224         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
10225         for -mcmodel=medium.
10226         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
10227         override user -mfp-in-toc or -msum-in-toc.  Default to
10228         -mno-fp-in-toc for -mcmodel=medium.
10230 2013-06-06  DJ Delorie  <dj@redhat.com>
10232         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
10233         TARGET_VALID_POINTER_MODE.
10235 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
10236             Pat Haugen  <pthaugen@us.ibm.com>
10237             Peter Bergner  <bergner@vnet.ibm.com>
10239         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
10240         Document new power8 builtins.
10242         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
10243         condition code register, to allow 128-bit logical operations to be
10244         done in the VSX or GPR registers.
10245         (nor<mode>3): Use the canonical form for nor.
10246         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
10247         vclz*, and vpopcnt* vector instructions.
10248         (nand<mode>3): Likewise.
10249         (orc<mode>3): Likewise.
10250         (clz<mode>2): LIkewise.
10251         (popcount<mode>2): Likewise.
10253         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
10254         that only the GPRs are recognized.
10256         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10257         support for new power8 builtins.
10259         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
10260         builtin functions.
10261         (xscvdpspn): Likewise.
10262         (vclz): Likewise.
10263         (vclzb): Likewise.
10264         (vclzh): Likewise.
10265         (vclzw): Likewise.
10266         (vclzd): Likewise.
10267         (vpopcnt): Likewise.
10268         (vpopcntb): Likewise.
10269         (vpopcnth): Likewise.
10270         (vpopcntw): Likewise.
10271         (vpopcntd): Likewise.
10272         (vgbbd): Likewise.
10273         (vmrgew): Likewise.
10274         (vmrgow): Likewise.
10275         (eqv): Likewise.
10276         (eqv_v16qi3): Likewise.
10277         (eqv_v8hi3): Likewise.
10278         (eqv_v4si3): Likewise.
10279         (eqv_v2di3): Likewise.
10280         (eqv_v4sf3): Likewise.
10281         (eqv_v2df3): Likewise.
10282         (nand): Likewise.
10283         (nand_v16qi3): Likewise.
10284         (nand_v8hi3): Likewise.
10285         (nand_v4si3): Likewise.
10286         (nand_v2di3): Likewise.
10287         (nand_v4sf3): Likewise.
10288         (nand_v2df3): Likewise.
10289         (orc): Likewise.
10290         (orc_v16qi3): Likewise.
10291         (orc_v8hi3): Likewise.
10292         (orc_v4si3): Likewise.
10293         (orc_v2di3): Likewise.
10294         (orc_v4sf3): Likewise.
10295         (orc_v2df3): Likewise.
10297         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
10298         allow power8 quad mode in 64-bit.
10299         (rs6000_builtin_vectorized_function): Add support to vectorize
10300         ISA 2.07 count leading zeros, population count builtins.
10301         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
10302         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
10303         (builtin_function_type): Add vgbbd builtin function which takes an
10304         unsigned argument.
10305         (altivec_expand_vec_perm_const): Add support for new power8 merge
10306         instructions.
10308         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
10309         that does not include TImdoe for use with 32-bit.
10310         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
10311         instructions.
10312         (UNSPEC_VSX_CVDPSPN): Likewise.
10313         (vsx_xscvdpspn): Likewise.
10314         (vsx_xscvspdpn): Likewise.
10315         (vsx_xscvdpspn_scalar): Likewise.
10316         (vsx_xscvspdpn_directmove): Likewise.
10317         (vsx_and<mode>3): Split logical operations into 32-bit and
10318         64-bit. Add support to do logical operations on TImode as well as
10319         VSX vector types.  Allow logical operations to be done in either
10320         VSX registers or in general purpose registers in 64-bit mode.  Add
10321         splitters if GPRs were used. For AND, add clobber of CCmode to
10322         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
10323         encoding.
10324         (vsx_and<mode>3_32bit): Likewise.
10325         (vsx_and<mode>3_64bit): Likewise.
10326         (vsx_ior<mode>3): Likewise.
10327         (vsx_ior<mode>3_32bit): Likewise.
10328         (vsx_ior<mode>3_64bit): Likewise.
10329         (vsx_xor<mode>3): Likewise.
10330         (vsx_xor<mode>3_32bit): Likewise.
10331         (vsx_xor<mode>3_64bit): Likewise.
10332         (vsx_one_cmpl<mode>2): Likewise.
10333         (vsx_one_cmpl<mode>2_32bit): Likewise.
10334         (vsx_one_cmpl<mode>2_64bit): Likewise.
10335         (vsx_nor<mode>3): Likewise.
10336         (vsx_nor<mode>3_32bit): Likewise.
10337         (vsx_nor<mode>3_64bit): Likewise.
10338         (vsx_andc<mode>3): Likewise.
10339         (vsx_andc<mode>3_32bit): Likewise.
10340         (vsx_andc<mode>3_64bit): Likewise.
10341         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
10342         and xxlorc instructions.
10343         (vsx_eqv<mode>3_64bit): Likewise.
10344         (vsx_nand<mode>3_32bit): Likewise.
10345         (vsx_nand<mode>3_64bit): Likewise.
10346         (vsx_orc<mode>3_32bit): Likewise.
10347         (vsx_orc<mode>3_64bit): Likewise.
10349         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
10351         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
10352         instruction.
10353         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
10354         (p8_vmrgow): Likewise.
10355         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
10356         GPRs to be split under VSX.
10357         (p8v_clz<mode>2): Add power8 count leading zero support.
10358         (p8v_popcount<mode>2): Add power8 population count support.
10359         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
10360         support.
10362         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
10363         instruction.
10365         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
10366         builtin functions.
10367         (vec_nand): Likewise.
10368         (vec_vclz): Likewise.
10369         (vec_vclzb): Likewise.
10370         (vec_vclzd): Likewise.
10371         (vec_vclzh): Likewise.
10372         (vec_vclzw): Likewise.
10373         (vec_vgbbd): Likewise.
10374         (vec_vmrgew): Likewise.
10375         (vec_vmrgow): Likewise.
10376         (vec_vpopcnt): Likewise.
10377         (vec_vpopcntb): Likewise.
10378         (vec_vpopcntd): Likewise.
10379         (vec_vpopcnth): Likewise.
10380         (vec_vpopcntw): Likewise.
10382 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
10384         PR rtl-optimization/57468
10385         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
10386         spilled pseudos.
10388 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
10390         PR rtl-optimization/57459
10391         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
10392         type when setting live regs.
10394 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
10396         * config/s390/s390.opt (mlra): New option.
10397         * config/s390/s390.c (s390_decompose_address): Check displacement
10398         for all registers for LRA.
10399         (s390_secondary_reload): Don't used secondary reloads for LRA.
10400         (s390_lra_p): New function.
10401         (TARGET_LRA_P): Define.
10402         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
10403         of attribute cpu_facility to zarch for the last alternative.
10404         (*cmpmem_short): Ditto.
10406 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
10408         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
10409         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
10410         (arm_expand_prologue): Likewise.
10412 2013-06-06  Teresa Johnson  <tejohnson@google.com>
10414         PR c++/53743
10415         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
10416         as this is now done by redirect_edge_and_branch_force.
10417         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
10418         barriers, and fix interaction with splitting.
10419         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
10420         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
10421         reflect changes made in the routine.
10422         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
10423         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
10424         since this is called in cfglayout mode, and replace partition fixup
10425         with assert as that is now done by force_nonfallthru_and_redirect.
10426         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
10427         already be marked with region crossing note.
10428         (insert_section_boundary_note): Make non-static, gate on flag
10429         has_bb_partition, rewrite to also check for multiple partitions.
10430         (rest_of_handle_reorder_blocks): Remove call to
10431         insert_section_boundary_note, now done later during free_cfg.
10432         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
10433         * bb-reorder.h (insert_section_boundary_note): Declare.
10434         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
10435         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
10436         invoke insert_section_boundary_note.
10437         (try_redirect_by_replacing_jump): Remove unnecessary
10438         check for region crossing note.
10439         (fixup_partition_crossing): New function.
10440         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
10441         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
10442         in non-cfglayout mode.
10443         (force_nonfallthru_and_redirect): Fixup partition boundaries,
10444         remove old code that tried to do this. Emit barrier correctly
10445         when we are in cfglayout mode.
10446         (last_bb_in_partition): New function.
10447         (rtl_split_edge): Correctly fixup partition boundaries.
10448         (commit_one_edge_insertion): Remove old code that tried to
10449         fixup region crossing edge since this is now handled in
10450         split_block, and set up insertion point correctly since
10451         block may now end in a jump.
10452         (verify_hot_cold_block_grouping): Guard against checking when not in
10453         linearized RTL mode.
10454         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
10455         notes.
10456         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
10457         rtl_verify_flow_info, so not called in cfglayout mode.
10458         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
10459         (fixup_reorder_chain): Remove old code that attempted to fixup region
10460         crossing note as this is now handled in force_nonfallthru_and_redirect.
10461         (duplicate_insn_chain): Don't duplicate switch section notes.
10462         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
10463         note.
10464         * basic-block.h (emit_barrier_after_bb): Declare.
10466 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10468         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
10469         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
10470         arm_usatsihi): Adjust alternatives for arm_restrict_it.
10472 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10474         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
10475         where appropriate.
10476         * config/arm/ldmstm.md: Regenerate.
10478 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10480         * config/arm/sync.md (atomic_loaddi_1):
10481         Disable predication for arm_restrict_it.
10482         (arm_load_exclusive<mode>): Likewise.
10483         (arm_load_exclusivesi): Likewise.
10484         (arm_load_exclusivedi): Likewise.
10485         (arm_load_acquire_exclusive<mode>): Likewise.
10486         (arm_load_acquire_exclusivesi): Likewise.
10487         (arm_load_acquire_exclusivedi): Likewise.
10488         (arm_store_exclusive<mode>): Likewise.
10489         (arm_store_exclusive<mode>): Likewise.
10490         (arm_store_release_exclusivedi): Likewise.
10491         (arm_store_release_exclusive<mode>): Likewise.
10493 2013-06-06  Richard Biener  <rguenther@suse.de>
10495         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
10496         after LTO_null.
10497         (lto_tag_is_tree_code_p): Adjust.
10498         (lto_tag_is_gimple_code_p): Likewise.
10499         (lto_gimple_code_to_tag): Likewise.
10500         (lto_tag_to_gimple_code): Likewise.
10501         (lto_tree_code_to_tag): Likewise.
10502         (lto_tag_to_tree_code): Likewise.
10503         * data-streamer.h (streamer_write_hwi_in_range): Use
10504         uhwi streaming to stream the normalized range.
10505         (streamer_read_hwi_in_range): Likewise.
10507 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10509         * config/arm/arm.md (enabled_for_depr_it): New attribute.
10510         (predicable_short_it): Likewise.
10511         (predicated): Likewise.
10512         (enabled): Handle above.
10513         (define_cond_exec): Set predicated attribute to yes.
10515 2013-06-05  Mike Stump  <mikestump@comcast.net>
10517         * gdbinit.in (__FUNCTION__): Add.
10519 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
10521         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
10522         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
10524 2013-06-05  Jan Hubicka  <jh@suse.cz>
10526         * varasm.c (mark_decl_referenced): Revert the removal until targets
10527         are fixed.
10529 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
10531         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
10532         instead of mark_decl_referenced.
10534 2013-06-05  Jan Hubicka  <jh@suse.cz>
10536         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
10537         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
10538         and symtab_used_from_object_file_p.
10539         (cgraph_make_node_local_1): Clear forced_by_abi.
10540         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
10541         * cgraph.h (symtab_node_base): Add forced_by_abi.
10542         (decide_is_variable_needed): Remove.
10543         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
10544         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
10545         (decide_is_symbol_needed): ... this one; handle symbols in general;
10546         always analyze virtuals; honnor forced_by_abi.
10547         (cgraph_finalize_function): Update.
10548         (varpool_finalize_decl): Update.
10549         (symbol_defined_and_needed): Remove.
10550         (analyze_functions): Update.
10551         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
10552         output_refs, input_overwrite_node): Handle forced_by_abi.
10553         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
10554         (address_taken_from_non_vtable_p): ... this one.
10555         (comdat_can_be_unshared_p_1): New function.
10556         (cgraph_comdat_can_be_unshared_p): Rename to ...
10557         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
10558         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
10559         (function_and_variable_visibility): Clear forced_by_abi as needed.
10560         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
10561         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
10562         * symtab.c (dump_symtab_base): Dump forced_by_abi.
10563         * varpool.c (decide_is_variable_needed): Remove.
10565 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10567         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
10568         (arm_option_override): Override arm_restrict_it where appropriate.
10569         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
10570         * config/arm/arm.opt (mrestrict-it): New command-line option.
10571         * doc/invoke.texi: Document -mrestrict-it.
10573 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
10575         * tsan.c (tsan_atomic_table): Make const.
10577 2013-06-05  Richard Biener  <rguenther@suse.de>
10579         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
10580         index associated with the tree we are supposed to replace.
10581         * tree-streamer-out.c (pack_ts_base_value_fields): Output
10582         TREE_ASM_WRITTEN as zero for everything but SSA names.
10584 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
10586         * tree-ssa-structalias.c (call_stmt_vars): Make static.
10588 2013-06-04  Jan Hubicka  <jh@suse.cz>
10590         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
10591         (input_node, input_varpool_node): Handle correctly external same
10592         body aliases.
10593         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
10594         nodes at ltrans stage.
10596 2013-06-04  Jan Hubicka  <jh@suse.cz>
10598         * ipa-inline.c (update_caller_keys): Fix availability test.
10599         (update_callee_keys): Likewise.
10600         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
10601         to follow ELF standard.
10603 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
10605         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
10606         (mips64r5900el-*-elf*): New configurations.
10607         * config/mips/mips-cpus.def (r5900): New processor.
10608         * config/mips/mips-tables.opt: Regenerate.
10609         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
10610         (mips_issue_rate): Handle PROCESSOR_R5900.
10611         (mips_reorg_process_insns): Force reorder mode for the R5900.
10612         * config/mips/mips.h (TARGET_MIPS5900): Define.
10613         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
10614         TARGET_MIPS5900.
10615         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
10616         TARGET_MIPS5900.
10617         * config/mips/mips.md (processor): Add r5900.
10618         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
10620 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
10622         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
10623         into function to generate MOVI instruction.
10624         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
10625         (aarch64_preferred_simd_mode): Turn into wrapper.
10626         (aarch64_output_scalar_simd_mov_immediate): New function.
10627         * config/aarch64/aarch64-protos.h: Add prototype for above.
10629 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
10631         * config/aarch64/aarch64.c (simd_immediate_info): Remove
10632         element_char member.
10633         (sizetochar): Return signed char.
10634         (aarch64_simd_valid_immediate): Remove elchar and other
10635         unnecessary variables.
10636         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
10637         Calculate element_char as required.
10638         * config/aarch64/aarch64-protos.h: Update and move prototype
10639         for aarch64_output_simd_mov_immediate.
10640         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
10641         Update arguments.
10643 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
10645         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
10646         information completed by aarch64_simd_valid_immediate.
10647         (aarch64_legitimate_constant_p): Update arguments.
10648         (aarch64_simd_valid_immediate): Work with struct rather than many
10649         pointers.
10650         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
10651         (aarch64_simd_make_constant): Update arguments.
10652         (aarch64_output_simd_mov_immediate): Work with struct rather than
10653         many pointers.  Output immediate directly rather than as operand.
10654         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
10655         Update prototype.
10656         * config/aarch64/constraints.md (Dn): Update arguments.
10658 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
10660         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
10661         longer static.
10662         (aarch64_simd_immediate_valid_for_move): Remove.
10663         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
10664         (aarch64_simd_make_constant): Update call.
10665         (aarch64_output_simd_mov_immediate): Update call.
10666         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
10667         Add prototype.
10668         * config/aarch64/constraints.md (Dn): Update call.
10670 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
10672         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
10673         return type to bool for prototype.
10674         (aarch64_legitimate_constant_p): Check for true instead of not -1.
10675         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
10676         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
10678 2013-06-04  Catherine Moore  <clm@codesourcery.com>
10680         * config/mips/mips.opt (meva): New.
10681         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
10682         (ASM_SPEC): Handle -meva.
10683         * doc/invoke.texi (meva):  Document.
10685 2013-06-04  Alan Modra  <amodra@gmail.com>
10687         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
10688         constant output.
10690 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10692         * rtl.def: Add extra fourth optional field to define_cond_exec.
10693         * gensupport.c (process_one_cond_exec): Process attributes from
10694         define_cond_exec.
10695         * doc/md.texi: Document fourth field in define_cond_exec.
10697 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
10699         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
10700         out the processing order as in store_bit_field_1.
10702 2013-06-04  Jan Hubicka  <jh@suse.cz>
10704         PR middle-end/57500
10705         * cgraphunit.c (cgraph_process_same_body_aliases): Create
10706         non-VAR_DECL node if it does not exist yet.
10708 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
10710         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
10711         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
10712         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
10713         target_cpu_default setting.
10715 2013-06-03  Teresa Johnson  <tejohnson@google.com>
10717         * dumpfile.c (opt_info_switch_p): Change -fopt-info
10718         default to -fopt-info=optimized instead of all.
10719         * doc/invoke.texi: Ditto.
10720         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
10721         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
10722         (execute_vect_slp): Emit BB vectorization success under
10723         MSG_OPTIMIZED_LOCATIONS.
10724         * tree-vect-slp.c (vect_slp_transform_bb): Change
10725         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
10726         * tree-vect-loop.c (vect_transform_loop): Ditto.
10728 2013-06-03  Jason Merrill  <jason@redhat.com>
10730         PR c++/57415
10731         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10732         Use TARGET_EXPR for C++.
10734 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
10736         PR rtl-optimization/57268
10737         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
10738         if DEBUG_INSN_P (insn).
10740         Reapply
10741         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
10743         PR rtl-optimization/57268
10744         * sched-deps.c (sched_analyze_2): Flush dependence lists if
10745         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
10747 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
10749         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
10750         (ix86_avoid_lea_for_addr): Likewise.
10751         (exact_dependency_1): Likewise.
10752         (ix86_adjust_cost): Likewise.
10753         (swap_top_of_ready_list): Fix formatting and !reload_completed check
10754         removed.
10755         (do_reorder_for_imul): Fix typo, formatting and
10756         !reload_completed check removed.
10757         (ix86_sched_reorder): Fix typo and formatting.
10758         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
10759         list.
10761 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
10763         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
10765 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
10767         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
10768         <STRING_CST>: Likewise.
10769         <VECTOR_CST>: Likewise.
10771 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
10772             Mikael Morin  <mikael@gcc.gnu.org>
10774         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
10775         * config.in: Regenerated.
10776         * configure: Regenerated.
10778 2013-06-01  Jan Hubicka  <jh@suse.cz>
10780         PR middle-end/57366
10781         * cgraphunit.c (compile): When weakref is not supported,
10782         set up transparent aliases before final output pass.
10783         * varasm.c (assemble_alias): Do not try to do it here.
10785 2013-06-01  Jan Hubicka  <jh@suse.cz>
10787         PR middle-end/57467
10788         * passes.c (for_per_function): Skip unanalyzed functions.
10790 2013-06-01  Jan Hubicka  <jh@suse.cz>
10792         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
10793         (lto_symtab_merge_symbols_1): ... this one.
10794         (lto_symtab_merge_cgraph_nodes): Rename to ...
10795         (lto_symtab_merge_symbols): ... this one; simplify.
10796         * cgraph.c (same_body_aliases_done): Rename to ...
10797         (cpp_implicit_aliases_done): ... this one.
10798         (cgraph_create_function_alias): Update.
10799         (cgraph_same_body_alias): Update.
10800         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
10801         (verify_edge_corresponds_to_fndecl): Simplify.
10802         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
10803         (cgraph_node): Remove same_body_alias.
10804         (varpool_node): Remove alias_of and extra_name_alias.
10805         (same_body_aliases_done): Rename to ..
10806         (cpp_implicit_aliases_done): ... this one.
10807         (symtab_alias_ultimate_target): Add default parameter.
10808         (symtab_resolve_alias): New function.
10809         (fixup_same_cpp_alias_visibility): Declare.
10810         (cgraph_function_node): Add default parameter.
10811         (cgraph_node_asm_name): Likewise.
10812         (cgraph_function_or_thunk_node): Add default parameter; do
10813         not ICE when it is NULL.
10814         (varpool_variable_node): Likewise.
10815         * tree-emutls.c (create_emultls_var): Update.
10816         (ipa_lower_emutls): Update.
10817         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
10818         (cgraph_reset_node): Reset alias info.
10819         (cgraph_finalize_function): Update.
10820         (fixup_same_cpp_alias_visibility): Move to symtab.c.
10821         (analyze_function): Simplify.
10822         (cgraph_process_same_body_aliases): Simplify.
10823         (analyze_functions): Fixup same body aliases.
10824         (handle_alias_pairs): Simplify.
10825         (assemble_thunk): Update.
10826         (assemble_thunks_and_aliases): Update.
10827         (output_weakrefs): Rewrite.
10828         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
10829         (lto_output_varpool_node): Likewise.
10830         (compute_ltrans_boundary): Remve assert.
10831         (get_alias_symbol): New functoin.
10832         (input_node): Rewrite alias handling.
10833         (input_varpool_node): Likewise.
10834         * ipa-pure-const.c (propagate_pure_const): Fix formating.
10835         * ipa.c (process_references): Handle weakrefs correctly.
10836         (symtab_remove_unreachable_nodes): Likewise.
10837         * trans-mem.c (get_cg_data): Update.
10838         (ipa_tm_create_version_alias): Update.
10839         (ipa_tm_execute): Update.
10840         * symtab.c (dump_symtab_base): Dump aliases.
10841         (verify_symtab_base): Verify aliases.
10842         (symtab_node_availability): New function.
10843         (symtab_alias_ultimate_target): Simplify.
10844         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
10845         handle all the fixup cases.
10846         (symtab_resolve_alias): New function.
10847         * passes.c (ipa_write_summaries): Handle weakrefs.
10848         * varpool.c (varpool_analyze_node): Simplify.
10849         (assemble_aliases): Update.
10850         (varpool_create_variable_alias): Simplify.
10851         (varpool_extra_name_alias): Simplify.
10852         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
10853         (lto_symtab_merge_symbols): ... this one.
10855 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
10857         Revert
10858         PR rtl-optimization/57268
10859         * sched-deps.c (sched_analyze_2): Flush dependence lists if
10860         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
10862 2013-06-01  Tobias Burnus  <burnus@net-b.de>
10864         Partially reverted:
10865         2013-05-31  Tobias Burnus  <burnus@net-b.de>
10867         PR middle-end/57073
10868         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
10869         further up.
10871 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
10873         PR rtl-optimization/57268
10874         * sched-deps.c (sched_analyze_2): Flush dependence lists if
10875         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
10877 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
10879         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
10880         unordered comparison operators when -fno-trapping-math is in effect
10881         on the e500.
10882         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
10883         and implement unordered comparison operators properly on the e500.
10885 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
10887         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
10888         for constant scalar integers.
10889         (simplify_relational_operation_1): Likewise.
10891 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
10893         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
10894         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
10895         Fix comment.
10897 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
10898             Igor Zamyatin  <igor.zamyatin@intel.com>
10900         Silvermont (SLM) architecture performance tuning.
10901         * config/i386/i386.h (enum ix86_tune_indices): Add
10902         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
10903         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
10905         * config/i386/i386.c (initial_ix86_tune_features)
10906         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
10907         (ix86_lea_outperforms): Handle Silvermont tuning.
10908         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
10909         call.
10910         (ix86_use_lea_for_mov): Likewise.
10911         (ix86_avoid_lea_for_addr): Likewise.
10912         (ix86_lea_for_add_ok): Likewise.
10913         (exact_dependency_1): New function.
10914         (exact_store_load_dependency): Likewise.
10915         (ix86_adjust_cost): Handle Silvermont tuning.
10916         (do_reoder_for_imul): Likewise.
10917         (swap_top_of_ready_list): New function.
10918         (ix86_sched_reorder): Changed to handle Silvermont tuning.
10920         * config/i386/i386.md (peepholes that split memory operand in fp
10921         converts): New.
10923 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
10925         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
10926         Remove un-necessary braces.
10928 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
10930         * config/aarch64/aarch64.c (aarch64_classify_symbol):
10931         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
10933 2013-05-31  Tobias Burnus  <burnus@net-b.de>
10935         PR middle-end/57073
10936         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
10938 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10940         PR target/56315
10941         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
10942         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
10943         * config/arm/neon.md (iordi3_neon): Remove.
10944         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
10945         * config/arm/predicates.md (imm_for_neon_logic_operand):
10946         Move to earlier in the file.
10947         (neon_logic_op2): Likewise.
10948         (arm_iordi_operand_neon): New predicate.
10950 2013-05-31  Richard Biener  <rguenther@suse.de>
10952         PR tree-optimization/57478
10953         PR tree-optimization/57453
10954         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
10955         are life as well.
10957 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
10959         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
10960         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
10962 2013-05-30  Tobias Burnus  <burnus@net-b.de>
10963             Thomas Koenig  <tkoenig@gcc.gnu.org>
10965         PR middle-end/57073
10966         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
10967         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
10969 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
10971         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
10973 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
10975         * target.def (register_usage_leveling_p): New hook.
10976         * targhooks.c (default_register_usage_leveling_p): New.
10977         * targhooks.h (default_register_usage_leveling_p): New prototype.
10978         * lra-assigns.c (register_usage_leveling_p): Use the hook.
10979         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
10980         * doc/tm.texi: Update.
10981         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
10983 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
10985         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
10986         (*insv_reg<mode>): New define_insn.
10988 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
10990         PR rtl-optimization/57439
10991         * postreload.c (move2add_valid_value_p): Check that we have
10992         a zero subreg_regno_offset when accessing the register in
10993         the requested mode.
10995 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
10996             Igor Zamyatin  <igor.zamyatin@intel.com>
10998         Silvermont (SLM) architecture pipeline model, tuning and
10999         insn selection.
11000         * config.gcc: Add slm config options and target.
11002         * config/i386/slm.md: New.
11004         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
11006         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
11007         PROCESSOR_SLM.
11008         (ix86_target_macros_internal): Likewise.
11010         * config/i386/i386.c (slm_cost): New cost.
11011         (m_SLM): New macro flag.
11012         (initial_ix86_tune_features): Set m_SLM.
11013         (x86_accumulate_outgoing_args): Likewise.
11014         (x86_arch_always_fancy_math_387): Likewise.
11015         (processor_target_table): Add slm cost.
11016         (cpu_names): Add slm cpu name.
11017         (x86_option_override_internal): Set SLM ISA.
11018         (ix86_issue_rate): New case PROCESSOR_SLM.
11019         (ia32_multipass_dfa_lookahead): Likewise.
11020         (fold_builtin_cpu): Add slm.
11022         * config/i386/i386.h (TARGET_SLM): New target macro.
11023         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
11024         (processor_type): Add PROCESSOR_SLM.
11026         * config/i386/i386.md (cpu): Add new value "slm".
11027         (slm.md): Include slm.md.
11029 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
11030             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
11032         * config/arm/arm-protos.h: Add and update function protos.
11033         * config/arm/arm.c (use_simple_return_p): New added.
11034         (thumb2_expand_return): Check simple_return flag.
11035         * config/arm/arm.md: Add simple_return and conditional simple_return.
11036         * config/arm/iterators.md: Add iterator for return and simple_return.
11038 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
11040         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
11041         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
11042         (arm_emit_vfp_multi_reg_pop): Likewise.
11043         (thumb2_emit_ldrd_pop): Likewise.
11044         (arm_expand_epilogue): Add misc REG_CFA notes.
11045         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
11047 2013-05-29  Lawrence Crowl  <crowl@google.com>
11049         * config/arm/t-arm: Update for below.
11051         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
11052         Change type to hash_table.  Update dependent calls and types.
11054         * config/i386/t-cygming: Update for below.
11056         * config/i386/t-interix: Update for below.
11058         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
11059         Change type to hash_table.  Update dependent calls and types.
11060         (i386_find_on_wrapper_list::wrappers): Likewise.
11062         * config/ia64/t-ia64: Update for below.
11064         * config/ia64/ia64.c (bundle_state_table):
11065         Change type to hash_table.  Update dependent calls and types.
11067         * config/mips/mips.c (mips_reorg_process_insns::htab):
11068         Change type to hash_table.  Update dependent calls and types.
11070         * config/sol2.c (solaris_comdat_htab):
11071         Change type to hash_table.  Update dependent calls and types.
11073         * config/t-sol2: Update for above.
11075 2013-05-29  Teresa Johnson  <tejohnson@google.com>
11077         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
11078         functions are not yet marked as defined.
11080 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
11081             Pat Haugen  <pthaugen@us.ibm.com>
11082             Peter Bergner  <bergner@vnet.ibm.com>
11084         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
11085         instructions.
11086         (VEC_A): Likewise.
11087         (VEC_C): Likewise.
11088         (vrotl<mode>3): Likewise.
11089         (vashl<mode>3): Likewise.
11090         (vlshr<mode>3): Likewise.
11091         (vashr<mode>3): Likewise.
11093         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11094         support for power8 V2DI builtins.
11096         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
11097         power8 V2DI builtins.
11098         (vupkhsw): Likewise.
11099         (vupklsw): Likewise.
11100         (vaddudm): Likewise.
11101         (vminsd): Likewise.
11102         (vmaxsd): Likewise.
11103         (vminud): Likewise.
11104         (vmaxud): Likewise.
11105         (vpkudum): Likewise.
11106         (vpksdss): Likewise.
11107         (vpkudus): Likewise.
11108         (vpksdus): Likewise.
11109         (vrld): Likewise.
11110         (vsld): Likewise.
11111         (vsrd): Likewise.
11112         (vsrad): Likewise.
11113         (vsubudm): Likewise.
11114         (vcmpequd): Likewise.
11115         (vcmpgtsd): Likewise.
11116         (vcmpgtud): Likewise.
11117         (vcmpequd_p): Likewise.
11118         (vcmpgtsd_p): Likewise.
11119         (vcmpgtud_p): Likewise.
11120         (vupkhsw): Likewise.
11121         (vupklsw): Likewise.
11122         (vaddudm): Likewise.
11123         (vmaxsd): Likewise.
11124         (vmaxud): Likewise.
11125         (vminsd): Likewise.
11126         (vminud): Likewise.
11127         (vpksdss): Likewise.
11128         (vpksdus): Likewise.
11129         (vpkudum): Likewise.
11130         (vpkudus): Likewise.
11131         (vrld): Likewise.
11132         (vsld): Likewise.
11133         (vsrad): Likewise.
11134         (vsrd): Likewise.
11135         (vsubudm): Likewise.
11137         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
11138         support for power8 V2DI instructions.
11140         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
11141         power8 V2DI instructions.  Combine pack and unpack insns to use an
11142         iterator for each mode.  Check whether a particular mode supports
11143         Altivec instructions instead of just checking TARGET_ALTIVEC.
11144         (UNSPEC_VPKUWUM): Likewise.
11145         (UNSPEC_VPKSHSS): Likewise.
11146         (UNSPEC_VPKSWSS): Likewise.
11147         (UNSPEC_VPKUHUS): Likewise.
11148         (UNSPEC_VPKSHUS): Likewise.
11149         (UNSPEC_VPKUWUS): Likewise.
11150         (UNSPEC_VPKSWUS): Likewise.
11151         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
11152         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
11153         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
11154         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
11155         (UNSPEC_VUPKHSB): Likewise.
11156         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
11157         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
11158         (UNSPEC_VUPKHSH): Likewise.
11159         (UNSPEC_VUPKLSB): Likewise.
11160         (UNSPEC_VUPKLSH): Likewise.
11161         (VI2): Likewise.
11162         (VI_char): Likewise.
11163         (VI_scalar): Likewise.
11164         (VI_unit): Likewise.
11165         (VP): Likewise.
11166         (VP_small): Likewise.
11167         (VP_small_lc): Likewise.
11168         (VU_char): Likewise.
11169         (add<mode>3): Likewise.
11170         (altivec_vaddcuw): Likewise.
11171         (altivec_vaddu<VI_char>s): Likewise.
11172         (altivec_vadds<VI_char>s): Likewise.
11173         (sub<mode>3): Likewise.
11174         (altivec_vsubcuw): Likewise.
11175         (altivec_vsubu<VI_char>s): Likewise.
11176         (altivec_vsubs<VI_char>s): Likewise.
11177         (altivec_vavgs<VI_char>): Likewise.
11178         (altivec_vcmpbfp): Likewise.
11179         (altivec_eq<mode>): Likewise.
11180         (altivec_gt<mode>): Likewise.
11181         (altivec_gtu<mode>): Likewise.
11182         (umax<mode>3): Likewise.
11183         (smax<mode>3): Likewise.
11184         (umin<mode>3): Likewise.
11185         (smin<mode>3): Likewise.
11186         (altivec_vpkuhum): Likewise.
11187         (altivec_vpkuwum): Likewise.
11188         (altivec_vpkshss): Likewise.
11189         (altivec_vpkswss): Likewise.
11190         (altivec_vpkuhus): Likewise.
11191         (altivec_vpkshus): Likewise.
11192         (altivec_vpkuwus): Likewise.
11193         (altivec_vpkswus): Likewise.
11194         (altivec_vpks<VI_char>ss): Likewise.
11195         (altivec_vpks<VI_char>us): Likewise.
11196         (altivec_vpku<VI_char>us): Likewise.
11197         (altivec_vpku<VI_char>um): Likewise.
11198         (altivec_vrl<VI_char>): Likewise.
11199         (altivec_vsl<VI_char>): Likewise.
11200         (altivec_vsr<VI_char>): Likewise.
11201         (altivec_vsra<VI_char>): Likewise.
11202         (altivec_vsldoi_<mode>): Likewise.
11203         (altivec_vupkhsb): Likewise.
11204         (altivec_vupkhs<VU_char>): Likewise.
11205         (altivec_vupkls<VU_char>): Likewise.
11206         (altivec_vupkhsh): Likewise.
11207         (altivec_vupklsb): Likewise.
11208         (altivec_vupklsh): Likewise.
11209         (altivec_vcmpequ<VI_char>_p): Likewise.
11210         (altivec_vcmpgts<VI_char>_p): Likewise.
11211         (altivec_vcmpgtu<VI_char>_p): Likewise.
11212         (abs<mode>2): Likewise.
11213         (vec_unpacks_hi_v16qi): Likewise.
11214         (vec_unpacks_hi_v8hi): Likewise.
11215         (vec_unpacks_lo_v16qi): Likewise.
11216         (vec_unpacks_hi_<VP_small_lc>): Likewise.
11217         (vec_unpacks_lo_v8hi): Likewise.
11218         (vec_unpacks_lo_<VP_small_lc>): Likewise.
11219         (vec_pack_trunc_v8h): Likewise.
11220         (vec_pack_trunc_v4si): Likewise.
11221         (vec_pack_trunc_<mode>): Likewise.
11223         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
11224         V2DI builtins.
11225         (vec_vmaxsd): Likewise.
11226         (vec_vmaxud): Likewise.
11227         (vec_vminsd): Likewise.
11228         (vec_vminud): Likewise.
11229         (vec_vpksdss): Likewise.
11230         (vec_vpksdus): Likewise.
11231         (vec_vpkudum): Likewise.
11232         (vec_vpkudus): Likewise.
11233         (vec_vrld): Likewise.
11234         (vec_vsld): Likewise.
11235         (vec_vsrad): Likewise.
11236         (vec_vsrd): Likewise.
11237         (vec_vsubudm): Likewise.
11238         (vec_vupkhsw): Likewise.
11239         (vec_vupklsw): Likewise.
11241 2013-05-29  Jan Hubicka  <jh@suse.cz>
11243         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
11244         flags; reorder rest of fields in more consistent way.
11245         (varpool_node): Remove analyzed, finalized and alias.
11246         (cgraph_ndoe): Likewise.
11247         (symtab_alias_ultimate_target): New function.
11248         (cgraph_function_node): Move offline.
11249         (cgraph_reset_node): Declare.
11250         (cgraph_comdat_can_be_unshared_p): Remove.
11251         (varpool_remove_initializer): Declare.
11252         (varpool_first_defined_variable, varpool_next_defined_variable
11253         cgraph_first_defined_function, cgraph_next_defined_function): Update.
11254         (cgraph_function_with_gimple_body_p): Update.
11255         (varpool_all_refs_explicit_p): Update.
11256         (symtab_alias_target): New function.
11257         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
11258         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
11259         (cgraph_function_or_thunk_node): Simplify using
11260         symtab_alias_ultimate_target.
11261         (varpool_variable_node): Likewise.
11262         * cgraph.c (cgraph_create_function_alias): Update.
11263         (cgraph_add_thunk): Update.
11264         (cgraph_remove_node): Update.
11265         (dump_cgraph_node): Do not dump removed flags.
11266         (cgraph_function_body_availability): Update.
11267         (cgraph_propagate_frequency): Update.
11268         (verify_cgraph_node): Check sanity of local flag.
11269         (cgraph_function_node): Move here from cgraph.h; revamp for
11270         cgraph_function_or_thunk_node.
11271         * lto-symtab.c (lto_varpool_replace_node): Update.
11272         (lto_symtab_resolve_can_prevail_p): Update.
11273         (lto_symtab_merge_cgraph_nodes): Update.
11274         * ipa-cp.c (determine_versionability, initialize_node_lattices,
11275         propagate_constants_accross_call, devirtualization_time_bonus,
11276         ipcp_propagate_stage): Update.
11277         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
11278         * ipa-inline-transform.c (clone_inlined_nodes,
11279         preserve_function_body_p): Update.
11280         * ipa-reference.c (propagate): Update.
11281         (write_node_summary_p): Update.
11282         * toplev.c (wrapup_global_declaration_2): Update.
11283         * cgraphunit.c (cgraph_analyze_function): Rename to ...
11284         (analyze_function) ... this one.
11285         (cgraph_process_new_functions): Update.
11286         (cgraph_reset_node): Export.
11287         (cgraph_finalize_function): Update.
11288         (cgraph_add_new_function): Update.
11289         (process_function_and_variable_attributes): Update.
11290         (varpool_finalize_decl): Update.
11291         (symbol_finalized): Remove.
11292         (symbol_finalized_and_needed): Rename to ...
11293         (symbol_defined_and_needed): ... update.
11294         (cgraph_analyze_functions): Update.
11295         (handle_alias_pairs): Update.
11296         (mark_functions_to_output): Update.
11297         (assemble_thunk): Update.
11298         (output_in_order): Update.
11299         (output_weakrefs): Update.
11300         (finalize_compilation_unit): Update.
11301         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
11302         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
11303         input_node, input_varpool_node): Update.
11304         * dbxout.c (dbxout_expand_expr): Update.
11305         * cgraphclones.c (cgraph_clone_node): Update.
11306         (cgraph_copy_node_for_versioning): Update.
11307         (cgraph_materialize_clone): Update.
11308         (cgraph_materialize_all_clones): Update.
11309         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
11310         propagate_pure_const, propagate_nothrow): Update.
11311         * lto-streamer-out.c (lto_output, write_symbol): Update.
11312         * ipa-utils.c (ipa_reverse_postorder): Update.
11313         * ipa-inline.c (can_inline_edge_p): Update.
11314         (update_caller_keys, ipa_inline): Update.
11315         * dwarf2out.c (reference_to_unused,
11316         premark_types_used_by_global_vars_helper): Update.
11317         * tree-eh.c (tree_could_trap_p): Update.
11318         * ipa-split.c (consider_split, execute_split_functions): Update.
11319         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
11320         has_addr_references_p): Update; move ahead in file for better
11321         readability.
11322         (process_references): Simplify.
11323         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
11324         bodies are removed.
11325         (cgraph_comdat_can_be_unshared_p): Make static.
11326         (cgraph_externally_visible_p): Update.
11327         (varpool_externally_visible_p): Update.
11328         (function_and_variable_visibility): Update.
11329         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
11330         ipa_tm_mark_force_output_node): Update.
11331         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
11332         estimate_edge_devirt_benefit, inline_generate_summary,
11333         inline_write_summary): Update.
11334         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
11335         * ipa-prop.c (ipa_compute_jump_functions): Update.
11336         (ipa_print_node_params, ipa_prop_read_section,
11337         ipa_update_after_lto_read, read_replacements_section): Update.
11338         * varasm.c (mark_decl_referenced): Update.
11339         (assemble_alias, dump_tm_clone_pairs): Update.
11340         * tree-inline.c (copy_bb): Update.
11341         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
11342         Update.
11343         * symtab.c (dump_symtab_base): Print new flags.
11344         (verify_symtab_base): Verify new flags.
11345         (symtab_alias_ultimate_target): New function.
11346         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
11347         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
11348         Update.
11349         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
11350         Update.
11351         * i386.c (ix86_get_function_versions_dispatcher,
11352         ix86_generate_version_dispatcher_body): Update.
11353         (fold_builtin_cpu): Use varpool_add_new_variable.
11354         * varpool.c (varpool_remove_initializer): Break out from ...
11355         (varpool_remove_node): ... this one.
11356         (dump_varpool_node, varpool_node_for_asm,
11357         cgraph_variable_initializer_availability, varpool_analyze_node,
11358         varpool_assemble_decl, varpool_remove_unreferenced_decls,
11359         varpool_finalize_named_section_flags, varpool_create_variable_alias):
11360         Update.
11362 2013-05-29  Jan Hubicka  <jh@suse.cz>
11364         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
11366 2013-05-29  Easwaran Raman  <eraman@google.com>
11368         PR tree-optimization/57442
11369         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
11370         when control exits the main loop.
11372 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
11374         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
11375         and RX600.
11376         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
11377         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
11378         * rx/t-rx: Add rx100 under multi library matches option for nofpu
11379         option.
11381 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11383         PR tree-optimization/57441
11384         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
11385         Don't limit size of incr_vec to number of candidates.
11387 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
11389         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
11390         and mips16 directories.
11391         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
11392         (MULTILIB_DIRNAMES): Ditto.
11393         (MULTILIB_EXCEPTIONS): Add new exceptions.
11394         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
11395         (MULTILIB_DIRNAMES): Ditto.
11396         (MULTILIB_EXCEPTIONS): Add new exceptions.
11398 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
11399             Marcus Shawcroft  <marcus.shawcroft@arm.com>
11401         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
11402         SYMBOL_TINY_ABSOLUTE.
11403         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
11404         SYMBOL_TINY_ABSOLUTE.
11405         (aarch64_expand_mov_immediate): Likewise.
11406         (aarch64_classify_symbol): Likewise.
11407         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
11408         Permit SYMBOL_TINY_ABSOLUTE.
11409         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
11411 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
11412             Marcus Shawcroft  <marcus.shawcroft@arm.com>
11414         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
11415         Refactor if/switch.  Replace gcc_assert with if.
11417 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
11419         * config/i386/i386.c (initial_ix86_tune_features): Enable
11420         FP Reassociation for AMD bdver1 and bdver2.
11422 2013-05-29  Martin Jambor  <mjambor@suse.cz>
11424         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
11425         and IMAGPART_EXPR do not occur within other handled_components.
11427 2013-05-29  Richard Biener  <rguenther@suse.de>
11429         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
11430         access on whether the use is in the BB we currently try to
11431         vectorize.
11432         (vect_bb_vectorization_profitable_p): Pass the BB we currently
11433         vectorize to vect_bb_slp_scalar_cost.
11435 2013-05-29  Richard Biener  <rguenther@suse.de>
11437         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
11438         computing scalar cost offsetted by stmts that are kept live
11439         by scalar uses.
11440         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
11441         for computation of scalar cost.
11443 2013-05-28  Steve Ellcey  <sellcey@mips.com>
11445         * config/mips/mips-cpus.def (mips32r2): Change processor type.
11447 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
11449         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
11450         array notation built-in reduction functions.
11451         * doc/passes.texi (Passes): Added documentation about changes done
11452         for Cilk Plus.
11453         * doc/invoke.texi (C Dialect Options): Added documentation about
11454         the -fcilkplus flag.
11455         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
11456         (BUILTINS_DEF): Depend on cilkplus.def.
11457         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
11458         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
11459         * cilkplus.def: New file.
11461 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
11463         PR rtl-optimization/57439
11464         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
11466 2013-05-28  Easwaran Raman  <eraman@google.com>
11468         PR tree-optimization/57337
11469         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
11470         (find_insert_point): Correctly identify the insertion point
11471         when two statements with the same UID is compared.
11473 2013-05-28  Richard Biener  <rguenther@suse.de>
11475         PR tree-optimization/56787
11476         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
11477         from the list of data references.
11478         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
11479         clobbers.
11480         (vect_analyze_loop_operations): Likewise.
11481         (vect_transform_loop): Remove clobbers.
11483 2013-05-28  Martin Jambor  <mjambor@suse.cz>
11485         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
11486         and REALPART_EXPRs have scalar type.
11488 2013-05-28  Richard Biener  <rguenther@suse.de>
11490         PR tree-optimization/57411
11491         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
11492         virtual operands.
11493         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
11494         virtual operand propagation.
11496 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
11498         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
11499         destination register for bmasksi_vis.
11500         (vector_init_bshuffle): Likewise.
11501         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
11503 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
11505         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
11506         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
11507         mode if the instruction isn't available in the original mode.
11508         * config/sparc/sparc.opt (mfix-ut699): New option.
11509         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
11510         (divdf3): Turn into expander.
11511         (divdf3_nofix): New insn.
11512         (divdf3_fix): Likewise.
11513         (divsf3): Disable if -mfix-ut699.
11514         (sqrtdf2): Turn into expander.
11515         (sqrtdf2_nofix): New insn.
11516         (sqrtdf2_fix): Likewise.
11517         (sqrtsf2): Disable if -mfix-ut699.
11519 2013-05-27  Richard Biener  <rguenther@suse.de>
11521         PR middle-end/57412
11522         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
11523         block for the new loop.
11525 2013-05-27  Richard Biener  <rguenther@suse.de>
11527         PR tree-optimization/57343
11528         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
11529         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
11530         (number_of_iterations_cond): Do not build the folded tree.
11532 2013-05-27  Richard Biener  <rguenther@suse.de>
11534         Revert
11535         PR middle-end/57381
11536         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
11537         OEP_CONSTANT_ADDRESS_OF retained.
11539         PR tree-optimization/57417
11540         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
11541         for unchanged base.
11542         (set_ssa_val_to): Compare addresses using
11543         get_addr_base_and_unit_offset.
11545 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
11547         PR rtl-optimization/56833
11548         * postreload.c (move2add_record_mode): New function.
11549         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
11550         (move2add_use_add2_insn): Use move2add_record_sym_value.
11551         (move2add_use_add3_insn): Likewise.
11552         (reload_cse_move2add): Use move2add_valid_value_p and
11553         move2add_record_mode.  Invalidate call-clobbered and REG_INC
11554         affected regs by setting reg_mode to VOIDmode.
11555         (move2add_note_store): Don't pretend the inside of a SUBREG is
11556         the actual destination.  Invalidate single/leading registers by
11557         setting reg_mode to VOIDmode.
11558         Use move2add_record_sym_value, move2add_valid_value_p and
11559         move2add_record_mode.
11561 2013-05-27  Richard Biener  <rguenther@suse.de>
11563         PR tree-optimization/57396
11564         * tree-affine.c (double_int_constant_multiple_p): Properly
11565         return false for val == 0 and div != 0.
11567 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
11569         * config/mips/mips.h: Use #elif in preprocessor conditions.
11571 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
11573         PR target/53916
11574         * config/mips/constraints.md (kl): New constraint.
11575         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
11576         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
11577         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
11578         from LO for MIPS16.
11579         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
11581 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
11583         PR target/55777
11584         * config/mips/mips.c (mips_can_inline_p): New function.
11585         (TARGET_CAN_INLINE_P): Define.
11587 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
11589         * sched-int.h (ds_t, dw_t): Make unsigned int.
11590         Fix documentation that describes how all the ds_t bits are used.
11591         Reserve the last bit for delayed-branch scheduling.
11592         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
11593         (BITS_PER_DEP_WEAK): Fix definition and documentation.
11594         (gen_dep_weak_1): Remove prototype.
11595         * sched-deps.c (get_dep_weak_1): Make static.
11596         * target.def (speculate_insn, needs_block_p, gen_spec_check,
11597         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
11598         * doc/tm.texi: Regenerate.
11599         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
11601 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
11603         PR debug/56950
11604         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
11606 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
11607             Sandra Loosemore  <sandra@codesourcery.com>
11609         * config.gcc (powerpc-*): Allow native for with-cpu.
11611 2013-05-24  Jeff Law  <law@redhat.com>
11613         PR tree-optimization/57124
11614         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
11615         conversion feeding a condition if the range has an overflow
11616         if -fstrict-overflow.  Add warnings for when we do make the
11617         transformation.
11619 2013-05-24  Dehao Chen  <dehao@google.com>
11621         * tree-cfg.c (locus_discrim_map): Fix the typo.
11622         (locus_discrim_hasher): Likewise.
11623         (locus_discrim_hasher::hash): Likewise.
11624         (locus_discrim_hasher::equal): Likewise.
11626 2013-05-24  Martin Jambor  <mjambor@suse.cz>
11628         PR tree-optimization/57294
11629         * cgraph.h (ipa_record_stmt_references): Declare.
11630         * cgraphbuild.c (ipa_record_stmt_references): New function.
11631         (build_cgraph_edges): Use ipa_record_stmt_references.
11632         (rebuild_cgraph_edges): Likewise.
11633         (cgraph_rebuild_references): Likewise.
11634         * ipa-prop.c (ipa_modify_call_arguments): Discard references
11635         associated with the old statement and build references from the
11636         newly built statements.
11637         * ipa-ref.c (ipa_remove_stmt_references): New function.
11638         * ipa-ref.h (ipa_remove_stmt_references): Declare.
11640 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
11642         * lra-constraints.c (emit_spill_move): Use smaller mode for
11643         mem-mem moves.
11644         (check_and_process_move): Consider mem-reg moves for secondary
11645         too.
11646         (curr_insn_transform): Don't lose insns emitted before for
11647         secondary memory moves.
11648         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
11649         reg set up in the current insn.
11651 2013-05-24  Dehao Chen  <dehao@google.com>
11653         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
11654         hash function.
11655         (locus_descrim_hasher::equal): Likewise.
11656         (build_gimple_cfg): New discriminator assignment algorithm.
11657         (make_edges): Likewise.
11658         (next_discriminator_for_locus): Likewise.
11659         (same_line_p): Likewise.
11660         (assign_discriminators): Likewise.
11661         (make_cond_expr_edges): Likewise.
11662         (make_gimple_switch_edges): Likewise.
11663         (make_goto_expr_edges): Likewise.
11664         (make_gimple_asm_edges): Likewise.
11666 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
11668         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
11669         X format specifier to only display bottom 16 bits.
11670         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
11671         immediate to match for operand 2, since it will be masked.
11673 2013-05-24  Richard Biener  <rguenther@suse.de>
11675         PR tree-optimization/57287
11676         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
11677         all SSA names that occur in abnormal PHIs.
11679 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
11681         PR tree-ssa/57385
11682         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
11683         that index is not negative.
11685 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
11687         PR rtl-optimization/55177
11688         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
11689         (simplify_byte_swapping_operation): New.
11690         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
11691         (simplify_relational_operation_1): Deal with BSWAP.
11693 2013-05-23  Richard Henderson  <rth@redhat.com>
11695         PR target/56742
11696         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
11697         (ix86_reorg): Call it.
11699 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
11701         PR target/57379
11702         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
11703         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
11704         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
11706 2013-05-23  Christian Bruel  <christian.bruel@st.com>
11708         PR debug/57351
11709         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
11711 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
11712             Marcus Shawcroft  <marcus.shawcroft@arm.com>
11714         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
11715         * config/aarch64/constraints.md (Usa): Remove.
11716         * doc/md.texi (AArch64 Usa): Remove.
11718 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
11719             Marcus Shawcroft  <marcus.shawcroft@arm.com>
11721         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
11722         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
11723         * config/aarch64/predicates.md (aarch64_const_address): Remove.
11724         (aarch64_mov_operand): Use aarch64_mov_operand_p.
11726 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
11728         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
11729         instruction (AdvSIMD).
11730         * config/aarch64/aarch64-builtins.c
11731         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
11732         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
11734 2013-05-23  Martin Jambor  <mjambor@suse.cz>
11736         PR middle-end/57347
11737         * tree.h (contains_bitfld_component_ref_p): Declare.
11738         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
11739         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
11740         caller.
11741         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
11742         not access a bit-field.  Assert all final offsets are byte-aligned.
11744 2013-05-23  Richard Biener  <rguenther@suse.de>
11746         PR tree-optimization/57380
11747         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
11748         least one invariant or re-used load.
11749         * passes.c (init_optimization_passes): Move pass_phiprop before
11750         pass_forwprop.
11752 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
11754         * config/aarch64/aarch64-simd.md
11755         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
11757 2013-05-23  Richard Biener  <rguenther@suse.de>
11759         PR middle-end/57381
11760         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
11761         OEP_CONSTANT_ADDRESS_OF retained.
11763 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
11765         PR middle-end/57344
11766         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
11767         don't lower unit.  Handle unit not being always BITS_PER_WORD.
11769 2013-05-23  Richard Biener  <rguenther@suse.de>
11771         PR rtl-optimization/57341
11772         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
11773         instead of true_dependence.
11775 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
11777         * bb-reorder.c (branch_threshold): Make const.
11778         (exec_threshold): Ditto.
11780 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
11781             Pat Haugen  <pthaugen@us.ibm.com>
11782             Peter Bergner  <bergner@vnet.ibm.com>
11784         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
11785         documentation for the power8 crypto builtins.
11787         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
11789         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
11790         macros for defining power8 builtin functions.
11791         (BU_P8V_AV_2): Likewise.
11792         (BU_P8V_AV_P): Likewise.
11793         (BU_P8V_VSX_1): Likewise.
11794         (BU_P8V_OVERLOAD_1): Likewise.
11795         (BU_P8V_OVERLOAD_2): Likewise.
11796         (BU_CRYPTO_1): Likewise.
11797         (BU_CRYPTO_2): Likewise.
11798         (BU_CRYPTO_3): Likewise.
11799         (BU_CRYPTO_OVERLOAD_1): Likewise.
11800         (BU_CRYPTO_OVERLOAD_2): Likewise.
11801         (XSCVSPDP): Fix typo, point to the correct instruction.
11802         (VCIPHER): Add power8 crypto builtins.
11803         (VCIPHERLAST): Likewise.
11804         (VNCIPHER): Likewise.
11805         (VNCIPHERLAST): Likewise.
11806         (VPMSUMB): Likewise.
11807         (VPMSUMH): Likewise.
11808         (VPMSUMW): Likewise.
11809         (VPERMXOR_V2DI): Likewise.
11810         (VPERMXOR_V4SI: Likewise.
11811         (VPERMXOR_V8HI: Likewise.
11812         (VPERMXOR_V16QI: Likewise.
11813         (VSHASIGMAW): Likewise.
11814         (VSHASIGMAD): Likewise.
11815         (VPMSUM): Likewise.
11816         (VPERMXOR): Likewise.
11817         (VSHASIGMA): Likewise.
11819         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
11820         __CRYPTO__ if the crypto instructions are available.
11821         (altivec_overloaded_builtins): Add support for overloaded power8
11822         builtins.
11824         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
11825         support for power8 crypto builtins.
11826         (builtin_function_type): Likewise.
11827         (altivec_init_builtins): Add support for builtins that take vector
11828         long long (V2DI) arguments.
11830         * config/rs6000/crypto.md: New file, define power8 crypto
11831         instructions.
11833 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
11834             Pat Haugen  <pthaugen@us.ibm.com>
11835             Peter Bergner  <bergner@vnet.ibm.com>
11837         * doc/invoke.texi (Option Summary): Add power8 options.
11838         (RS/6000 and PowerPC Options): Likewise.
11840         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
11841         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
11842         wm, wn, wr documentation.
11844         * config/rs6000/constraints.md (wm): New constraint for VSX
11845         registers if direct move instructions are enabled.
11846         (wn): New constraint for no registers.
11847         (wq): New constraint for quad word even GPR registers.
11848         (wr): New constraint if 64-bit instructions are enabled.
11849         (wv): New constraint if power8 vector instructions are enabled.
11850         (wQ): New constraint for quad word memory locations.
11852         * config/rs6000/predicates.md (const_0_to_15_operand): New
11853         constraint for 0..15 for crypto instructions.
11854         (gpc_reg_operand): If VSX allow registers in VSX registers as well
11855         as GPR and floating point registers.
11856         (int_reg_operand): New predicate to match only GPR registers.
11857         (base_reg_operand): New predicate to match base registers.
11858         (quad_int_reg_operand): New predicate to match even GPR registers
11859         for quad memory operations.
11860         (vsx_reg_or_cint_operand): New predicate to allow vector logical
11861         operations in both GPR and VSX registers.
11862         (quad_memory_operand): New predicate for quad memory operations.
11863         (reg_or_indexed_operand): New predicate for direct move support.
11865         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
11866         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
11867         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
11868         (POWERPC_MASKS): Add power8 options.
11869         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
11870         various options.
11872         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
11873         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
11875         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
11876         (-mpower8-fusion): New power8 options.
11877         (-mpower8-fusion-sign): Likewise.
11878         (-mpower8-vector): Likewise.
11879         (-mcrypto): Likewise.
11880         (-mdirect-move): Likewise.
11881         (-mquad-memory): Likewise.
11883         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
11884         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
11885         registers.
11886         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
11887         (rs6000_debug_vector_unit): Add p8_vector.
11888         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
11889         definitions.  Also print fusion state.
11890         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
11891         (rs6000_builtin_mask_calculate): Add power8 builtin support.
11892         (rs6000_option_override_internal): Add support for power8.
11893         (rs6000_common_init_builtins): Add debugging for skipped builtins
11894         if -mdebug=builtin.
11895         (rs6000_adjust_cost): Add power8 support.
11896         (rs6000_issue_rate): Likewise.
11897         (insn_must_be_first_in_group): Likewise.
11898         (insn_must_be_last_in_group): Likewise.
11899         (force_new_group): Likewise.
11900         (rs6000_register_move_cost): Likewise.
11901         (rs6000_opt_masks): Likewise.
11903         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
11904         power8 capable assembler, default to power7 options.
11905         (TARGET_DIRECT_MOVE): Likewise.
11906         (TARGET_CRYPTO): Likewise.
11907         (TARGET_P8_VECTOR): Likewise.
11908         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
11909         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
11910         (VECTOR_MEM_P8_VECTOR_P): Likewise.
11911         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
11912         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
11913         (TARGET_XSCVDPSPN): Likewise.
11914         (TARGET_XSCVSPDPN): Likewsie.
11915         (TARGET_SYNC_HI_QI): Likewise.
11916         (TARGET_SYNC_TI): Likewise.
11917         (MASK_CRYPTO): Likewise.
11918         (MASK_DIRECT_MOVE): Likewise.
11919         (MASK_P8_FUSION): Likewise.
11920         (MASK_P8_VECTOR): Likewise.
11921         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
11922         temporary used by some of the direct move instructions to get two FP
11923         temporary registers does not force creation of a stack frame.
11924         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
11925         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
11926         that any VSX registers are tieable, even if they are also an
11927         Altivec vector mode.
11928         (r6000_reg_class_enum): Add wm, wr, wv constraints.
11929         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
11930         (RS6000_BTM_CRYPTO): Likewise.
11931         (RS6000_BTM_COMMON): Likewise.
11933         * config/rs6000/rs6000.md (cpu attribute): Add power8.
11934         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
11935         (enum rs6000_vector): Add power8 vector support.
11937 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11939         PR target/19599
11940         PR target/57340
11941         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
11942         (any_sibcall_could_use_r3): this and handle indirect calls.
11943         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
11945 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11947         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
11949 2013-05-22  Richard Biener  <rguenther@suse.de>
11951         PR middle-end/57349
11952         * profile.c (branch_prob): Do not split blocks that are
11953         abnormally receiving from ECF_RETURNS_TWICE functions.
11955 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
11957         * recog.c (offsettable_address_addr_space_p): Fix calculation of
11958         address mode.  Move pointer mode initialization to the same place.
11960 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
11962         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
11963         while it has any effect.
11965 2013-05-21  Easwaran Raman  <eraman@google.com>
11967         PR tree-optimization/57322
11968         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
11969         UID of the statement added to the BB to be 1.
11971 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
11973         PR tree-optimization/57331
11974         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
11975         of conversion from pointer type to integral type with integer.
11977 2013-05-21  Martin Jambor  <mjambor@suse.cz>
11979         PR lto/57289
11980         * ipa-prop.c (ipa_read_node_info): Process param_used and
11981         controlled_uses in the same order as when writing.
11983 2013-05-21  Magnus Granberg  <baldrick@free.fr>
11985         PR plugins/56754
11986         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
11988 2013-05-21  Richard Biener  <rguenther@suse.de>
11990         PR tree-optimization/57318
11991         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
11992         estimate stmts with side-effects as likely eliminated.
11994 2013-05-21  Richard Biener  <rguenther@suse.de>
11996         PR tree-optimization/57330
11997         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
11998         preserve the call stmts fntype.
12000 2013-05-21  Richard Biener  <rguenther@suse.de>
12002         PR tree-optimization/57303
12003         * tree-ssa-sink.c (statement_sink_location): Improve killing
12004         stmt detection and properly handle self-assignments.
12006 2013-05-21  Christian Bruel  <christian.bruel@st.com>
12008         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
12009         spanning registers. LEAF_REG_REMAP is supported only for contiguous
12010         registers. Set register size out of the PARALLEL loop.
12012 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
12014         PR target/56547
12015         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
12016         (*fmasf4, *fmasf4_media): New insns.
12018 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
12020         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
12021         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
12022         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
12023         (mips_idiv_insns): Update the comments to say that the returned
12024         instruction counts are in units of BASE_INSN_LENGTH.
12025         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
12026         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
12027         using 2 rather than 4 as the length of indirect MIPS16 and
12028         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
12029         length of a NOP.  Don't divide MIPS16 lengths by 2.
12030         (mips16_split_long_branches): Assume a branch is long if the
12031         length is greater than 4 rather than 8.
12032         * config/mips/mips.md (length): Give MIPS16 lengths directly,
12033         rather than multiplying them by 2.  Multiply instruction counts
12034         by BASE_INSN_LENGTH rather than 4.
12035         (*jump_mips16, tls_get_tp_mips16_<mode>)
12036         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
12038 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
12040         * config/mips/mips.md (extended_mips16): Remove branch case.
12041         (length): Remove duplicated extended_mips16 test.
12043 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
12045         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
12047 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
12049         * recog.h (Recog_data): Rename to...
12050         (recog_data_d): ...this.
12051         (recog_data): Update accordingly.
12052         * recog.c (recog_data): Likewise.
12053         * reload.c (save_recog_data): Likewise.
12054         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
12055         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
12057 2013-05-17  Julian Brown  <julian@codesourcery.com>
12059         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
12060         found in a REG_EQUAL note, invalidate it.
12062 2013-05-17  Easwaran Raman  <eraman@google.com>
12064         * tree-ssa-reassoc.c (find_insert_point): New function.
12065         (insert_stmt_after): Likewise.
12066         (get_def_stmt): Likewise.
12067         (ensure_ops_are_available): Likewise.
12068         (not_dominated_by): Likewise.
12069         (rewrite_expr_tree): Do not move statements beyond what is
12070         necessary. Remove call to swap_ops_for_binary_stmt...
12071         (reassociate_bb): ... and move it here.
12072         (build_and_add_sum): Assign UIDs for new statements.
12073         (linearize_expr): Likewise.
12074         (do_reassoc): Renumber gimple statement UIDs.
12076 2013-05-17  Jan Hubicka  <jh@suse.cz>
12078         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
12079         weakrefs.
12080         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
12081         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
12082         weakrefs.
12083         (output_weakrefs): Update.
12085 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
12086             Martin Jambor  <mjambor@suse.cz>
12088         PR middle-end/57276
12089         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
12090         value that corresponds to the given aggval is found in values vector.
12092 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
12094         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
12095         sse, sse2, sse3, ssse3 and sse4a flags to options.
12097 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
12099         * gengtype-state.c: (s_expr_writer): New class, to handle
12100         prettifying of output layout of s-expressions.
12101         (state_writer): New class, to write out gtype.state.
12102         (state_written_type_count): Move this variable into member data of
12103         state_writer.
12104         (s_expr_writer::s_expr_writer): New code: constructor for new class
12105         (state_writer::state_writer(): ditto
12106         (s_expr_writer::write_new_line): New function
12107         (s_expr_writer::write_any_indent): ditto
12108         (s_expr_writer::begin_s_expr): ditto
12109         (s_expr_writer::end_s_expr): ditto
12110         (write_state_fileloc): convert to method of state_writer...
12111         (state_writer:: write_state_fileloc): ...and use methods of
12112         s_expr_writer to write indentation into the gtype.state output file
12113         to visually represent the hierarchical structure of the list
12114         structures
12115         (write_state_fields): ditto, renaming to...
12116         (state_writer::write_state_fields)
12117         (write_state_a_string): ditto, renaming to...
12118         (state_writer::write_state_a_string)
12119         (write_state_string_option): ditto, renaming to...
12120         (state_writer::write_state_string_option)
12121         (write_state_type_option): ditto, renaming to...
12122         (state_writer::write_state_type_option)
12123         (write_state_nested_option): ditto, renaming to...
12124         (state_writer::write_state_nested_option)
12125         (write_state_option): ditto, renaming to...
12126         (state_writer::write_state_option)
12127         (write_state_options): ditto, renaming to...
12128         (state_writer::write_state_options)
12129         (write_state_lang_bitmap): ditto, renaming to...
12130         (state_writer::write_state_lang_bitmap)
12131         (write_state_version): ditto, renaming to...
12132         (state_writer::write_state_version)
12133         (write_state_scalar_type): ditto, renaming to...
12134         (state_writer::write_state_scalar_type)
12135         (write_state_string_type): ditto, renaming to...
12136         (state_writer::write_state_string_type)
12137         (write_state_undefined_type): ditto, renaming to...
12138         (state_writer::write_state_undefined_type)
12139         (write_state_struct_union_type): ditto, renaming to...
12140         (state_writer::write_state_struct_union_type)
12141         (write_state_struct_type): ditto, renaming to...
12142         (state_writer::write_state_struct_type)
12143         (write_state_user_struct_type): ditto, renaming to...
12144         (state_writer::write_state_user_struct_type)
12145         (write_state_lang_struct_type): ditto, renaming to...
12146         (state_writer::write_state_lang_struct_type)
12147         (write_state_param_struct_type): ditto, renaming to...
12148         (state_writer::write_state_param_struct_type)
12149         (write_state_pointer_type): ditto, renaming to...
12150         (state_writer::write_state_pointer_type)
12151         (write_state_array_type): ditto, renaming to...
12152         (state_writer::write_state_array_type)
12153         (write_state_gc_used): ditto, renaming to...
12154         (state_writer::write_state_gc_used)
12155         (write_state_common_type_content): ditto, renaming to...
12156         (state_writer::write_state_common_type_content)
12157         (write_state_type): ditto, renaming to...
12158         (state_writer::write_state_type)
12159         (write_state_pair_list): ditto, renaming to...
12160         (state_writer::write_state_pair_list)
12161         (write_state_pair): ditto, renaming to...
12162         (state_writer::write_state_pair)
12163         (write_state_typedefs): ditto, renaming to...
12164         (state_writer::write_state_typedefs)
12165         (write_state_structures): ditto, renaming to...
12166         (state_writer::write_state_structures)
12167         (write_state_param_structs): ditto, renaming to...
12168         (state_writer::write_state_param_structs)
12169         (write_state_variables): ditto, renaming to...
12170         (state_writer::write_state_variables)
12171         (write_state_srcdir): ditto, renaming to...
12172         (state_writer::write_state_srcdir)
12173         (write_state_files_list): ditto, renaming to...
12174         (state_writer::write_state_files_list)
12175         (write_state_languages): ditto, renaming to...
12176         (state_writer::write_state_languages)
12177         (write_state): create a state_writer instance and use it when
12178         writing out the state file
12180 2013-05-17  Mike Stump  <mikestump@comcast.net>
12182         PR rtl-optimization/57304
12183         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
12184         accessing DF_REF_REAL_LOC.
12186 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
12188         PR rtl-optimization/57281
12189         PR rtl-optimization/57300
12190         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
12191         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
12192         what the other splitter did if the registers are dead.
12194 2013-05-17  Richard Biener  <rguenther@suse.de>
12196         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
12197         MEM_REF offsets.
12199 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
12201         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
12202         linking.
12204 2013-05-17  Marek Polacek  <polacek@redhat.com>
12206         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
12207         length when doing non-zero store of storing '\0' to '\0'.
12209 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
12211         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
12212         vect_external_def oprnd1 with loop_vinfo, try to emit
12213         optional cast, negation and and stmts on the loop preheader
12214         edge instead of into the pattern def seq.
12216         PR tree-optimization/57051
12217         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
12218         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
12220 2013-05-16  Nick Clifton  <nickc@redhat.com>
12222         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
12223         (rl78_is_naked_func): New function.
12224         (rl78_expand_prologue): Skip prologue generation for naked functions.
12225         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
12226         * doc/extend.texi (naked): Add RL78 to the list of processors
12227         that supports this attribute.
12229 2013-05-16  Jeff Law  <law@redhat.com>
12231         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
12233 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
12235         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
12236         cache parameters using detect_caches_amd also for CYRIX,
12237         NSC and TM2 signatures.
12239 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
12240             Dzianis Kahanovich  <mahatma@eu.by>
12242         PR target/45359
12243         PR target/46396
12244         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
12245         VIA/Centaur processors and determine their cache parameters
12246         using detect_caches_amd.
12248 2013-05-16  Teresa Johnson  <tejohnson@google.com>
12250         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
12251         (rtl_verify_edges): New function.
12252         (rtl_verify_bb_insns): Ditto.
12253         (rtl_verify_bb_pointers): Ditto.
12254         (rtl_verify_bb_insn_chain): Ditto.
12255         (rtl_verify_fallthru): Ditto.
12256         (rtl_verify_bb_layout): Ditto.
12257         (rtl_verify_flow_info_1): Outline checks into new functions.
12258         (rtl_verify_flow_info): Ditto.
12260 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
12262         * cfghooks.c (copy_bbs): Add update_dominance argument.
12263         * cfghooks.h (copy_bbs): Update prototype.
12264         * tree-cfg.c (gimple_duplicate_sese_region):
12265         Add update_dominance argument.
12266         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
12267         * tree-ssa-loop-ch.c (copy_loop_headers): Update
12268         gimple_duplicate_sese_region call.
12269         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
12270         Update copy_bbs call.
12271         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
12272         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
12274 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
12276         * tree-vectorizer.h (NUM_PATTERNS): Increment.
12277         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
12278         vect_recog_rotate_pattern.
12279         (vect_recog_rotate_pattern): New function.
12281 2013-05-16  Jason Merrill  <jason@redhat.com>
12283         * Makefile.in (LLINKER): New variable.
12284         (mostlyclean): Remove link mutex.
12285         * configure.ac: Handle --enable-link-mutex.
12286         * lock-and-run.sh: New script.
12288 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12290         PR target/19599
12291         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
12292         for NULL decl.
12294 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
12296         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
12298 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
12300         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
12301         * config/arm/arm.c (next_consecutive_mem): New function.
12302         (gen_movmem_ldrd_strd): Likewise.
12303         * config/arm/arm.md (movmemqi): Update condition and code.
12304         (unaligned_loaddi, unaligned_storedi): New patterns.
12306 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
12308         * config.gcc: Obsolete *-*-solaris2.9*.
12309         * doc/install.texi (Specific, *-*-solaris2*): Document it.
12311 2013-05-16  Richard Biener  <rguenther@suse.de>
12313         * passes.c (init_optimization_passes): Move pass_parallelize_loops
12314         earlier, after GRAPHITE transforms and IV canonicalization.
12316 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
12318         * omp-low.c (extract_omp_for_data): For collapsed loops,
12319         if at least one of the loops is known at compile time to
12320         iterate zero times, set count to 0.
12321         (expand_omp_regimplify_p): New function.
12322         (expand_omp_for_generic): For collapsed loops, if at least
12323         one of the loops isn't known to iterate at least once,
12324         add runtime check with setting count to 0.
12325         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
12326         For unsigned types if it isn't known at compile time that
12327         the loop will iterate at least once, add runtime check to bypass
12328         the whole loop if initial condition isn't true.
12330 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
12332         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
12334 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
12336         PR middle-end/57286
12337         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
12338         transformations to avoid an infinite loop.
12340 2013-05-16  Marek Polacek  <polacek@redhat.com>
12342         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
12344 2013-05-15  Leif Ekblad  <leif@rdos.net>
12346         * config/i386/i386.c (ix86_decompose_address): Use
12347         DEFAULT_TLS_SEG_REG to access TLS segment register.
12348         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
12349         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
12350         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
12352 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
12354         PR target/57260
12355         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
12356         sibling calls to functions that would normally be lazily bound,
12357         unless $gp is call-clobbered.
12359 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
12361         * config/i386/i386.c (ix86_option_override_internal): Update
12362         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
12363         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
12364         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
12365         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
12366         of TARGET_3DNOW.
12367         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
12369 2013-05-15  Andreas Schwab  <schwab@suse.de>
12371         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
12372         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
12373         third operand.
12375 2013-05-15  Teresa Johnson  <tejohnson@google.com>
12377         * loop-unroll.c (report_unroll_peel): Check decision before
12378         emitting unroll/peel message.
12380 2013-05-15  Teresa Johnson  <tejohnson@google.com>
12382         * function.h (has_bb_partition): New rtl_data flag.
12383         (bb_reorder_complete): Ditto.
12384         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
12385         instead of flag_reorder_blocks_and_partition.
12386         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
12387         with some enhancements.
12388         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
12389         * bb-reorder.c (connect_traces): Check for has_bb_partition
12390         instead of flag_reorder_blocks_and_partition.
12391         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
12392         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
12393         verify_hot_cold_block_grouping.
12394         (partition_hot_cold_basic_blocks): Set has_bb_partition.
12396 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12398         PR target/19599
12399         * config/arm/predicates.md (call_insn_operand): New predicate.
12400         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
12401         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
12402         if insn is not a tail call.
12403         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
12404         registers.
12405         * config/arm/arm.h (enum reg_class): New caller save register class.
12406         (REG_CLASS_NAMES): Likewise.
12407         (REG_CLASS_CONTENTS): Likewise.
12408         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
12409         without decls.
12411 2013-05-15  Richard Biener  <rguenther@suse.de>
12413         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
12414         of MSG_OPTIMIZED_LOCATIONS.
12415         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
12416         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
12417         message.
12418         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
12419         of MSG_OPTIMIZED_LOCATIONS.
12420         (execute_vect_slp): Likewise.
12421         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
12422         (vect_create_cond_for_alias_checks): Likewise.
12423         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
12424         (vect_recog_widen_mult_pattern): Likewise.
12425         (vect_recog_widen_sum_pattern): Likewise.
12426         (vect_recog_over_widening_pattern): Likewise.
12427         (vect_recog_widen_shift_pattern): Likewise.
12428         (vect_recog_vector_vector_shift_pattern): Likewise.
12429         (vect_recog_divmod_pattern): Likewise.
12430         (vect_recog_mixed_size_cond_pattern): Likewise.
12431         (vect_recog_bool_pattern): Likewise.
12432         (vect_pattern_recog_1): Likewise.
12434 2013-05-15  Martin Jambor  <mjambor@suse.cz>
12436         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
12437         non-functions to builtin_unreachable.
12438         * ipa-inline-transform.c (inline_call): Do not assert estimates were
12439         correct when new direct edges were discovered.
12441 2013-05-15  Martin Jambor  <mjambor@suse.cz>
12443         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
12444         header, print symbol order instead of node uid, print more information
12445         about indirect edge targets.
12446         (ipa_make_edge_direct_to_target): Print symbol order instead of node
12447         uids.
12448         (ipa_make_edge_direct_to_target): Likewise.
12449         (remove_described_reference): Likewise.
12450         (propagate_controlled_uses): Likewise.
12451         (ipa_print_node_params): Also print symbol order.
12452         (ipcp_transform_function): Print symbol order instead of node uids.
12453         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
12454         (cgraph_get_create_real_symbol_node): Likewise.
12455         * ipa-cp.c (print_lattice): Likewise.
12456         (print_all_lattices): Likewise.
12457         (determine_versionability): Likewise.
12458         (initialize_node_lattices): Likewise.
12459         (estimate_local_effects): Likewise.
12460         (update_profiling_info): Likewise.
12461         (create_specialized_node): Likewise.
12462         (perhaps_add_new_callers): Likewise.
12463         (decide_about_value): Likewise.
12464         (decide_whether_version_node): Likewise.
12465         (identify_dead_nodes): Likewise.
12466         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
12467         (dump_inline_summary): Likewise.
12468         (estimate_node_size_and_time): Likewise.
12469         (inline_analyze_function): Likewise.
12470         * ipa-inline.c (report_inline_failed_reason): Likewise.
12471         (want_early_inline_function_p): Likewise.
12472         (edge_badness): Likewise.
12473         (update_edge_key): Likewise.
12474         (inline_small_functions): Likewise.  Add dumping of order to two other
12475         dumps.
12476         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
12477         instead of node uids.
12478         (propagate_pure_const): Likewise.
12479         (propagate_pure_const): Likewise.
12480         * ipa-utils.c (dump_cgraph_node_set): Likewise.
12481         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
12482         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
12483         of node uids.
12484         * tree-pretty-print.c (dump_function_header): Likewise.
12485         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
12486         Print symbol order instead of node uids.
12488 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12490         * config/s390/s390.c (s390_register_move_cost): Don't impose the
12491         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
12493 2013-05-15  Richard Biener  <rguenther@suse.de>
12495         PR tree-optimization/57275
12496         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
12497         return value for fail to do runtime alias checks for gather loads.
12499 2013-05-15  Jan Hubicka  <jh@suse.cz>
12501         PR lto/57038
12502         PR lto/47375
12503         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
12504         weakrefs are not external.
12505         (lto_symtab_merge_decls): Fix thinko when dealing with
12506         non-lto_symtab decls.
12507         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
12508         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
12509         * varpool.c (dump_varpool_node): Dump more flags.
12511 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
12513         * config/i386/i386.c (processor_alias_table): Add instruction
12514         FSGSBASE for AMD bdver3 architecture.
12516 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
12518         * tree.c (warn_deprecated_use): Print file:line using locus color.
12519         * diagnostic.c (diagnostic_report_current_module): Print file:line
12520         and file:line:column using locus color.
12522 2013-05-14  Mike Stump  <mikestump@comcast.net>
12524         * gdbinit.in: Add __null.
12526 2013-05-14  Mike Stump  <mikestump@comcast.net>
12528         * recog.h: Rename struct recog_data to Recog_data.
12529         * recog.c: Likewise.
12530         * reload.c (can_reload_into): Likewise.
12531         * config/picochip/picochip.c: Likewise.
12533 2013-05-14  Mike Stump  <mikestump@comcast.net>
12535         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
12537 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
12539         * resource.h (struct resources): Remove unch_memory member.
12540         (CLEAR_RESOURCE): Don't clear unch_memory.
12541         * resource.c (mark_referenced_resources): Don't set it.
12542         (mark_set_resources): Likewise.
12543         (mark_target_live_regs): Don't clear it.
12544         (init_resource_info): Likewise.
12545         * reorg.c (resource_conflicts_p): Don't compare it.
12546         (redundant_insn): Don't set it.
12548         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
12549         Remove prototypes.
12550         * emit-rtl.c (next_label): Remove unused function.
12551         (skip_consecutive_labels, link_cc0_insns): Move to ...
12552         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
12553         only place where these functions are used, and make them static.
12555 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
12557         * fold-const.c (fold_negate_expr): Handle vectors.
12558         (fold_truth_not_expr): Make it static.
12559         (fold_invert_truthvalue): New static function.
12560         (invert_truthvalue_loc): Handle vectors. Do not call
12561         fold_truth_not_expr directly.
12562         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
12563         <TRUTH_NOT_EXPR>: Do not cast to boolean.
12564         (fold_comparison): Handle vector constants.
12565         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
12566         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
12567         * tree.h (fold_truth_not_expr): Remove declaration.
12569 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
12571         * config/aarch64/aarch64-simd.md
12572         (aarch64_vcond_internal<mode>): Rename to...
12573         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
12574         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
12575         float modes. Clarify all iterator modes.
12576         (vcond<mode><mode>): Use new name for vcond expanders.
12577         (vcond<v_cmp_result><mode>): Likewise.
12578         (vcondu<mode><mode>: Likewise.
12579         * config/aarch64/iterators.md (VDQF_COND): New.
12581 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
12583         PR bootstrap/57266
12584         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
12585         variable for the shift amount. Check that we shift by non-negative
12586         amounts.
12588 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
12590         PR target/42017
12591         * config/arm/arm.h (EPILOGUE_USES): Only return true
12592         for LR_REGNUM after epilogue_completed.
12594 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
12596         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
12597         is error_mark_node.
12599 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
12601         PR target/57261
12602         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
12603         and Solaris 11+/x86 with gld.
12604         * configure: Regenerate.
12606 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
12608         * expmed.c (expand_shift_1): Canonicalize rotates by
12609         constant bitsize / 2 to bitsize - 1.
12610         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
12611         case ROTATERT>: Likewise.
12613         Revert:
12614         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
12616         * config/i386/i386.md (rotateinv): New code attr.
12617         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
12618         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
12619         roll $31, %eax, etc.
12621 2013-05-14  Richard Biener  <rguenther@suse.de>
12623         PR middle-end/57235
12624         * tree-eh.c (sink_clobbers): Give up for successors with
12625         multiple predecessors and no virtual uses.
12627 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
12629         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
12630         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
12632 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
12634         PR middle-end/57251
12635         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
12636         the case when both op0 and op1 have VOIDmode.
12638 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
12640         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
12641         in multiply-accumulate mode.
12643 2013-05-13  Guozhi Wei  <carrot@google.com>
12645         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
12647 2013-05-13  Kai Tietz  <ktietz@redhat.com>
12649         PR target/56975
12650         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
12651         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
12652         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
12653         * config/i386/i386.c (ix86_option_override_internal): Likewise.
12654         (ix86_expand_prologue): Likewise.
12655         (ix86_expand_split_stack_prologue): Likewise.
12656         (legitimate_pic_address_disp_p): Likewise.
12657         (legitimize_pic_address): Likewise.
12658         (legitimize_tls_address): Likewise.
12659         (legitimize_pe_coff_symbol): Likewise.
12660         (output_pic_addr_const): Likewise.
12661         (construct_plt_address): Likewise.
12662         (ix86_expand_call): Likewise.
12663         (x86_output_mi_thunk): Likewise.
12664         (x86_function_profiler): Likewise.
12666 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
12668         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
12669         similar switch cases.
12670         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
12671         (aarch64_simd_mov_to_<mode>low): Delete.
12672         (aarch64_simd_mov_to_<mode>high): Delete.
12673         (move_lo_quad_<mode>): Add w<-r alternative.
12674         (aarch64_simd_move_hi_quad_<mode>): Likewise.
12675         (aarch64_simd_mov_from_*): Update type attribute.
12676         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
12677         statement.
12679 2013-05-13  Jan Hubicka  <jh@suse.cz>
12681         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
12682         * config/i386/i386.c (ix86_compute_frame_layout,
12683         ix86_expand_epilogue, emit_i387_cw_initialization,
12684         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
12685         ix86_local_alignment): Fix use of size/speed predicates.
12687 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
12689         PR tree-optimization/45216
12690         PR tree-optimization/57157
12691         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
12692         the (-Y) & (B - 1) variant if OP is |.
12693         * expmed.c (expand_shift_1): For rotations by const0_rtx just
12694         return shifted.  Use (-op1) & (prec - 1) as other_amount
12695         instead of prec - op1.
12697 2013-05-13  Martin Jambor  <mjambor@suse.cz>
12699         PR middle-end/42371
12700         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
12701         (ipa_constant_data): New type.
12702         (ipa_jump_func): Use ipa_constant_data to hold information about
12703         constant jump functions.
12704         (ipa_get_jf_constant): Adjust to jump function type changes.
12705         (ipa_get_jf_constant_rdesc): New function.
12706         (ipa_param_descriptor): New field controlled_uses.
12707         (ipa_get_controlled_uses): New function.
12708         (ipa_set_controlled_uses): Likewise.
12709         * ipa-ref.h (ipa_find_reference): Declare.
12710         * ipa-prop.c (ipa_cst_ref_desc): New type.
12711         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
12712         changes.
12713         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
12714         New parameter cs.  Adjust all callers.
12715         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
12716         (remove_described_reference): New function.
12717         (jfunc_rdesc_usable): Likewise.
12718         (try_make_edge_direct_simple_call): Decrement controlled use count,
12719         attempt to remove reference if it hits zero.
12720         (combine_controlled_uses_counters): New function.
12721         (propagate_controlled_uses): Likewise.
12722         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
12723         (ipa_edge_duplication_hook): Duplicate reference descriptions.
12724         (ipa_print_node_params): Print described use counter.
12725         (ipa_write_jump_function): Adjust to jump function type changes.
12726         (ipa_read_jump_function): New parameter CS, pass it to
12727         ipa_set_jf_constant.  Adjust caller.
12728         (ipa_write_node_info): Stream controlled use count
12729         (ipa_read_node_info): Likewise.
12730         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
12731         asserting.
12732         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
12733         count.  Remove cloning-added reference if it reaches zero.
12734         * ipa-ref.c (ipa_find_reference): New function.
12736 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
12738         * config/i386/i386.c (processor_target_table): Modified default
12739         alignment values for AMD BD and BT architectures.
12741 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
12743         * tree-vect-generic.c (uniform_vector_p): Move ...
12744         * tree.c (uniform_vector_p): ... here.
12745         * tree.h (uniform_vector_p): Declare it.
12746         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
12747         into a scalar.
12749 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
12751         PR tree-optimization/57230
12752         * tree-ssa-strlen.c (handle_char_store): Record length for
12753         array store from STRING_CST.
12755         PR tree-optimization/57230
12756         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
12757         check.
12759 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
12761         * config/epiphany/epiphany.c (epiphany_init): Check size of
12762         NUM_MODES_FOR_MODE_SWITCHING.
12763         (epiphany_expand_prologue):
12764         Remove CONFIG_REGNUM initial value handling code.
12765         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
12766         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
12767         (emit_set_fp_mode, epiphany_mode_after): Likewise.
12768         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
12769         Don't return 1 for FP_MODE_NONE.
12770         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
12771         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
12772         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
12773         * config/epiphany/epiphany.md (save_config): New pattern.
12775 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
12777         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
12779 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
12781         * config/i386/i386.md (memory): Handle sseishft1.
12782         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
12783         (*vec_extractv2di_1): Ditto.
12785 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
12787         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
12788         saved registers.
12790 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
12792         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
12793         Add mthumb/march=armv7-a multilib.
12794         Add mthumb/march=armv7-r multilib.
12795         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
12797 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
12799         * config/v850/t-rtems: Add more multilibs.
12801 2013-05-10  Richard Biener  <rguenther@suse.de>
12803         PR tree-optimization/57214
12804         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
12805         not propagate from SSA names that occur in abnormal PHI nodes.
12807 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
12809         * stor-layout.c (element_precision): New function.
12810         * machmode.h (element_precision): Declare it.
12811         * tree.c (build_minus_one_cst): New function.
12812         (element_precision): Likewise.
12813         * tree.h (build_minus_one_cst): Declare new function.
12814         (element_precision): Likewise.
12815         * fold-const.c (operand_equal_p): Use element_precision.
12816         (fold_binary_loc): Handle vector types.
12817         * convert.c (convert_to_integer): Use element_precision.
12818         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
12819         separately.
12821 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
12823         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
12824         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
12825         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
12826         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
12827         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
12828         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
12829         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
12830         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
12831         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
12832         (Uuw8): New constraints.
12833         (Usb4): Move into alphabetical order.
12834         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
12835         (sd8_operand, ub8_operand, uw8_operand): New predicates.
12836         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
12837         previously unnamed patterns.
12838         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
12839         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
12840         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
12841         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
12842         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
12843         of specific lengths.
12845 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
12847         * config/i386/i386.md (rotateinv): New code attr.
12848         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
12849         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
12850         roll $31, %eax, etc.
12852         PR tree-optimization/45216
12853         PR tree-optimization/57157
12854         * tree-ssa-forwprop.c (simplify_rotate): New function.
12855         (ssa_forward_propagate_and_combine): Call it.
12857 2013-05-10  Richard Biener  <rguenther@suse.de>
12859         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
12860         disable peeling when we version for aliasing.
12861         (vector_alignment_reachable_p): Honor explicit user alignment.
12862         (vect_supportable_dr_alignment): Likewise.
12863         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
12864         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
12865         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
12866         then peeling to arrange for the cost-model check to come first.
12868 2013-05-10  Alan Modra  <amodra@gmail.com>
12870         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
12871         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
12872         * configure: Regenerate.
12874 2013-05-10  Alan Modra  <amodra@gmail.com>
12876         PR target/55033
12877         * varasm.c (default_elf_select_section): Move !DECL_P check..
12878         (get_named_section): ..to here before calling get_section_name.
12879         Adjust assertion.
12880         (default_section_type_flags): Add DECL_P check.
12881         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
12882         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
12884 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
12886         * config/epiphany/epiphany.c (epiphany_expand_prologue):
12887         When using gen_stack_adjust_str with a register offset, add a
12888         REG_FRAME_RELATED_EXPR note.
12890 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
12892         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
12893         (*vec_extractv4si_zext_mem): Ditto.
12894         (*vec_extractv2di): Add 0->x and x->x alternatives.
12895         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
12896         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
12898 2013-05-09  Jason Merrill  <jason@redhat.com>
12900         N3639 C++1y VLA support
12901         * gimplify.c (gimplify_vla_decl): Don't touch an existing
12902         DECL_VALUE_EXPR.
12904         * tree.c (build_constructor_va): New.
12905         * tree.h: Declare it.
12907 2013-05-09  Martin Jambor  <mjambor@suse.cz>
12909         PR lto/57084
12910         * gimple-fold.c (canonicalize_constructor_val): Call
12911         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
12913 2013-05-09  Jan Hubicka  <jh@suse.cz>
12914             Richard Biener  <rguenther@suse.de>
12916         PR lto/54095
12917         * symtab.c (symtab_make_decl_local): Do not add private names.
12919 2013-05-09  Jan Hubicka  <jh@suse.cz>
12921         PR lto/54095
12922         * symtab.c (insert_to_assembler_name_hash): Handle clones.
12923         (unlink_from_assembler_name_hash): Likewise.
12924         (symtab_prevail_in_asm_name_hash, symtab_register_node,
12925         symtab_unregister_node, symtab_initialize_asm_name_hash,
12926         change_decl_assembler_name): Update.
12928 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
12930         * config/aarch64/aarch64.md: New movtf split.
12931         (*movtf_aarch64): Update.
12932         (aarch64_movdi_tilow): Handle TF modes and rename to
12933         aarch64_movdi_<mode>low.
12934         (aarch64_movdi_tihigh): Handle TF modes and rename to
12935         aarch64_movdi_<mode>high
12936         (aarch64_movtihigh_di): Handle TF modes and rename to
12937         aarch64_mov<mode>high_di
12938         (aarch64_movtilow_di): Handle TF modes and rename to
12939         aarch64_mov<mode>low_di
12940         (aarch64_movtilow_tilow): Remove spurious whitespace.
12941         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
12942         splits.
12943         (aarch64_print_operand): Update.
12945 2013-05-09  Alan Modra  <amodra@gmail.com>
12947         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
12948         powerpc64le.
12949         * configure: Regenerate.
12951 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
12953         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
12954         splitter preparation statements.
12955         * config/i386/sse.md (*vec_extract* splitters): Ditto.
12956         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
12957         adjust_address_nv.
12959 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12961         * gimple-ssa-strength-reduction.c (count_candidates): Change
12962         return value to int.
12963         (analyze_candidates_and_replace): Change type of length to int.
12965 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
12967         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
12968         (*vec_extract<mode>): Use VI12_128 mode iterator.
12969         (*vec_extract<mode>_mem): Ditto.
12970         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
12971         attribute.
12973 2013-05-08  Diego Novillo  <dnovillo@google.com>
12975         PR bootstrap/54659
12977         Revert:
12978         2012-08-17  Diego Novillo  <dnovillo@google.com>
12980         PR bootstrap/54281
12981         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
12982         * config.in: Regenerate.
12983         * configure: Regenerate.
12984         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
12986 2013-05-08  Jan Hubicka  <jh@suse.cz>
12988         PR lto/54095
12989         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
12990         * cgraph.h (symtab_node_base): Add unique_name.
12991         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
12992         input_overwrite_node, input_varpool_node): Stream unique_name.
12993         * cgraphclones.c (cgraph_create_virtual_clone,
12994         cgraph_function_versioning): Set unique_name.
12995         * ipa.c (function_and_variable_visibility): Set unique_name.
12997 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12999         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
13000         (alloc_cand_and_find_basis): Restrict conditional candidate
13001         processing to CAND_MULTs.
13003 2013-05-08  Jan Hubicka  <jh@suse.cz>
13005         PR lto/54095
13006         lto-symtab.c (lto_symtab_symbol_p): New function.
13007         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
13008         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
13009         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
13010         Skip static symbols.
13012 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
13014         PR tree-optimization/57200
13015         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
13016         Only call inform if the preceding warning_at returns true.
13018 2013-05-07  Han Shen  <shenhan@google.com>
13020         * cfgexpand.c (record_or_union_type_has_array_p): New function.
13021         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
13022         * common.opt (fstack-protector-strong): New option.
13023         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
13024         * doc/invoke.texi (Optimization Options): Document
13025         "-fstack-protector-strong".
13026         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
13028 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
13030         * config/mips/mips.c (mips_machine_reorg2): Return 0.
13032 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
13034         * ira.c (update_equiv_regs): Add insn having equiv memory even if
13035         it is not lhs of the insn.
13036         (setup_reg_equiv): Remove insn having equiv memory which it is not
13037         lhs of the insn.
13038         * lra-constraints.c (process_address): Try to improve generation
13039         code for address base + disp.
13040         (lra_constraints): Make correct the code for checking insn setting
13041         up backward equivalence.  Remove insn only if it is in the init
13042         insn list.
13043         * lra-eliminations.c (update_reg_eliminate): Change return value.
13044         (lra_eliminate): Use the result.
13046 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
13048         * config/i386/sse.md (ssescalarnummask): New mode attribute.
13049         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
13050         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
13051         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
13052         register target operands.
13053         (*vec_extractv8hi_sse2): New pattern.
13054         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
13055         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
13056         (*vec_extract<mode>_mem): New insn and split pattern.
13058 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
13060         * config/arm/arm.c (arm_asan_shadow_offset): New function.
13061         (TARGET_ASAN_SHADOW_OFFSET): Define.
13062         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
13063         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
13065 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13067         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
13068         (incr_vec_index): Return -1 if increment not found.
13069         (create_add_on_incoming_edge): Assert if increment not found.
13070         (record_increment): Limit number of increments recorded.
13071         (all_phi_incrs_profitable): Return false if an increment not found.
13072         (replace_profitable_candidates): Don't process increments that were
13073         not recorded.
13074         (analyze_candidates_and_replace): Limit size of incr_vec.
13076 2013-05-07  Richard Biener  <rguenther@suse.de>
13078         * calls.c (special_function_p): setjmp-like functions are leaf.
13079         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
13080         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
13082 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
13084         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
13085         (aarch64_simd_mov<mode>): New expander.
13086         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
13087         (aarch64_simd_mov_to_<mode>high): Likewise.
13088         (aarch64_simd_mov_from_<mode>low): Likewise.
13089         (aarch64_simd_mov_from_<mode>high): Likewise.
13090         (aarch64_dup_lane<mode>): Update.
13091         (aarch64_dup_lanedi): New instruction pattern.
13092         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
13093         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
13095 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13097         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
13098         (replace_mult_candidate): Remove unnecessary argument; remove
13099         unnecessary parameter from call to introduce_cast_before_cand.
13100         (replace_unconditional_candidate): Remove unnecessary parameter
13101         from call to replace_mult_candidate.
13102         (replace_conditional_candidate): Likewise.
13103         (insert_initializers): Use make_temp_ssa_name.
13104         (introduce_cast_before_cand): Remove unnecessary argument; use
13105         make_temp_ssa_name.
13106         (replace_one_candidate): Remove unnecessary argument; remove
13107         unnecessary parameter from calls to introduce_cast_before_cand.
13108         (replace_profitable_candidates): Remove unnecessary parameters
13109         from calls to replace_one_candidate.
13111 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13113         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13114         phi def as possibly hiding a basis for a CAND_ADD whose operands
13115         have been commuted in the analysis.
13116         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
13118 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13120         * config/aarch64/aarch64.md
13121         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
13122         shift value between 0-4.
13124 2013-05-07  Richard Biener  <rguenther@suse.de>
13126         * double-int.h (rshift): New overload.
13127         * double-int.c (rshift): New function.
13128         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
13129         (create_reference_ops_from_ref): Remove.
13130         (vn_reference_insert): Use shared ops for constructing the
13131         reference and copy it.
13133 2013-05-07  Richard Biener  <rguenther@suse.de>
13135         PR middle-end/57190
13136         * tree-eh.c (sink_clobbers): Properly propagate
13137         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
13139 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
13141         PR tree-optimization/57149
13142         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
13143         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
13144         collect_phi_def_edges, execute_late_warn_uninitialized): Use
13145         uninit_undefined_value_p instead of ssa_undefined_value_p.
13147         PR debug/57184
13148         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
13149         for modifier == EXPAND_INITIALIZER.
13151 2013-05-07  Anton Blanchard  <anton@samba.org>
13153         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
13154         for powerpc64 little endian.
13155         * configure: Regenerate.
13157 2013-05-06  Graham Stott  <grahams@btinternet.com>
13159         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
13160         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
13161         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
13162         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
13164 2013-05-06  Graham Stott  <grahams@btinternet.com>
13166         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
13167         codes which allow non-lvalues.
13169 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
13171         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
13172         components are all 1s.
13173         (integer_minus_onep): New function.
13174         * tree.h (integer_minus_onep): Declare it.
13175         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
13176         integer_minus_onep instead of integer_all_onesp.
13178 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
13180         PR target/52933
13181         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
13182         variations of these patterns.
13184 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
13186         * config/i386/i386.md (isa): Add x64_sse4 member.
13187         (enabled): Handle x64_sse4.
13188         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
13189         instruction for 64bit SSE4_1 targets.  Update insn attributes.
13190         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
13191         instruction for SSE4_1 targets.  Update insn attributes.
13192         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
13193         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
13194         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
13195         const_1 selector.
13196         (*vec_extractv4si): Rename from *sse4_1_pextrd.
13197         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
13198         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
13200 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
13202         PR target/57108
13203         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
13205 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
13207         * final.c (do_assembler_dialects): Don't handle curly braces and
13208         vertical bar escaped by % as dialect delimiters.
13209         (output_asm_insn): Print curly braces and vertical bar if escaped
13210         by % and ASSEMBLER_DIALECT defined.
13211         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
13212         * doc/tm.texi: Regenerated.
13214 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
13216         * config/mips/mips.c: Include tree-pass.h.
13217         (mips_reorg): Split in pre- and post-dbr_schedule parts.
13218         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
13219         (pass_mips_machine_reorg2): New machine specific pass.
13220         (insert_pass_mips_machine_reorg2): New pass plugin definition.
13221         (mips_option_override): Register the new pass.
13222         * rtl.h (cleanup_barriers): Remove prototype.
13223         (dbr_schedule): Likewise.
13224         * jump.c (cleanup_barriers): Make static.
13225         * reorg.c (dbr_schedule): Likewise.
13227 2013-05-06  Richard Biener  <rguenther@suse.de>
13229         PR tree-optimization/57185
13230         * tree-parloops.c (add_field_for_reduction): Handle anonymous
13231         SSA names properly.
13233 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
13235         PR target/57106
13236         * config/i386/i386.c (add_parameter_dependencies): Add dependence
13237         between "first_arg" and "insn", not "last" and "insn".
13239 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13241         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
13242         (find_candidates_in_block): Re-enable slsr_process_phi.
13243         (create_phi_basis): Fix double counting of candidate adjustment.
13245 2013-05-06  Richard Biener  <rguenther@suse.de>
13247         PR middle-end/57147
13248         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
13249         the edge is also fallthru, preserve it and just clear the
13250         abnormal flag.
13251         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
13252         also complex, preserve that and just clear the fallthru flag.
13253         * tree-inline.c (update_ssa_across_abnormal_edges): Also
13254         update virtual operands.
13256 2013-05-06  Alan Modra  <amodra@gmail.com>
13258         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
13259         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
13260         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
13261         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
13262         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
13263         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
13265 2013-05-06  Alan Modra  <amodra@gmail.com>
13267         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
13268         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
13269         (DEFAULT_ASM_ENDIAN): Define.
13270         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
13271         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
13272         Update -K PIC clause from sysv4.h.
13273         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
13274         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
13276 2013-05-06  Alan Modra  <amodra@gmail.com>
13278         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
13279         twice for little-endian.
13280         (ashrdi3_no_power, ashrdi3): Support little-endian.
13282 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
13284         PR target/55303
13285         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
13286         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
13287         related expanders.
13288         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
13289         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
13290         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
13291         New predicates.
13293 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
13294             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
13296         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
13297         * config/pa/pa.opt: Make mbig-switch a no-op.
13298         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
13299         (CASE_VECTOR_MODE): Always return SImode.
13300         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
13301         for the !TARGET_BIG_SWITCH case.
13302         * config/pa/pa-linux.h: Likewise.
13303         * config/pa/pa-openbsd.h: Likewise.
13304         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
13305         * config/pa/pa.md (short_jump): Remove define_insn.
13306         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
13307         (casesi0): Remove define_insn.
13308         (type): Remove btable_branch.
13309         (pa_combine_type): Likewise.
13310         (in_nullified_branch_delay): Likewise.
13311         (in_call_delay): Likewise.
13312         (define_delay): Likewise.
13313         (define_insn_reservation "Z3"): Likewise.
13314         (define_insn_reservation "Z4"): Likewise.
13315         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
13316         (pa_adjust_insn_length): Remove adjustment for btable branches.
13317         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
13318         and mno-big-switch
13320 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
13322         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
13323         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
13324         Add m->r,x alternatives.
13325         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
13326         splitters using SWI48x mode iterator.
13327         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
13328         TARGET_64BIT.  Add m->x alternative.
13329         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
13330         Add o->x alternative.  Enable for TARGET_SSE.
13331         (sse_storeq): Remove expander.
13332         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
13333         with memory input operand.
13334         (*vec_extractv2di_1 splitter): New.
13335         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
13336         * config/i386/i386.md (ssevecmodelower): New mode attribute.
13338 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
13340         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
13341         (INT_LOWPART): Delete.
13342         (extract_MB): Adjust.
13343         (extract_ME): Adjust.
13344         (print_operand): Adjust.
13346 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
13348         * config/rs6000/predicates.md (reg_or_add_cint_operand,
13349         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
13350         (reg_or_logical_cint_operand, easy_fp_constant,
13351         logical_const_operand): Delete "CONST_DOUBLE" case.
13352         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
13353         "HOST_BITS_PER_WIDE_INT == 64" test.
13354         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
13355         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
13356         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
13357         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
13358         test.
13359         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
13360         CONST_DOUBLE DImode/VOIDmode case.
13361         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
13362         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
13363         CONST_DOUBLE VOIDmode case.
13364         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
13365         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
13366         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
13367         Delete CONST_DOUBLE case.
13368         (splitters for mov FMOVE64 const_double): Delete
13369         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
13370         "HOST_BITS_PER_WIDE_INT >= 64" test.
13371         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
13372         case.
13373         (mov DI const_double): Delete.
13375 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
13377         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
13378         on op shows all bits zero in mode of a lowpart subreg, return zero.
13380 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
13382         PR target/57150
13383         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
13384         to save TFmode registers and DImode to save TImode registers for
13385         caller save operations.
13386         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
13387         mark being partially clobbered since they only use the first
13388         double word.
13390         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
13391         and TDmode only use the upper 64-bits of each VSX register.
13393 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13395         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
13396         (find_candidates_in_block): Disable slsr_process_phi.
13398 2013-05-03  Guozhi Wei  <carrot@google.com>
13400         * coverage.c (coverage_obj_init): Move the construction of gcov
13401         constructor to ...
13402         (build_init_ctor): ... here.
13404 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13406         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
13407         (slsr_cand_d): Redefine def_phi.
13408         (stride_status, phi_adjust_status, count_phis_status): New enums.
13409         (find_phi_def): New.
13410         (find_basis_for_base_expr): New.
13411         (find_basis_for_candidate): Handle hidden bases.
13412         (alloc_cand_and_find_basis): Handle phi candidates.
13413         (slsr_process_phi): New.
13414         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
13415         (create_mul_imm_cand): Likewise.
13416         (create_add_ssa_cand): Exclude phi base candidates.
13417         (create_add_imm_cand): Likewise.
13418         (slsr_process_cast): Likewise.
13419         (slsr_process_copy): Likewise.
13420         (find_candidates_in_block): Handle phi candidates.
13421         (dump_candidate): Likewise.
13422         (unconditional_cands): Delete.
13423         (unconditional_cands_with_known_stride_p): Delete.
13424         (phi_dependent_cand_p): New.
13425         (cand_increment): Handle phi-dependent candidates.
13426         (replace_dependent): Delete.
13427         (replace_mult_candidate): New.
13428         (replace_unconditional_candidate): New.
13429         (incr_vec_index): Move to avoid forward reference.
13430         (create_add_on_incoming_edge): New.
13431         (create_phi_basis): New.
13432         (replace_dependents): Delete.
13433         (replace_conditional_candidate): New.
13434         (phi_add_costs): New.
13435         (replace_uncond_cands_and_profitable_phis): New.
13436         (record_increment): Handle phi adjustments.
13437         (record_phi_increments): New.
13438         (record_increments): Handle phi adjustments.
13439         (phi_incr_cost): New.
13440         (lowest_cost_path): Handle phis.
13441         (total_savings): Likewise.
13442         (analyze_increments): Likewise.
13443         (ncd_with_phi): New.
13444         (ncd_of_cand_and_phis): New.
13445         (nearest_common_dominator_for_cands): Handle phi increments.
13446         (all_phi_incrs_profitable): New.
13447         (replace_profitable_candidates): Handle phi-dependent candidates.
13448         (analyze_candidates_and_replace): Likewise.
13450 2013-05-03  Teresa Johnson  <tejohnson@google.com>
13452         PR bootstrap/57154
13453         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
13454         do not exceed REG_BR_PROB_BASE.
13456 2013-05-03  Jeff Law  <law@redhat.com>
13458         PR tree-optimization/57144
13459         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
13460         operand of the condition will bit into the new type when eliminating
13461         a cast feeding a condition.
13463 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
13465         PR rtl-optimization/57130
13466         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
13467         of COMPARE as in_code to the recursive call if needed.
13469 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
13471         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
13472         (enabled): Handle new members.
13473         * config/i386/sse.md (*vec_concatv2si): Merge from
13474         *vec_concatv2si_sse2 and vec_concatv2si_sse.
13475         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
13477 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
13479         PR tree-optimization/57027
13480         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
13481         for fnms opportunity, check we got the prerequisite kind
13482         of tree / gimple before using accessor functions.
13484 2013-05-03  Richard Biener  <rguenther@suse.de>
13486         * double-int.h (lshift): New overload without precision
13487         and arith argument.
13488         (operator *=, operator +=, operator -=): Move ...
13489         * double-int.c (operator *=, operator +=, operator -=): ... here
13490         and implement more efficiently.
13491         (mul_double_with_sign): Remove.
13492         (lshift_double): Adjust to take unsinged shift argument, push
13493         dispatching code to callers.
13494         (mul_double_wide_with_sign): Add early out for callers that
13495         are not interested in high parts or overflow.
13496         (lshift): New function.
13497         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
13498         dispatch code here.
13499         (lrotate, rrotate): Use logical shifts.
13500         * expr.c (get_inner_reference): Use lshift.
13501         * fixed-value.c (do_fixed_divide): Likewise.
13502         * tree-dfa.c (get_ref_base_and_extent): Likewise.
13503         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
13504         (indirect_refs_may_alias_p): Likewise.
13505         (stmt_kills_ref_p_1): Likewise.
13507 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
13509         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
13511 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
13513         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
13514         scalar form of FABD instruction.
13516 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
13518         * lra-constraints.c (process_alt_operands): Add checking alt
13519         number to choose the best alternative.
13521 2013-05-02  Richard Biener  <rguenther@suse.de>
13523         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
13524         bitmap and its handling.
13525         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
13527 2013-05-02  Richard Biener  <rguenther@suse.de>
13529         PR middle-end/57140
13530         * tree-inline.c (copy_loops): Properly handle removed loops.
13531         (copy_cfg_body): Mark destination loops for fixup if source
13532         loops needed fixup.
13534 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
13536         PR target/56732
13537         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
13538         generating simple_return for naked functions.
13540 2013-05-02  Martin Jambor  <mjambor@suse.cz>
13542         PR middle-end/56988
13543         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
13544         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
13545         flags match.
13546         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
13547         ipa_agg_replacement_value structures.
13548         (known_aggs_to_agg_replacement_list): Likewise.
13549         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
13550         (read_agg_replacement_chain): Likewise.
13551         (ipcp_transform_function): Also check that by_ref flags match.
13553 2013-05-02  Richard Biener  <rguenther@suse.de>
13555         * graphds.h (struct graph): Add obstack member.
13556         * graphds.c (new_graph): Initialize obstack and allocate
13557         vertices from it.
13558         (add_edge): Allocate edge from the obstack.
13559         (free_graph): Free the obstack instead of all edges and vertices.
13561 2013-05-02  Teresa Johnson  <tejohnson@google.com>
13563         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
13564         divides.
13565         * cfg.c (update_bb_profile_for_threading): Ditto.
13566         * tree-inline.c (copy_bb): Ditto.
13567         (copy_edges_for_bb): Ditto.
13568         (initialize_cfun): Ditto.
13569         (copy_cfg_body): Ditto.
13570         (expand_call_inline): Ditto.
13571         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
13572         (estimate_node_size_and_time): Ditto.
13573         (inline_merge_summary): Ditto.
13574         * cgraphclones.c (cgraph_clone_edge): Ditto.
13575         (cgraph_clone_node): Ditto.
13576         * sched-rgn.c (compute_dom_prob_ps): Ditto.
13577         (compute_trg_info): Ditto.
13579 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
13581         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
13582         S reg when fp attribute set.
13583         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
13585 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
13587         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
13588         New pattern.
13589         (*and_one_cmplsi3_compare0_uxtw): Likewise.
13590         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
13591         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
13593 2013-05-02  Richard Biener  <rguenther@suse.de>
13595         * tree-scalar-evolution.c (scev_info_hasher): Remove.
13596         (struct instantiate_cache_entry): New type.
13597         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
13598         (struct instantiate_cache_type): New type.
13599         (set_instantiated_value, get_instantiated_value): Remove.
13600         (get_instantiated_value_entry): New function.
13601         (instantiate_scev_name): Use the new cache and adjust.
13602         (instantiate_scev_poly): Adjust.
13603         (instantiate_scev_binary): Likewise.
13604         (instantiate_array_ref): Likewise.
13605         (instantiate_scev_convert): Likewise.
13606         (instantiate_scev_not): Likewise.
13607         (instantiate_scev_3): Likewise.
13608         (instantiate_scev_2): Likewise.
13609         (instantiate_scev_r): Likewise.
13610         (instantiate_scev): Likewise.
13611         (resolve_mixers): Likewise.
13613 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
13615         PR target/57091
13616         * lra-constraints.c (best_small_class_operands_num): Remove.
13617         (process_alt_operands): Remove small_class_operands_num.  Take
13618         small classes operands into losers and only if the operand is not
13619         matched.  Modify debugging output.
13620         (curr_insn_transform): Remove best_small_class_operands_num.
13621         Print insn name.
13623 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13625         * config/aarch64/aarch64-builtins.c
13626         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
13627         * config/aarch64/aarch64-simd-builtins.def
13628         (reduc_splus_): Add new modes.
13629         (reduc_uplus_): New.
13630         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
13631         (reduc_uplus_v4sf): Likewise.
13632         (reduc_splus_v4sf): Likewise.
13633         (aarch64_addv<mode>): Likewise.
13634         (reduc_uplus_<mode>): Likewise.
13635         (reduc_splus_<mode>): Likewise.
13636         (aarch64_addvv2di): Likewise.
13637         (reduc_uplus_v2di): Likewise.
13638         (reduc_splus_v2di): Likewise.
13639         (aarch64_addvv2si): Likewise.
13640         (reduc_uplus_v2si): Likewise.
13641         (reduc_splus_v2si): Likewise.
13642         (reduc_<sur>plus_<mode>): New.
13643         (reduc_<sur>plus_v2di): Likewise.
13644         (reduc_<sur>plus_v2si): Likewise.
13645         (reduc_<sur>plus_v4sf): Likewise.
13646         (aarch64_addpv4sf): Likewise.
13647         * config/aarch64/arm_neon.h
13648         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
13649         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
13650         add UNSPEC_SADDV, UNSPEC_UADDV.
13651         (SUADDV): New.
13652         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
13654 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13656         * config/aarch64/arm_neon.h
13657         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
13659 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13661         * config/aarch64/aarch64-builtins
13662         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
13664 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13666         * config/aarch64/aarch64-simd-builtins.def
13667         (reduc_smax_): New.
13668         (reduc_smin_): Likewise.
13669         (reduc_umax_): Likewise.
13670         (reduc_umin_): Likewise.
13671         (reduc_smax_nan_): Likewise.
13672         (reduc_smin_nan_): Likewise.
13673         (fmax): Remove.
13674         (fmin): Likewise.
13675         (smax): Update for V2SF, V4SF and V2DF modes.
13676         (smin): Likewise.
13677         (smax_nan): New.
13678         (smin_nan): Likewise.
13679         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
13680         (<su><maxmin><mode>3): ...This, refactor.
13681         (s<maxmin><mode>3): New.
13682         (<maxmin_uns><mode>3): Likewise.
13683         (reduc_<maxmin_uns>_<mode>): Refactor.
13684         (reduc_<maxmin_uns>_v4sf): Likewise.
13685         (reduc_<maxmin_uns>_v2si): Likewise.
13686         (aarch64_<fmaxmin><mode>: Remove.
13687         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
13688         new builtin names.
13689         (vmin<q>_f<32,64>): Likewise.
13690         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
13691         (FMAXMIN): New.
13692         (su): Add mappings for smax, smin, umax, umin.
13693         (maxmin): New.
13694         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
13695         (FMAXMIN): Rename as...
13696         (FMAXMIN_UNS): ...This.
13697         (maxminv): Remove.
13698         (fmaxminv): Likewise.
13699         (fmaxmin): Likewise.
13700         (maxmin_uns): New.
13701         (maxmin_uns_op): Likewise.
13703 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13705         * config/aarch64/arm_neon.h
13706         (vac<ge, gt><sd>_f<32, 64>): Rename to...
13707         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
13708         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
13710 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13712         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
13713         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
13715 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13717         * config/aarch64/aarch64-simd.md
13718         (vcond<mode>_internal): Handle special cases for constant masks.
13719         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
13720         (vcondu<mode><mode>): Likewise.
13721         (vcond<v_cmp_result><mode>): New.
13723 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13725         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
13726         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
13727         * config/aarch64/aarch64-simd-builtins.def
13728         (cmeq): Update to BUILTIN_VALLDI.
13729         (cmgt): Likewise.
13730         (cmge): Likewise.
13731         (cmle): Likewise.
13732         (cmlt): Likewise.
13733         * config/aarch64/arm_neon.h
13734         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
13735         to builtins or C as appropriate.
13737 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13739         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
13740         (cmgeu): ...This.
13741         (cmhi): Rename to...
13742         (cmgtu): ...This.
13743         * config/aarch64/aarch64-simd.md
13744         (simd_mode): Add SF.
13745         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
13746         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
13747         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
13748         (cstore<mode>_neg): ...This.
13749         * config/aarch64/iterators.md
13750         (VALLF): new.
13751         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
13752         (COMPARISONS): New.
13753         (UCOMPARISONS): Likewise.
13754         (optab): Add missing comparisons.
13755         (n_optab): New.
13756         (cmp_1): Likewise.
13757         (cmp_2): Likewise.
13758         (CMP): Likewise.
13759         (cmp): Remove.
13760         (VCMP_S): Likewise.
13761         (VCMP_U): Likewise.
13762         (V_cmp_result): Add DF, SF modes.
13763         (v_cmp_result): Likewise.
13764         (v): Likewise.
13765         (vmtype): Likewise.
13766         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
13768 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
13770         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
13771         define_insn to define_insn_and_split.
13772         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
13773         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
13774         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
13775         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
13776         (thumb2_negscc): Likewise.
13778 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
13780         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
13782 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
13784         * config/arm/thumb2.md: Remove trailing whitespaces.
13786 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
13788         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
13789         Use gen_int_mode rather than GEN_INT.
13791 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
13793         * value-prof.c (stream_in_histogram_value): Remove the strayed
13794         debug_gimple_stmt.
13796 2013-04-30  Richard Biener  <rguenther@suse.de>
13798         PR middle-end/57122
13799         * cfghooks.c (split_edge): Properly check for the loop latch edge.
13801 2013-04-30  Richard Biener  <rguenther@suse.de>
13803         PR middle-end/57107
13804         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
13806 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
13808         PR rtl-optimization/56957
13809         PR rtl-optimization/57105
13810         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
13811         variable.  Use just INSN_UID for determining whether an insn
13812         should be only disconnected from the insn stream.
13813         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
13815 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
13817         PR tree-optimization/57104
13818         * tsan.c (instrument_expr): Don't instrument accesses to
13819         DECL_HARD_REGISTER VAR_DECLs.
13821 2013-04-30  Richard Biener  <rguenther@suse.de>
13823         * function.h (loops_for_fn): New inline function.
13824         (set_loops_for_fn): Likewise.
13825         * cfgloop.h (place_new_loop): Add struct function parameter.
13826         (get_loop): Likewise.
13827         (get_loops): Likewise.
13828         (number_of_loops): Likewise.
13829         (fel_next): Adjust.
13830         (fel_init): Likewise.
13831         * cfg.c (get_loop_copy): Adjust.
13832         * cfgloop.c (flow_loops_dump): Likewise.
13833         (record_loop_exits): Likewise.
13834         (verify_loop_structure): Likewise.
13835         * cfgloopanal.c (mark_irreducible_loops): Likewise.
13836         (estimate_reg_pressure_cost): Likewise.
13837         (mark_loop_exit_edges): Likewise.
13838         * cfgloopmanip.c (place_new_loop): Likewise.
13839         (add_loop): Likewise.
13840         (duplicate_loop): Likewise.
13841         * graph.c (draw_cfg_nodes): Likewise.
13842         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
13843         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
13844         (extract_affine_chrec): Likewise.
13845         (build_scop_iteration_domain): Likewise.
13846         * graphite.c (graphite_initialize): Likewise.
13847         * ira-build.c (create_loop_tree_nodes): Likewise.
13848         (more_one_region_p): Likewise.
13849         (rebuild_regno_allocno_maps): Likewise.
13850         (mark_loops_for_removal): Likewise.
13851         (mark_all_loops_for_removal): Likewise.
13852         (remove_unnecessary_regions): Likewise.
13853         (ira_build): Likewise.
13854         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
13855         * loop-init.c (fix_loop_structure): Likewise.
13856         (gate_rtl_move_loop_invariants): Likewise.
13857         (gate_rtl_unswitch): Likewise.
13858         (gate_rtl_unroll_and_peel_loops): Likewise.
13859         (rtl_doloop): Likewise.
13860         * lto-streamer-in.c (input_cfg): Likewise.
13861         * lto-streamer-out.c (output_cfg): Likewise.
13862         * modulo-sched.c (sms_schedule): Likewise.
13863         * predict.c (tree_estimate_probability): Likewise.
13864         (tree_estimate_probability_driver): Likewise.
13865         (estimate_loops): Likewise.
13866         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
13867         (move_sese_region_to_fn): Likewise.
13868         (debug_loop_num): Likewise.
13869         * tree-chrec.c (chrec_evaluate): Likewise.
13870         (hide_evolution_in_other_loops_than_loop): Likewise.
13871         (chrec_component_in_loop_num): Likewise.
13872         (reset_evolution_in_loop): Likewise.
13873         (evolution_function_is_invariant_rec_p): Likewise.
13874         * tree-if-conv.c (main_tree_if_conversion): Likewise.
13875         * tree-inline.c (copy_loops): Likewise.
13876         (copy_cfg_body): Likewise.
13877         (tree_function_versioning): Likewise.
13878         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
13879         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
13880         Likewise.
13881         (add_to_evolution_1): Likewise.
13882         (scev_const_prop): Likewise.
13883         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
13884         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
13885         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
13886         (tree_ssa_lim_initialize): Likewise.
13887         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
13888         (verify_loop_closed_ssa): Likewise.
13889         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
13890         (tree_ssa_loop_im): Likewise.
13891         (tree_ssa_loop_unswitch): Likewise.
13892         (tree_vectorize): Likewise.
13893         (check_data_deps): Likewise.
13894         (tree_ssa_loop_ivcanon): Likewise.
13895         (tree_ssa_loop_bounds): Likewise.
13896         (tree_complete_unroll): Likewise.
13897         (tree_complete_unroll_inner): Likewise.
13898         (tree_parallelize_loops): Likewise.
13899         (tree_ssa_loop_prefetch): Likewise.
13900         (tree_ssa_loop_ivopts): Likewise.
13901         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
13902         * tree-vectorizer.c (vectorize_loops): Likewise.
13904 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
13906         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
13907         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
13908         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
13909         with EABI_LINK_SPEC.
13911 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
13913         PR target/44578
13914         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
13915         alternative.
13917 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
13919         PR target/57097
13920         * lra-constraints.c (process_alt_operands): Discourage a bit more
13921         using memory for pseudos.  Print cost dump for alternatives.
13922         Modify cost values for conflicts with early clobbers.
13923         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
13925 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
13927         PR target/57098
13928         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
13930 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
13932         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
13933         from/to S register.
13934         (movdi_aarch64): Support LDR/STR from/to D register.
13936 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
13938         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
13939         or higher by default.
13941 2013-04-29  Richard Biener  <rguenther@suse.de>
13943         PR middle-end/57075
13944         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
13945         even if not adding abnormal edges for calls that can make
13946         abnormal gotos.
13948 2013-04-29  Richard Biener  <rguenther@suse.de>
13950         PR middle-end/57103
13951         * tree-cfg.c (move_stmt_op): Fix condition under which to update
13952         TREE_BLOCK.
13953         (move_stmt_r): Remove redundant checking.
13955 2013-04-29  Teresa Johnson  <tejohnson@google.com>
13957         PR bootstrap/57077
13958         * basic-block.h (apply_scale): New function.
13959         (apply_probability): Use apply_scale.
13960         * gimple-streamer-in.c (input_bb): Ditto.
13961         * lto-streamer-in.c (input_cfg): Ditto.
13962         * lto-cgraph.c (merge_profile_summaries): Ditto.
13963         * tree-optimize.c (execute_fixup_cfg): Ditto.
13964         * tree-inline.c (copy_bb): Update comment to use apply_scale.
13965         (copy_edges_for_bb): Ditto.
13966         (copy_cfg_body): Ditto.
13968 2013-04-29  Tom de Vries  <tom@codesourcery.com>
13970         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
13971         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
13972         (tail_merge_optimize): Handle current_loops == NULL.
13974 2013-04-26  Jeff Law  <law@redhat.com>
13976         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
13977         (simplify_cond_using_ranges): Generalize code to simplify
13978         COND_EXPRs where one argument is a constant and the other
13979         is an SSA_NAME created by an integral type conversion.
13981 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13983         * config/arm/arm.md (store_minmaxsi): Use only when
13984         optimize_insn_for_size_p.
13986 2013-04-29  Christian Bruel  <christian.bruel@st.com>
13988         PR target/57108
13989         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
13991 2013-04-29  Richard Biener  <rguenther@suse.de>
13993         PR middle-end/57089
13994         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
13995         loop tree make sure to schedule a fixup for the child as well.
13996         (expand_omp_for_generic): Properly add loops.
13997         (expand_omp_for_static_nochunk): Likewise.
13998         (expand_omp_for_static_chunk): Likewise.
13999         (expand_omp_for): For the degenerate case fixup loops.
14000         (expand_omp_sections): Fix default bb placement in loops.
14001         (expand_omp_atomic_pipeline): Properly add loops.
14003 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14005         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
14007 2013-04-29  Tom de Vries  <tom@codesourcery.com>
14009         * tree-ssa-tail-merge.c: Update header comment.
14011 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14013         * config/aarch64/arm_neon.h
14014         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
14015         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
14016         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
14017         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
14018         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
14019         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
14020         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
14021         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
14023 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14025         * config/aarch64/aarch64-simd.md
14026         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
14027         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
14028         fix_trunc, fixuns_trunc.
14029         (ftrunc<VDQF:mode>2): New.
14030         * config/aarch64/iterators.md (optab): Add fix, fixuns.
14031         (fix_trunc_optab): New.
14033 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14035         * config/aarch64/aarch64-builtins.c
14036         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
14037         iceilf, lround, iroundf.
14039 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
14041         PR target/54349
14042         * config/i386/i386.h (enum ix86_tune_indices)
14043         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
14044         New, split from X86_TUNE_INTER_UNIT_MOVES.
14045         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
14046         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
14047         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
14048         (TARGET_INTER_UNIT_MOVES): Remove.
14049         * config/i386/i386.c (initial_ix86_tune_features): Update.
14050         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
14051         (ix86_expand_convert_uns_didf_sse): Use
14052         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
14053         (ix86_expand_vector_init_one_nonzero): Ditto.
14054         (ix86_expand_vector_init_interleave): Ditto.
14055         (inline_secondary_memory_needed): Return true for moves from SSE class
14056         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
14057         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
14058         * config/i386/constraints.md (Yi, Ym): Depend on
14059         TARGET_INTER_UNIT_MOVES_TO_VEC.
14060         (Yj, Yn): New constraints.
14061         * config/i386/i386.md (*movdi_internal): Change constraints of
14062         operand 1 from Yi to Yj and from Ym to Yn.
14063         (*movsi_internal): Ditto.
14064         (*movdf_internal): Ditto.
14065         (*movsf_internal): Ditto.
14066         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
14067         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
14068         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
14069         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
14070         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
14071         * config/i386/sse.md (movdi_to_sse): Ditto.
14072         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
14073         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
14074         TARGET_INTER_UNIT_MOVES.
14075         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
14076         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
14077         instead of TARGET_INTER_UNIT_MOVES.
14078         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
14079         operand 1 from Yi to Yj and from Ym to Yn.
14081 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14083         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
14084         (float_truncate_hi_): Likewise.
14085         (float_extend_lo_): Likewise.
14086         (float_truncate_lo_): Likewise.
14087         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
14088         (aarch64_float_extend_lo_v2df): Likewise.
14089         (vec_unpacks_hi_v4sf): Likewise.
14090         (aarch64_float_truncate_lo_v2sf): Likewise.
14091         (aarch64_float_truncate_hi_v4sf): Likewise.
14092         (vec_pack_trunc_v2df): Likewise.
14093         (vec_pack_trunc_df): Likewise.
14095 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14097         * config/aarch64/aarch64-builtins.c
14098         (aarch64_fold_builtin): Fold float conversions.
14099         * config/aarch64/aarch64-simd-builtins.def
14100         (floatv2si, floatv4si, floatv2di): New.
14101         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
14102         * config/aarch64/aarch64-simd.md
14103         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
14104         * config/aarch64/iterators.md (FLOATUORS): New.
14105         (optab): Add float, floatuns.
14106         (su_optab): Likewise.
14108 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14110         * config/aarch64/aarch64-builtins.c
14111         (aarch64_builtin_vectorized_function): Use new names for
14112         fcvt builtins.
14113         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
14114         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
14115         (fcvtzu): Split as...
14116         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
14117         (fcvtas): Split as...
14118         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
14119         (fcvtau): Split as...
14120         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
14121         (fcvtps): Split as...
14122         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
14123         (fcvtpu): Split as...
14124         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
14125         (fcvtms): Split as...
14126         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
14127         (fcvtmu): Split as...
14128         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
14129         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
14130         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
14131         (lfrintnusf, lfrintnudf): Likewise.
14132         * config/aarch64/aarch64-simd.md
14133         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
14134         define_insn.
14135         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
14136         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
14137         (fcvt_pattern): Likewise.
14139 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14141         * config/aarch64/aarch64-simd.md
14142         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
14143         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
14145 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14147         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
14148         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
14149         (vrnd<a,m,n,p>_f32): Implement using builtins.
14150         (vrnd<i,x><q>_f<32, 64>): New.
14152 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14154         * config/aarch64/aarch64-builtins.c
14155         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
14156         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
14157         (frintz): Rename to...
14158         (btrunc): ...this.
14159         (frintp): Rename to...
14160         (ceil): ...this.
14161         (frintm): Rename to...
14162         (floor): ...this.
14163         (frinti): Rename to...
14164         (nearbyint): ...this.
14165         (frintx): Rename to...
14166         (rint): ...this.
14167         (frinta): Rename to...
14168         (round): ...this.
14169         * config/aarch64/aarch64-simd.md
14170         (aarch64_frint<frint_suffix><mode>): Delete.
14171         (<frint_pattern><mode>2): Convert to insn.
14172         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
14173         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
14174         (frint_pattern): Likewise.
14175         (frint_suffix): Likewise.
14177 2013-04-29  Richard Biener  <rguenther@suse.de>
14179         PR tree-optimization/57081
14180         * loop-init.c: Include tree-flow.h.
14181         (loop_optimizer_finalize): Free number of iteration estimates.
14182         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
14184 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
14186         PR tree-optimization/57083
14187         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
14188         non-singleton shift count range, zero extend low_bound for uns case.
14190         * config/i386/predicates.md (general_vector_operand): New predicate.
14191         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
14192         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
14193         if they aren't nonimmediate operands.  If their original values
14194         satisfy const_vector_equal_evenodd_p, don't shift them.
14195         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
14196         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
14197         if not nonimmediate_operand.
14198         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
14199         instead of register_operand.
14200         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
14202 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
14204         * stor-layout.c (finalize_size_functions): Allocate a structure and
14205         reset cfun before dumping the functions.
14207 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
14209         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
14211         PR target/56866
14212         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
14213         use xop_pmacsdqh if uns_p.
14214         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
14215         the immediate rotate count.
14217 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
14219         * rtl.h (struct rtx_def): Add comment for field jump.
14220         (LRA_SUBREG_P): New macro.
14221         * recog.c (register_operand): Check LRA_SUBREG_P.
14222         * lra.c (lra): Add note at the end of RTL code. Align non-empty
14223         stack frame.
14224         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
14225         (lra_final_code_change): Skip subreg change for operators.
14226         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
14227         if there are no operand changes.
14228         * lra-constraints.c (curr_insn_set): New.
14229         (match_reload): Set LRA_SUBREG_P.
14230         (emit_spill_move): Ditto.
14231         (check_and_process_move): Use curr_insn_set. Process only single
14232         set insns.  Don't initialize sec_mem_p and change_p.
14233         (simplify_operand_subreg): Use LRA_SUBREG_P.
14234         (reg_in_class_p): New function.
14235         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
14236         of #ifdef.  Add code to remove cycling.
14237         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
14238         non-null disp.  Reload inner instead of disp when base and index
14239         are null.  Try to put lo_sum into register.
14240         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
14241         (check_and_process_move): Move code for move cost check to
14242         simple_move_p.  Remove equiv_substitution.
14243         (simple_move_p): New function.
14244         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
14245         curr_insn_set.  Call check_and_process_move only for single set
14246         insns.  Use the new function.  Move call of check_and_process_move
14247         after operand equiv substitution and address process.
14249 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
14251         PR go/57045
14252         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
14253         with nonlocal goto receivers or returns twice calls, ignore
14254         unininitialized values from abnormal edges to nl goto receiver
14255         or returns twice call.
14257 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
14259         PR tree-optimization/57051
14260         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
14261         and VEC_RSHIFT_EXPR if shift count is a multiple of element
14262         bitsize.
14264 2013-04-26  Richard Biener  <rguenther@suse.de>
14266         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
14267         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
14268         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
14269         (fixup_loop_arrays_after_move): New function.
14270         (move_sese_region_to_fn): Properly outline the loop tree parts
14271         of the SESE region.
14273 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
14275         * config/i386/i386.md (type, unit): Fix long lines.
14277 2013-04-26  Richard Biener  <rguenther@suse.de>
14279         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
14280         (lto-streamer-out.o): Likewise.
14281         * cfgloop.c (init_loops_structure): Export, add struct function
14282         argument and adjust.
14283         (flow_loops_find): Adjust.
14284         * cfgloop.h (enum loop_estimation): Add EST_LAST.
14285         (init_loops_structure): Declare.
14286         * lto-streamer-in.c: Include cfgloop.h.
14287         (input_cfg): Input the loop tree.
14288         * lto-streamer-out.c: Include cfgloop.h.
14289         (output_cfg): Output the loop tree.
14290         (output_struct_function_base): Do not drop PROP_loops.
14292 2013-03-26  Richard Biener  <rguenther@suse.de>
14294         * tree-cfg.c (execute_build_cfg): Build the loop tree.
14295         (pass_build_cfg): Provide PROP_loops.
14296         (move_sese_region_to_fn): Remove loops that are outlined into fn
14297         for now.
14298         * tree-inline.c: Include cfgloop.h.
14299         (initialize_cfun): Do not drop PROP_loops.
14300         (copy_loops): New function.
14301         (copy_cfg_body): Copy loop structure.
14302         (tree_function_versioning): Initialize destination loop tree.
14303         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
14304         (pass_parallelize_loops): Do IL verification.
14305         * loop-init.c (loop_optimizer_init): Fixup loops if required.
14306         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
14307         the CFG make sure we fixup loops as well.
14308         * tree-ssa-tail-merge.c: Include cfgloop.h.
14309         (replace_block_by): When merging loop latches mark loops for fixup.
14310         * lto-streamer-out.c (output_struct_function_base): Drop
14311         PROP_loops for now.
14312         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
14313         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
14314         * ipa-split.c: Include cfgloop.h.
14315         (split_function): Add the new return block to the loop tree root.
14316         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
14317         whether we have removed the forwarder block.
14318         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
14319         * cfgloop.h (place_new_loop): Declare.
14320         * cfgloopmanip.c (place_new_loop): Export.
14321         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
14322         (tree-switch-conversion.o): Likewise.
14323         (tree-complex.o): Likewise.
14324         (tree-inline.o): Likewise.
14325         (tree-ssa-tailmerge.o): Likewise.
14326         (ipa-split.o): Likewise.
14327         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
14328         (tree-ssa-copy.o): Likewise.
14329         * tree-switch-conversion.c: Include cfgloop.h
14330         (process_switch): If we emit a bit-test cascade, schedule loops
14331         for fixup.
14332         * tree-complex.c: Include cfgloop.h.
14333         (expand_complex_div_wide): Properly add new basic-blocks to loops.
14334         * asan.c: Include cfgloop.h.
14335         (create_cond_insert_point): Properly add new basic-blocks to
14336         loops, schedule loop fixup.
14337         * cfgloop.c (verify_loop_structure): Check that looks are not
14338         marked for fixup.
14339         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
14340         to loops.
14341         (expand_omp_for_generic): Likewise.
14342         (expand_omp_sections): Likewise.
14343         (expand_omp_atomic_pipeline): Schedule loops for fixup.
14344         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
14345         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
14346         is initialized, not when loops are present.
14347         * tree-parloops.c (parallelize_loops): Remove checking here.
14348         * passes.c (init_optimization_passes): Schedule a copy-propagation
14349         pass before complete unrolling of inner loops.
14351 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
14353         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
14354         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
14355         (colorize_init): Add argument to _WIN32 version.
14356         * toplev.c: Include diagnostic-color.h.
14357         (process_options): Default to -fdiagnostics-color=auto if
14358         GCC_COLORS env var is in the environment.
14359         * common.opt (fdiagnostics-color=): Add Var and Init.
14360         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
14361         env var is in the environment, the default is auto rather than never.
14363         * diagnostic.h (file_name_as_prefix): Add context argument.
14364         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
14365         the string as locus.
14366         * langhooks.c (lhd_print_error_function): Adjust caller.
14368 2013-04-25  Lawrence Crowl  <crowl@google.com>
14370         * var-tracking.c (shared_hash_def::htab):
14371         Change type to hash_table.  Update dependent calls and types.
14373 2013-04-25  Lawrence Crowl  <crowl@google.com>
14375         * Makefile.in: Update as needed below.
14377         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
14378         Move declaration to after the type's method definitons.
14380         * attribs.c (htab_t scoped_attributes::attribute_hash):
14381         Change type to hash_table.  Update dependent calls and types.
14383         * bitmap.c (htab_t bitmap_desc_hash):
14384         Change type to hash_table.  Update dependent calls and types.
14386         * cselib.c (htab_t cselib_hash_table):
14387         Change type to hash_table.  Update dependent calls and types.
14389         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
14390         (hash_string_slot_node): Move implementation into lto-streamer.h
14391         struct string_slot_hasher.
14392         (eq_string_slot_node): Likewise.
14394         * data-streamer-out.c: Update output_block::string_hash_table
14395         dependent calls and types.
14397         * dwarf2cfi.c (htab_t trace_index):
14398         Change type to hash_table.  Update dependent calls and types.
14400         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
14401         Change type to hash_table.  Update dependent calls and types.
14402         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
14403         (htab_t optimize_external_refs::map): Likewise.
14404         (htab_t output_comp_unit::extern_map): Likewise.
14405         (htab_t output_comdat_type_unit::extern_map): Likewise.
14406         (htab_t output_macinfo::macinfo_htab): Likewise.
14407         (htab_t optimize_location_lists::htab): Likewise.
14408         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
14410         * except.c (htab_t ehspec_hash_type):
14411         Change type to hash_table.  Update dependent calls and types.
14412         (assign_filter_values::ttypes): Likewise.
14413         (assign_filter_values::ehspec): Likewise.
14414         (sjlj_assign_call_site_values::ar_hash): Likewise.
14415         (convert_to_eh_region_ranges::ar_hash): Likewise.
14417         * gcse.c (htab_t pre_ldst_table):
14418         Change type to hash_table.  Update dependent calls and types.
14420         * ggc-common.c (htab_t saving_htab):
14421         Change type to hash_table.  Update dependent calls and types.
14422         (htab_t loc_hash): Likewise.
14423         (htab_t ptr_hash): Likewise.
14424         (call_count): Rename ggc_call_count.
14425         (call_alloc): Rename ggc_call_alloc.
14426         (loc_descriptor): Rename make_loc_descriptor.
14427         (add_statistics): Rename ggc_add_statistics.
14429         * ggc-common.c (saving_htab):
14430         Change type to hash_table.  Update dependent calls and types.
14432         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
14433         (push_gimplify_context): Likewise.
14434         (pop_gimplify_context): Likewise.
14435         (struct gimple_temp_hash_elt): Added.
14436         (struct gimplify_hasher): Likewise.
14437         (struct gimplify_ctx.temp_htab):
14438         Change type to hash_table.  Update dependent calls and types.
14440         * gimple-fold.c: Include gimplify-ctx.h.
14442         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
14443         Change type to hash_table.  Update dependent calls and types.
14444         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
14445         avoid potential global name collision.
14447         * gimplify.c: Include gimplify-ctx.h.
14448         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
14449         (htab_t gimplify_ctx::temp_htab):
14450         Update dependent calls and types for new type hash_table.
14451         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
14452         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
14454         * gimplify-ctx.h: New.
14455         (struct gimple_temp_hash_elt): Move from gimplify.c.
14456         (class gimplify_hasher): New.
14457         (struct gimplify_ctx): Move from gimple.h.
14458         (htab_t gimplify_ctx::temp_htab):
14459         Change type to hash_table.  Update dependent calls and types.
14461         * graphite-clast-to-gimple.c: Include graphite-htab.h.
14462         (htab_t ivs_params::newivs_index):
14463         Change type to hash_table.  Update dependent calls and types.
14464         (htab_t ivs_params::params_index): Likewise.
14465         (htab_t print_generated_program::params_index): Likewise.
14466         (htab_t gloog::newivs_index): Likewise.
14467         (htab_t gloog::params_index): Likewise.
14469         * graphite.c: Include graphite-htab.h.
14470         4htab_t graphite_transform_loops::bb_pbb_mapping):
14471         Change type to hash_table.  Update dependent calls and types.
14473         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
14474         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
14475         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
14477         * graphite-dependences.c: Include graphite-htab.h.
14478         (loop_is_parallel_p): Change hash table type of parameter.
14480         * graphite-htab.h: New.
14481         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
14482         (extern find_pbb_via_hash): Move from graphite-poly.h.
14483         (extern loop_is_parallel_p): Move from graphite-poly.h.
14484         (extern get_loop_body_pbbs): Move from graphite-poly.h.
14486         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
14487         (extern loop_is_parallel_p): Move to graphite-htab.h.
14488         (extern get_loop_body_pbbs): Move to graphite-htab.h.
14490         * haifa-sched.c (htab_t delay_htab):
14491         Change type to hash_table.  Update dependent calls and types.
14492         (htab_t delay_htab_i2): Likewise.
14494         * ira-color.c (htab_t allocno_hard_regs_htab):
14495         Change type to hash_table.  Update dependent calls and types.
14497         * ira-costs.c (htab_t cost_classes_htab):
14498         Change type to hash_table.  Update dependent calls and types.
14500         * loop-invariant.c (htab_t merge_identical_invariants::eq):
14501         Change type to hash_table.  Update dependent calls and types.
14503         * loop-iv.c (htab_t bivs):
14504         Change type to hash_table.  Update dependent calls and types.
14506         * loop-unroll.c (htab_t opt_info::insns_to_split):
14507         Change type to hash_table.  Update dependent calls and types.
14508         (htab_t opt_info::insns_with_var_to_expand): Likewise.
14510         * lto-streamer.h (struct string_slot): Move from data-streamer.h
14511         (struct string_slot_hasher): New.
14512         (htab_t output_block::string_hash_table):
14513         Change type to hash_table.  Update dependent calls and types.
14515         * lto-streamer-in.c (freeing_string_slot_hasher): New.
14516         (htab_t file_name_hash_table):
14517         Change type to hash_table.  Update dependent calls and types.
14519         * lto-streamer-out.c: Update output_block::string_hash_table dependent
14520         calls and types.
14522         * lto-streamer.c (htab_t tree_htab):
14523         Change type to hash_table.  Update dependent calls and types.
14525         * omp-low.c: Include gimplify-ctx.h.
14527         * passes.c (htab_t name_to_pass_map):
14528         Change type to hash_table.  Update dependent calls and types.
14529         (pass_traverse): Rename to passes_pass_traverse.
14531         * plugin.c (htab_t event_tab):
14532         Change type to hash_table.  Update dependent calls and types.
14534         * postreload-gcse.c (htab_t expr_table):
14535         Change type to hash_table.  Update dependent calls and types.
14536         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
14538         * sese.c (debug_rename_map_1): Make extern.
14539         (htab_t copy_bb_and_scalar_dependences::rename_map):
14540         Change type to hash_table.  Update dependent calls and types.
14542         * sese.h (extern debug_rename_map): Move to .c file.
14544         * store-motion.c (htab_t store_motion_mems_table):
14545         Change type to hash_table.  Update dependent calls and types.
14547         * trans-mem.c (htab_t tm_new_mem_hash):
14548         Change type to hash_table.  Update dependent calls and types.
14550         * tree-browser.c (htab_t TB_up_ht):
14551         Change type to hash_table.  Update dependent calls and types.
14553         * tree-cfg.c (htab_t discriminator_per_locus):
14554         Change type to hash_table.  Update dependent calls and types.
14556         * tree-complex.c: Include tree-hasher.h
14557         (htab_t complex_variable_components):
14558         Change type to hash_table.  Update dependent calls and types.
14560         * tree-eh.c (htab_t finally_tree):
14561         Change type to hash_table.  Update dependent calls and types.
14563         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
14564         struct int_tree_hasher.
14565         (extern int_tree_map_eq): Likewise.
14566         (uid_decl_map_hash): Removed.
14567         (extern decl_tree_map_eq): Likewise.
14569         * tree-hasher.h: New.
14570         (struct int_tree_hasher): New.
14571         (typedef int_tree_htab_type): New.
14573         * tree-inline.c: Include gimplify-ctx.h.
14575         * tree-mudflap.c: Include gimplify-ctx.h.
14577         * tree-parloops.c: Include tree-hasher.h.
14578         (htab_t eliminate_local_variables_stmt::decl_address):
14579         Change type to hash_table.  Update dependent calls and types.
14580         (htab_t separate_decls_in_region::decl_copies): Likewise.
14582         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
14583         Change type to hash_table.  Update dependent calls and types.
14585         * tree-sra.c (candidates):
14586         Change type to hash_table.  Update dependent calls and types.
14588         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
14589         in tree-flow.h.
14590         (int_tree_map_hash): Likewise.
14592         * tree-ssa-dom.c (htab_t avail_exprs):
14593         Change type to hash_table.  Update dependent calls and types.
14595         * tree-ssa-live.c (var_map_base_init::tree_to_index):
14596         Change type to hash_table.  Update dependent calls and types.
14598         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
14599         Change type to hash_table.  Update dependent calls and types.
14601         * tree-ssa-phiopt.c (seen_ssa_names):
14602         Change type to hash_table.  Update dependent calls and types.
14604         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
14605         Change type to hash_table.  Update dependent calls and types.
14607         * tree-ssa-uncprop.c (equiv):
14608         Change type to hash_table.  Update dependent calls and types.
14610 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
14612         PR rtl-optimization/57003
14613         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
14614         call note_stores with kill_clobbered_value callback again after
14615         killing regs_invalidated_by_call.
14617 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
14619         * config/aarch64/aarch64-simd.md
14620         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
14621         (aarch64_simd_bsl<mode>): Likewise.
14622         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
14624 2013-04-25  Marek Polacek  <polacek@redhat.com>
14626         PR tree-optimization/57066
14627         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
14629 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
14631         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
14633 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
14635         * config/aarch64/aarch64-builtins.c
14636         (aarch64_fold_builtin): New.
14637         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
14638         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
14639         * config/aarch64/aarch64-simd-builtins.def (abs): New.
14640         * config/aarch64/arm_neon.h
14641         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
14643 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
14644             Tejas Belagod  <tejas.belagod@arm.com>
14646         * config/aarch64/aarch64-builtins.c
14647         (aarch64_gimple_fold_builtin): New.
14648         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
14649         * config/aarch64/aarch64-simd-builtins.def (addv): New.
14650         * config/aarch64/aarch64-simd.md (addpv4sf): New.
14651         (addvv4sf): Update.
14652         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
14654 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14656         * config/aarch64/aarch64.md
14657         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
14659 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14661         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
14662         (*ngcsi_uxtw): New pattern.
14664 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14665             Julian Brown  <julian@codesourcery.com>
14667         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
14668         (TB_DREG): Add T_V4HF.
14669         (v4hf_UP): New macro.
14670         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
14671         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
14672         Handle initialisation of V4HF. Adjust initialisation of reinterpret
14673         built-ins.
14674         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
14675         (arm_vector_mode_supported_p): Handle V4HF.
14676         (arm_mangle_map): Handle V4HFmode.
14677         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
14678         * config/arm/arm_neon_builtins.def: Add entries for
14679         vcvtv4hfv4sf, vcvtv4sfv4hf.
14680         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
14681         (neon_vcvtv4hfv4sf): Likewise.
14682         * config/arm/neon-gen.ml: Handle half-precision floating point
14683         features.
14684         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
14685         * config/arm/arm_neon.h: Regenerate.
14686         * config/arm/neon.ml (type elts): Add F16.
14687         (type vectype): Add T_float16x4, T_floatHF.
14688         (type vecmode): Add V4HF.
14689         (type features): Add Requires_FP_bit feature.
14690         (elt_width): Handle F16.
14691         (elt_class): Likewise.
14692         (elt_of_class_width): Likewise.
14693         (mode_of_elt): Refactor.
14694         (type_for_elt): Handle F16, fix error messages.
14695         (vectype_size): Handle T_float16x4.
14696         (vcvt_sh): New function.
14697         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
14698         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
14699         (string_of_mode): Handle V4HF.
14700         * doc/arm-neon-intrinsics.texi: Regenerate.
14702 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
14704         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
14705         format specifier in 'X' case.
14707 2013-04-25  Alan Modra  <amodra@gmail.com>
14709         PR target/57052
14710         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
14711         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
14712         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
14713         Repeat for many other rotate/shift and mask patterns using subregs.
14714         Name lshiftrt insns.
14715         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
14716         on WORDS_BIG_ENDIAN.
14718 2013-04-25  Alan Modra  <amodra@gmail.com>
14720         * config.gcc: Support little-endian powerpc-linux targets.
14721         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
14722         (LINK_OS_LINUX_SPEC): Define.
14723         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
14724         Preserve MASK_LITTLE_ENDIAN.
14725         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
14726         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
14727         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
14728         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
14729         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
14730         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
14731         Correct fp word order for little-endian.  Don't shift toc entries
14732         smaller than a word for little-endian.
14733         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
14734         (bswapdi2 splits): Correct low-part subreg for little-endian.
14735         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
14736         low/high where such is correct only for be.
14737         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
14738         little-endian for -mcall-aixdesc.
14740 2013-04-25  Alan Modra  <amodra@gmail.com>
14742         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
14743         replace_equiv_address_nv.
14745 2013-04-25  Alan Modra  <amodra@gmail.com>
14747         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
14749 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
14751         Revert:
14752         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
14753         * rtl.h (struct rtx_def): ...
14755 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
14757         PR rtl-optimizations/57046
14758         * lra-constraints (split_reg): Set up lra_risky_transformations_p
14759         for multi-reg splits.
14761 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
14763         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
14765 2013-04-24  Sterling Augustine  <saugustine@google.com>
14767         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
14768         (comp_dir_string, debug_str_dwo_section): New.
14769         (DEBUG_STR_DWO_SECTION): Rename to ...
14770         (DEBUG_DWO_STR_SECTION): ... this.
14771         (DEBUG_NORM_STR_SECTION): Delete.
14772         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
14773         (DEBUG_STR_DWO_SECTION_FLAGS): New.
14774         (find_AT_string): Move most logic to ...
14775         (find_AT_string_in_table): ... here.  New.
14776         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
14777         add_skeleton_AT_string.  Delete logic.
14778         (output_skeleton_debug_sections): Remove call to
14779         add_top_level_skeleton_die_attrs.
14780         (add_comp_dir_attribute): Move logic to comp_dir_string.
14781         (dwarf2out_init): Initialize debug_str_dwo_section.
14782         (output_indirect_string): Call find_string_form.
14783         (output_indirect_strings): Rewrite.
14784         (prune_unused_types): Empty skeleton_debug_str_hash.
14785         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
14786         (dwarf2out_finish):  Call output_indirect_strings.
14788 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
14790         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
14792 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
14794         * rtl.h (struct rtx_def): Add comment for field jump.
14795         (LRA_SUBREG_P): New macro.
14796         * recog.c (register_operand): Check LRA_SUBREG_P.
14797         * lra.c (lra): Add note at the end of RTL code. Align non-empty
14798         stack frame.
14799         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
14800         (lra_final_code_change): Skip subreg change for operators.
14801         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
14802         if there are no operand changes.
14803         * lra-constraints.c (curr_insn_set): New.
14804         (match_reload): Set LRA_SUBREG_P.
14805         (emit_spill_move): Ditto.
14806         (check_and_process_move): Use curr_insn_set. Process only single
14807         set insns.  Don't initialize sec_mem_p and change_p.
14808         (simplify_operand_subreg): Use LRA_SUBREG_P.
14809         (reg_in_class_p): New function.
14810         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
14811         of #ifdef.  Add code to remove cycling.
14812         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
14813         non-null disp.  Reload inner instead of disp when base and index
14814         are null.  Try to put lo_sum into register.
14815         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
14816         (check_and_process_move): Move code for move cost check to
14817         simple_move_p.  Remove equiv_substitution.
14818         (simple_move_p): New function.
14819         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
14820         curr_insn_set.  Call check_and_process_move only for single set
14821         insns.  Use the new function.  Move call of check_and_process_move
14822         after operand equiv substitution and address process.
14824 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
14826         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
14827         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
14828         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
14830 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
14832         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
14834 2013-04-24  Marek Polacek  <polacek@redhat.com>
14836         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
14837         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
14838         (select_loops_exit_conditions): Likewise.
14839         (number_of_iterations_for_all_loops): Likewise.
14840         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
14841         (scev_analysis): Likewise.
14843 2013-04-02  Catherine Moore  <clm@codesourcery.com>
14844             Chao-ying Fu  <fu@mips.com>
14846         * config/mips/micromips.md (jraddiusp): New pattern.
14847         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
14848         instruction if possible.
14850 2013-04-24  Alan Modra  <amodra@gmail.com>
14852         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
14854 2013-04-24  Julian Brown  <julian@codesourcery.com>
14855             Chung-Lin Tang  <cltang@codesourcery.com>
14857         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
14858         dependency behavior in enumeration type DIE generation. Add TODO note
14859         to comments about future DW_FORM_sdata/udata re-work of related code.
14861 2013-04-23  Lawrence Crowl  <crowl@google.com>
14863         * Makefile.in: Update as needed below.
14865         * hash-table.h (class hash_table):
14866         Correct many methods with parameter types compare_type to the correct
14867         value_type.  (Correct code was unlikely to notice the change.)
14868         (hash_table::elements_with_deleted) New.
14869         (class hashtable::iterator): New.
14870         (hashtable::begin()): New.
14871         (hashtable::end()): New.
14872         (FOR_EACH_HASH_TABLE_ELEMENT): New.
14874         * statistics.c (statistics_hashes):
14875         Change type to hash_table.  Update dependent calls and types.
14877         * tree-into-ssa.c (var_infos):
14878         Change type to hash_table.  Update dependent calls and types.
14880         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
14881         Change type to hash_table.  Update dependent calls and types.
14883         * tree-ssa-loop-im.c (struct mem_ref.refs):
14884         Change type to hash_table.  Update dependent calls and types.
14886         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
14887         Change type to hash_table.  Update dependent calls and types.
14889         * tree-ssa-sccvn.c (vn_tables_s::nary):
14890         Change type to hash_table.  Update dependent calls and types.
14891         (vn_tables_s::phis): Likewise.
14892         (vn_tables_s::references): Likewise.
14894         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
14895         (vn_reference_eq): Update parameter and return types.
14897         * tree-ssa-structalias.c (pointer_equiv_class_table):
14898         Change type to hash_table.  Update dependent calls and types.
14899         (location_equiv_class_table): Likewise.
14901         * tree-vect-data-refs.c: Consequential changes for making
14902         peeling a hash_table.
14904         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
14905         (destroy_loop_vec_info): Dependent hash_table update.
14907         * tree-vectorizer.h (peeling_htab):
14908         Change type to hash_table.  Update dependent calls and types.
14910 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
14912         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
14913         to check the register content is equal or not.
14914         * lra-constraints.c (match_reload): Use lra_assign_reg_val
14915         to assign register content record.
14916         * lra-eliminations.c (update_reg_eliminate): Use
14917         lra_update_reg_val_offset to update register content offset.
14918         * lra-int.h (struct lra_reg): Add offset member.
14919         (lra_reg_val_equal_p): New static inline function.
14920         (lra_update_reg_val_offset): New static inline function.
14921         (lra_assign_reg_val): New static inline function.
14922         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
14923         to assign register content record.
14924         (initialize_lra_reg_info_element): Initial offset to zero.
14926 2013-04-23  Catherine Moore  <clm@codesourcery.com>
14928         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
14929         operands.  Record compression.
14931 2013-04-23  Xinliang David Li  <davidxl@google.com>
14933         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
14935 2013-04-23  Richard Biener  <rguenther@suse.de>
14937         PR middle-end/57036
14938         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
14939         parameter, only add abnormal goto edges from the copied body
14940         if the call could perform abnormal gotos.
14941         (copy_cfg_body): Adjust.
14943 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
14945         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
14947 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
14949         * coretypes.h (gimple_stmt_iterator): Add struct to make
14950         compatible with C.
14952 2013-04-23  Richard Biener  <rguenther@suse.de>
14954         PR tree-optimization/57026
14955         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
14956         from SSA names occuring in abnormal PHI nodes.
14958 2013-04-22  Andi Kleen  <ak@linux.intel.com>
14960         * lto/lto.c (print_lto_report_1): Fix LTO report names.
14962 2013-04-22  Andi Kleen  <ak@linux.intel.com>
14964         * lto/lto.c (print_lto_report_1): Declare early.
14965         (read_cgraph_and_symbols): Call print_lto_report_1 early.
14967 2013-04-22  Andi Kleen  <ak@linux.intel.com>
14969         * common.opt (-flto-report-wpa): Add.
14970         * doc/invoke.texi (-flto-report-wpa): Add.
14971         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
14972         (lto_main): dito.
14974 2013-04-22  Xinliang David Li  <davidxl@google.com>
14976         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
14977         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
14978         * Makefile.in: New dependency
14980             David Daney <ddaney.cavm@gmail.com>
14982         * configure.ac (gcc_cv_as_micromips_support):  Use the
14983         --fatal-warnings option.
14984         * configure: Regenerate.
14986 2013-04-22  Marek Polacek  <polacek@redhat.com>
14988         PR sanitizer/56990
14989         * tsan.c (instrument_expr): Don't instrument expression
14990         in case its size is zero.
14992 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
14994         PR target/57032
14995         Revert:
14996         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
14998         * config/alpha/alpha.c (TARGET_LRA_P): New define.
15000 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
15002         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
15003         (gimple_stmt_iterator): New typedef.
15004         * gimple.h (gimple_stmt_iterator): Rename to...
15005         (gimple_stmt_iterator_d): ... This.
15006         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
15007         trees be valid for GIMPLE and GENERIC.
15008         (TARGET_GIMPLE_FOLD_BUILTIN): New.
15009         * gimple-fold.c (gimple_fold_call): Call target hook
15010         gimple_fold_builtin.
15011         * hooks.c (hook_bool_gsiptr_false): New.
15012         * hooks.h (hook_bool_gsiptr_false): New.
15013         * target.def (fold_stmt): New.
15014         * doc/tm.texi: Regenerate.
15016 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
15018         PR target/57018
15019         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
15020         a set sp if no stack realignment.
15022 2013-04-22  Nick Clifton  <nickc@redhat.com>
15024         * config.gcc (tilegx-linux): Extend extra_objs rather than
15025         overwriting it.
15026         (tilepro-linux): Likewise.
15028 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
15030         * config/aarch64/aarch64-builtins.c
15031         (CF): Remove.
15032         (CF0, CF1, CF2, CF3, CF4, CF10): New.
15033         (VAR<1-12>): Add MAP parameter.
15034         (BUILTIN_*): Likewise.
15035         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
15036         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
15037         (aarch64_ushl_n<mode>): Likewise.
15038         (aarch64_sshr_n<mode>): Likewise.
15039         (aarch64_ushr_n<mode>): Likewise.
15040         (aarch64_<maxmin><mode>): Likewise.
15041         (aarch64_sqrt<mode>): Likewise.
15042         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
15043         (vshr<q>_n_*): Likewise.
15045 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
15047         * config/aarch64/aarch64-builtins.c
15048         (aarch64_simd_builtin_type_mode): Handle SF types.
15049         (sf_UP): Define.
15050         (BUILTIN_GPF): Define.
15051         (aarch64_init_simd_builtins): Handle SF types.
15052         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
15053         (frecps): Likewise.
15054         (frecpx): Likewise.
15055         * config/aarch64/aarch64-simd.md
15056         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
15057         (aarch64_frecpe<mode>): New.
15058         (aarch64_frecps<mode>): Likewise.
15059         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
15060         (v8type): Add frecp<esx>.
15061         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
15062         (aarch64_frecps<mode>): Likewise.
15063         * config/aarch64/iterators.md (FRECP): New.
15064         (frecp_suffix): Likewise.
15065         * config/aarch64/arm_neon.h
15066         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
15068 2013-04-22  Christian Bruel  <christian.bruel@st.com>
15070         PR target/56995
15071         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
15072         (REG_CLASS_NAMES): Idem.
15073         (REG_CLASS_CONTENTS): Idem.
15074         (REGCLASS_HAS_FP_REG): Idem.
15075         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
15076         (sh_conditional_register_usage): Idem.
15078 2013-04-21  Jeff Law  <law@redhat.com>
15080         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
15081         (ssa_forward_propagate_and_combine): Use it.
15083 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
15085         * lra.c: Update the flow chart diagram.
15087 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
15089         PR rtl-optimization/56847
15090         * lra-constraints.c (process_alt_operands): Discourage alternative
15091         with non-matche doffsettable memory constraint fro memory with
15092         known offset.
15094 2013-04-19  Richard Biener  <rguenther@suse.de>
15096         PR tree-optimization/56982
15097         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
15098         function.
15099         * gimplify.c (gimplify_call_expr): Notice special calls.
15100         (gimplify_modify_expr): Likewise.
15101         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
15102         abnormal control flow receivers.
15103         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
15104         in the same way as cfun->has_nonlocal_labels.
15105         (gimple_purge_dead_abnormal_call_edges): Likewise.
15106         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
15107         receivers start a basic-block.
15109 2013-04-19  Richard Biener  <rguenther@suse.de>
15111         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
15112         member ...
15113         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
15114         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
15115         (SLP_TREE_LOAD_PERMUTATION): Add.
15116         (vect_transform_slp_perm_load): Adjust prototype.
15117         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
15118         (vect_free_slp_instance): Likewise.
15119         (vect_create_new_slp_node): Likewise.
15120         (vect_supported_slp_permutation_p): Remove.
15121         (vect_slp_rearrange_stmts): Adjust.
15122         (vect_supported_load_permutation_p): Likewise.  Inline
15123         vect_supported_slp_permutation_p here.
15124         (vect_analyze_slp_instance): Compute load permutations per
15125         slp node instead of per instance.
15126         (vect_get_slp_defs): Adjust.
15127         (vect_transform_slp_perm_load): Likewise.
15128         (vect_schedule_slp_instance): Remove redundant code.
15129         (vect_schedule_slp): Remove hack for PR56270, add it ...
15130         * tree-vect-stmts.c (vectorizable_load): ... here, do not
15131         CSE loads for SLP.  Adjust.
15133 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
15135         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
15136         spelling in two comments.
15138 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
15140         PR target/56797
15141         * config/arm/arm.c (load_multiple_sequence): Require SP
15142         as base register for loads if SP is in the register list.
15144 2013-04-19  Martin Jambor  <mjambor@suse.cz>
15146         PR tree-optimization/56718
15147         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
15148         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
15149         and made public.  Adjusted all callers.
15150         (ipa_intraprocedural_devirtualization): New function.
15151         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
15152         (ipa_intraprocedural_devirtualization): Likewise.
15153         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
15155 2013-04-19  Richard Biener  <rguenther@suse.de>
15157         PR tree-optimization/57000
15158         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
15160 2013-04-19  Terry Guo  <terry.guo@arm.com>
15162         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
15163         Replace with ...
15164         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
15165         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
15166         (cortex_m4_fmacs): Use new reservations.
15167         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
15169 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
15171         PR rtl-optimization/56999
15172         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
15173         related code.
15174         (lra_coalesce): Remove split_origin_bitmap and related code.
15175         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
15176         ranges if necessary.
15178 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
15180         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
15181         New array.
15182         (ix86_expand_call): Remove clobbered_registers array and use
15183         x86_64_ms_sysv_extra_clobbered_registers instead.
15184         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
15185         Declare here.
15186         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
15187         predicate.
15188         * config/i386/i386.md (*call_rex64_ms_sysv): Use
15189         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
15190         (*call_value_rex64_ms_sysv): Ditto.
15192 2013-04-18  Cary Coutant  <ccoutant@google.com>
15194         * dwarf2out.c (output_pubnames): Check die_perennial_p of
15195         parent instead of die_mark.
15197 2013-04-18  Diego Novillo  <dnovillo@google.com>
15199         * gimple.c (create_gimple_tmp): New.
15200         (get_expr_type): New.
15201         (build_assign): New.
15202         (build_type_cast): New.
15203         * gimple.h (enum ssa_mode): Define.
15204         (gimple_seq_set_location): New.
15205         * asan.c (build_check_stmt): Change some gimple_build_* calls
15206         to use build_assign and build_type_cast.
15208 2013-04-18  Richard Biener  <rguenther@suse.de>
15210         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
15211         handle negative step.  Remove redundant checks.
15212         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
15213         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
15214         for negative step and grouped loads fail to vectorize.
15216 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
15218         * emit-rtl.c (reset_insn_used_flags): New function.
15219         (reset_all_used_flags): Use it.
15220         (verify_insn_sharing): New function.
15221         (verify_rtl_sharing): Fix verification for SEQUENCEs.
15223 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
15225         PR tree-optimization/56984
15226         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
15227         and (x >> M) >= N don't register any assertion if N << M is the
15228         minimum value.
15230 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
15232         * lower-subreg.c (resolve_simple_move): If called self-recursive,
15233         do not delete_insn insns that have not yet been emitted, only
15234         unlink them with remove_insn.
15235         * df-scan.c (df_insn_delete): Revert r197492.
15237 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
15239         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
15240         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
15242 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
15244         * config/arm/arm.md (movsicc_insn): Convert define_insn into
15245         define_insn_and_split.
15246         (and_scc,ior_scc,negscc): Likewise.
15247         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
15249 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
15251         * config/arm/arm.c (use_return_insn): Return 0 for targets that
15252         can benefit from using a sequence of LDRD instructions in epilogue
15253         instead of a single LDM instruction.
15255 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
15257         PR 45688
15258         * doc/extend.texi: Fix typo.
15260 2013-04-17  Richard Biener  <rguenther@suse.de>
15262         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
15263         (vect_build_slp_tree): ... here.
15264         (vect_build_slp_tree_1): Compute which stmts of the SLP group
15265         match.  Remove special-casing of mismatched complex loads.
15266         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
15267         re-try the match with swapped commutative operands.
15268         (vect_supported_load_permutation_p): Remove special-casing of
15269         mismatched complex loads.
15270         (vect_analyze_slp_instance): Adjust.
15272 2013-04-17  Richard Biener  <rguenther@suse.de>
15274         PR rtl-optimization/56921
15275         * cfgloop.h (struct loop): Add simple_loop_desc member.
15276         (struct niter_desc): Mark with GTY(()).
15277         (simple_loop_desc): Do not use aux field but simple_loop_desc.
15278         * loop-iv.c (get_simple_loop_desc): Likewise.
15279         (free_simple_loop_desc): Likewise.
15281         Revert
15282         2013-04-16  Richard Biener  <rguenther@suse.de>
15284         PR rtl-optimization/56921
15285         * loop-init.c (pass_rtl_move_loop_invariants): Add
15286         TODO_do_not_ggc_collect to todo_flags_finish.
15287         (pass_rtl_unswitch): Same.
15288         (pass_rtl_unroll_and_peel_loops): Same.
15289         (pass_rtl_doloop): Same.
15291 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
15293         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
15294         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
15295         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
15296         references.
15297         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
15298         * tree-streamer.c (record_common_node): Adjust reference in comment.
15300 2013-04-17  Terry Guo  <terry.guo@arm.com>
15302         * config/arm/cortex-m4.md: Add a new bypass.
15304 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
15306         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
15307         New pattern.
15308         (*subs_<optab><mode>_multp2): New pattern.
15309         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
15310         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
15312 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
15314         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
15315         (*subs_mul_imm_<mode>): New pattern.
15317 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
15319         PR target/56948
15320         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
15321         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
15322         (vsx_movti_32bit): Change j->wa to O->wa.
15324 2013-04-16  Richard Biener  <rguenther@suse.de>
15326         PR rtl-optimization/56921
15327         * loop-init.c (pass_rtl_move_loop_invariants): Add
15328         TODO_do_not_ggc_collect to todo_flags_finish.
15329         (pass_rtl_unswitch): Same.
15330         (pass_rtl_unroll_and_peel_loops): Same.
15331         (pass_rtl_doloop): Same.
15333 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
15335         * config/arm/arm.c (emit_multi_reg_push): New declaration
15336         for an existing function.
15337         (arm_emit_strd_push): New function.
15338         (arm_expand_prologue): Used here.
15339         (arm_emit_ldrd_pop): New function.
15340         (arm_expand_epilogue): Used here.
15341         (arm_get_frame_offsets): Update condition.
15342         (arm_emit_multi_reg_pop): Add a special case for load of a single
15343         register with writeback.
15345 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
15347         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
15348         description.
15350 2013-04-16  Richard Biener  <rguenther@suse.de>
15352         PR tree-optimization/56756
15353         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
15354         (first_mem_ref_loc): New.
15355         (execute_sm): Place the load temporarily before a previous
15356         access instead of in the latch edge to ensure its SSA dependencies
15357         are defined at points dominating the load.
15359 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
15361         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
15362         correct fix by moving header and footer insn to the footer of
15363         the merged basic block.  Clear BB_END of the merged-away block.
15365         PR middle-end/43631
15366         * emit-rtl.c (make_note_raw): New function.
15367         (link_insn_into_chain): New static inline function.
15368         (add_insn): Use it.
15369         (add_insn_before, add_insn_after): Factor insn chain linking code...
15370         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
15371         using link_insn_into_chain.
15372         (note_outside_basic_block_p): New helper function for emit_note_after
15373         and emit_note_before.
15374         (emit_note_after): Use nobb variant of add_insn_after if the note
15375         should not be contained in a basic block.
15376         (emit_note_before): Use nobb variant of add_insn_before if the note
15377         should not be contained in a basic block.
15378         (emit_note_copy): Use make_note_raw.
15379         (emit_note): Likewise.
15380         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
15381         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
15382         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
15383         the moved barrier the tail of the basic block it follows.
15384         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
15386 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
15388         PR tree-optimization/56962
15389         * gimple-ssa-strength-reduction.c (record_increment): Only set
15390         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
15391         either rhs1 or rhs2 is equal to c->base_expr.
15393 2013-04-15  Richard Biener  <rguenther@suse.de>
15395         PR tree-optimization/56933
15396         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
15397         member.
15398         (GROUP_READ_WRITE_DEPENDENCE): Remove.
15399         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
15400         * tree-vect-data-refs.c (vect_analyze_group_access): Move
15401         dependence check ...
15402         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
15403         ... here.
15404         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
15405         GROUP_READ_WRITE_DEPENDENCE.
15407 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15409         * emit-rtl.c (reset_all_used_flags): New function.
15410         (verify_rtl_sharing): Call reset_all_used_flags before and after
15411         performing the checks.
15413 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15415         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
15416         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
15417         * config/arm/constraints.md (De): New constraint.
15418         * config/arm/neon.md (anddi3_neon): Delete.
15419         (neon_vand<mode>): Expand to standard anddi3 pattern.
15420         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
15421         Move earlier in the file.
15422         (neon_inv_logic_op2): Likewise.
15423         (arm_anddi_operand_neon): New predicate.
15425 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
15427         * configure.ac (gcc_cv_ld_as_needed): Set
15428         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
15429         Use -z ignore, -z record on *-*-solaris2*.
15430         (HAVE_LD_AS_NEEDED): Update comment.
15431         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
15432         * configure: Regenerate.
15433         * config.in: Regenerate.
15434         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
15435         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
15436         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
15437         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
15438         equivalents.  Fix markup.
15439         * doc/tm.texi: Regenerate.
15441 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
15443         * config/i386/i386.opt: New option mstack-protector-guard=.
15444         * config/i386/i386-opts.h: Add enum stack_protector_guard.
15445         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
15446         TARGET_SSP_TLS_GUARD.
15447         * config/i386/i386.c (ix86_option_override_internal): Set
15448         ix86_stack_protector_guard.
15449         * config/i386/i386.md (stack_protect_set): Enable for
15450         TARGET_SSP_TLS_GUARD only.
15451         (stack_protect_set_<mode>): Ditto.
15452         (stack_protect_test): Ditto.
15453         (stack_protect_test_<mode>): Ditto.
15454         * doc/invoke.texi (i386 Option): Document.
15456 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
15458         PR target/56890
15459         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
15460         (S_MODES): Set H_MODE bit.
15461         (SF_MODES): Set only S_MODE and SF_MODE bits.
15462         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
15463         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
15464         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
15465         <MODE_FLOAT>: Likewise.
15467 2013-04-15  Joey Ye  <joey.ye@arm.com>
15469         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
15471 2013-04-15  Joey Ye  <joey.ye@arm.com>
15473         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
15474         for real far jump.
15475         (thumb_far_jump_used_p): Count instruction size and set
15476         far_jump_used.
15478 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
15480         * reorg.c (fill_simple_delay_slots): Reindent block of code.
15481         * resource.c (mark_target_live_regs): Reformat conditional block.
15483 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
15485         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
15486         notes, they are emitted only just before final.
15487         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
15489 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
15491         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
15492         * cfgrtl.c (delete_insn): Call it here instead.
15493         * lra-spills.c (lra_final_code_change): Use delete_insn.
15494         * haifa-sched.c (sched_remove_insn): Likewise.
15495         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
15496         returning to the nop pool.
15497         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
15498         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
15500 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
15502         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
15503         * doc/tm.texi: Regenerated.
15505 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
15507         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
15508         QImode checks.
15510 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
15512         * df-core.c (df_find_def): Compare register numbers.
15513         (df_find_use): Likewise.
15515 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
15517         PR target/56903
15518         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
15519         lra_in_progress for return.
15521 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
15523         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
15524         define_insn into define_insn_and_split and emit movsicc patterns.
15526 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
15528         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
15530 2013-04-12  Richard Biener  <rguenther@suse.de>
15532         * tree-pass.h (TODO_do_not_ggc_collect): New.
15533         * passes.c (execute_one_ipa_transform_pass): Honor
15534         TODO_do_not_ggc_collect.
15535         (execute_one_pass): Likewise.
15537         Revert
15538         2013-04-10  Richard Biener  <rguenther@suse.de>
15540         * passes.c (init_optimization_passes): Remove reload pass.
15541         * ira.c (do_reload): Merge into ...
15542         (ira): ... this.
15543         (rest_of_handle_reload): Remove.
15544         (pass_reload): Likewise.
15545         * config/i386/i386.c (ix86_option_override): Refer to ira instead
15546         of reload for vzeroupper pass placement.
15548 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
15550         PR tree-optimization/56918
15551         PR tree-optimization/56920
15552         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
15553         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
15554         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
15555         use wide_mul_with_sign method.
15557 2013-04-12  Richard Biener  <rguenther@suse.de>
15559         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
15560         not be considered a gimple constant.
15562 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
15564         * fold-const.c (const_binop): Handle vector shifts by a scalar.
15565         (fold_binary_loc): Call const_binop also for mixed vector-scalar
15566         operations.
15568 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
15569             Jakub Jelinek  <jakub@redhat.com>
15571         * opts.c: Include diagnostic-color.h.
15572         (common_handle_option): Handle OPT_fdiagnostics_color_.
15573         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
15574         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
15575         (diagnostic-color.o): New.
15576         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
15577         (diagnostic_color_rule): New enum.
15578         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
15579         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
15580         the location string.
15581         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
15582         either NULL, or color kind.
15583         * diagnostic-color.c: New file.
15584         * diagnostic-color.h: New file.
15585         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
15586         arguments.
15587         * doc/invoke.texi (-fdiagnostics-color): Document.
15588         * pretty-print.h (pp_show_color): Define.
15589         (struct pretty_print_info): Add show_color field.
15590         * diagnostic.c: Include diagnostic-color.h.
15591         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
15592         macros.  Colorize error:, warning: etc. strings and also the location
15593         string.
15594         (diagnostic_show_locus): Colorize the caret line.
15595         * pretty-print.c: Include diagnostic-color.h.
15596         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
15597         inside of %< %> quotes or quoted through q format modifier.
15599 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15601         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
15603 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
15605         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
15606         code in CC_NZ mode.
15607         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
15608         pattern.
15610 2013-04-11  Marek Polacek  <polacek@redhat.com>
15612         PR tree-optimization/48184
15613         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
15615 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
15617         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
15618         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
15619         (skip_simple_arithmetic): Tidy up.
15620         * tree.h (skip_simple_constant_arithmetic): Declare.
15622 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
15624         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
15626 2013-04-11  Richard Biener  <rguenther@suse.de>
15628         * tree-vect-loop.c (get_initial_def_for_induction): Properly
15629         generate vector constants.
15631 2013-04-11  Richard Biener  <rguenther@suse.de>
15633         PR tree-optimization/56878
15634         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
15635         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
15636         New function.
15637         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
15638         Prefer to align the DR with the most invariant base address.
15640 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
15642         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
15643         comment.
15645 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
15647         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
15648         floating-point vector comparisons against 0.
15650 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
15652         PR tree-optimization/56899
15653         * fold-const.c (extract_muldiv_1): Apply distributive law
15654         only if TYPE_OVERFLOW_WRAPS (ctype).
15656 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
15658         PR target/56124
15659         * ira-costs.c (scan_one_insn): Check whether the source rtx of
15660         loading has side effect.
15662 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
15664         * config/sparc/sparc.c: Include tree-pass.h.
15665         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
15666         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
15667         head of file.  Change return type.  Split off gate function.
15668         (sparc_gate_work_around_errata): New function.
15669         (pass_work_around_errata): New pass definition.
15670         (insert_pass_work_around_errata) New pass insert definition to
15671         insert pass_work_around_errata just after delayed-branch scheduling.
15672         (sparc_option_override): Insert the pass.
15673         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
15675 2013-04-10  David S. Miller  <davem@davemloft.net>
15677         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
15678         or -mcpu=hypersparc.
15680         * target.def (cstore_mode): New hook.
15681         * target.h: Include insn-codes.h
15682         * targhooks.c: Likewise.
15683         (default_cstore_mode): New function.
15684         * targhooks.h: Declare it.
15685         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
15686         * doc/tm.texi: Rebuild.
15687         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
15688         target hook, rather than inspecting the insn_data.
15689         * config/sparc/sparc.c (sparc_cstore_mode): New function.
15690         (TARGET_CSTORE_MODE): Redefine.
15691         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
15692         result patterns.
15693         * config/sparc/predicates.md (cstore_result_operand): New special
15694         predicate.
15695         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
15696         Use it for operand 0.
15697         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
15698         (*snesi_special): Likewise.
15699         (*snesi_zero): Likewise.
15700         (*seqsi_zero): Likewise.
15701         (*sltu_insn): Likewise.
15702         (*sgeu_insn): Likewise.
15703         (*seqdi_special): Make operand 0 and comparison operation be of
15704         DImode.
15705         (*snedi_special): Likewise.
15706         (*snedi_special_vis3): Likewise.
15707         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
15708         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
15709         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
15710         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
15711         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
15712         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
15713         (*sltu_extend_sp64): Likewise.
15714         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
15715         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
15716         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
15717         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
15718         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
15720 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
15722         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
15723         (aarch64_start_file): Use the new function.
15725 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
15726             Jason Merrill  <jason@redhat.com>
15728         * common.opt: Add -gdwarf.
15729         * opts.c (common_handle_option): Handle it.
15730         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
15732 2013-04-10  Richard Biener  <rguenther@suse.de>
15734         * passes.c (execute_todo): Do not call ggc_collect conditional here.
15735         (execute_one_ipa_transform_pass): But unconditionally here.
15736         (execute_one_pass): And here.
15737         (init_optimization_passes): Remove reload pass.
15738         * tree-pass.h (TODO_ggc_collect): Remove.
15739         (pass_reload): Likewise.
15740         * ira.c (do_reload): Merge into ...
15741         (ira): ... this.
15742         (rest_of_handle_reload): Remove.
15743         (pass_reload): Likewise.
15744         * config/i386/i386.c (ix86_option_override): Refer to ira instead
15745         of reload for vzeroupper pass placement.
15746         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
15747         and todo_flags_finish of all passes.
15749 2013-04-10  Richard Biener  <rguenther@suse.de>
15751         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
15752         first_const_oprnd field, rename first_def_type to first_op_type.
15753         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
15754         (vect_get_and_check_slp_defs): Always use the type of the
15755         operand.  Allow mixed vect_external_def, vect_constant_def types.
15756         (vect_get_constant_vectors): Handle mixed vect_external_def,
15757         vect_constant_def types.
15759 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
15761         PR tree-optimization/55524
15762         * tree-ssa-math-opts.c
15763         (convert_mult_to_fma): Don't use an fms construct
15764         when we don't have an fms operation, but fnma, and it looks
15765         likely that we'll be able to use the latter.
15767 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
15769         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
15770         function.
15771         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
15772         inline fail caused by overwritable functions.
15774 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
15776         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
15777         unnecessary bits in the constant power of two case.
15779 2013-04-10  Richard Biener  <rguenther@suse.de>
15781         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
15782         broken code swapping operands.
15783         (vect_build_slp_tree): Do not compute load permutations here.
15784         (vect_analyze_slp_instance): Compute load permutations here,
15785         after building the SLP tree.
15787 2013-04-09  Christian Bruel  <christian.bruel@st.com>
15789         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
15790         of next/prev_real_insn.
15792 2013-04-09  Jan Hubicka  <jh@suse.cz>
15794         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
15795         Drop aliased parameter.
15796         (function_and_variable_visibility): Do not handle alias pairs.
15797         * cgraph.c (varpool_externally_visible_p): Update prototype.
15798         * varpool.c (varpool_add_new_variable): Update.
15800 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15802         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
15804 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
15806         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
15808         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
15810 2013-04-09  Marek Polacek  <polacek@redhat.com>
15812         PR tree-optimization/48762
15813         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
15815 2013-04-09  Richard Biener  <rguenther@suse.de>
15817         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
15818         dealing with cost.
15819         (vect_build_slp_tree): Likewise.
15820         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
15821         calculating the cost of a SLP instance.
15822         (vect_analyze_slp_instance): Use it from here, after building
15823         the SLP tree.
15825 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
15827         PR middle-end/56883
15828         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
15829         expand_omp_for_static_chunk): Use simple_p = true in
15830         force_gimple_operand_gsi calls when assigning to addressable decls.
15832 2013-04-09  Jeff Law  <law@redhat.com>
15834         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
15835         when the boolean was created by converting a wider object which
15836         had a boolean range.
15838 2013-04-09  Richard Biener  <rguenther@suse.de>
15840         * tree-vectorizer.h (slp_void_p): Remove.
15841         (slp_tree): Typedef before _slp_tree declaration.
15842         (struct _slp_tree): Use a vector of slp_tree as children.
15843         (vect_get_place_in_interleaving_chain): Remove.
15844         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
15845         Move ...
15846         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
15847         and make static.
15848         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
15849         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
15850         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
15851         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
15852         Use slp_node instead of slp_void_p and adjust.
15854 2013-04-09  Richard Biener  <rguenther@suse.de>
15856         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
15857         work that is not necessary.
15859 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
15861         PR tree-optimization/56854
15862         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
15863         forward into clobber stmts if it would change MEM_REF lhs into
15864         non-MEM_REF.
15866 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
15868         * tree.c (type_hash_lookup, type_hash_add): Make static.
15869         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
15871 2013-04-09  Richard Biener  <rguenther@suse.de>
15873         * tree.h (unsave_expr_now): Remove.
15874         * tree-inline.c (mark_local_for_remap_r): Remove.
15875         (unsave_expr_1): Likewise.
15876         (unsave_r): Likewise.
15877         (unsave_expr_now): Likewise.
15878         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
15879         (propagate_tree_value): Likewise.
15881 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
15883         * doc/rtl.texi (sequence): Rewrite documentation to match the
15884         current use of SEQUENCE rtl objects.
15885         * rtl.def (SEQUENCE): Likewise.
15887         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
15888         Update documentation.
15889         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
15890         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
15892         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
15894 2013-04-08  Teresa Johnson  <tejohnson@google.com>
15896         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
15897         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
15898         methods.
15899         (estimate_edge_size_and_time): Add comment to suggest using rounding
15900         methods.
15901         (estimate_node_size_and_time): Ditto.
15902         (remap_edge_change_prob): Use helper rounding divide methods.
15903         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
15904         (gimple_mod_pow2_value_transform): Ditto.
15905         (gimple_mod_subtract_transform): Ditto.
15906         (gimple_ic_transform): Ditto.
15907         (gimple_stringops_transform): Ditto.
15908         * stmt.c (conditional_probability): Ditto.
15909         (emit_case_dispatch_table): Ditto.
15910         * lto-cgraph.c (merge_profile_summaries): Ditto.
15911         * tree-optimize.c (execute_fixup_cfg): Ditto.
15912         * cfgcleanup.c (try_forward_edges): Ditto.
15913         * cfgloopmanip.c (scale_loop_profile): Ditto.
15914         (loopify): Ditto.
15915         (duplicate_loop_to_header_edge): Ditto.
15916         (lv_adjust_loop_entry_edge): Ditto.
15917         * tree-vect-loop.c (vect_transform_loop): Ditto.
15918         * profile.c (compute_branch_probabilities): Ditto.
15919         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
15920         * lto-streamer-in.c (input_cfg): Ditto.
15921         * gimple-streamer-in.c (input_bb): Ditto.
15922         * ipa-cp.c (update_profiling_info): Ditto.
15923         (update_specialized_profile): Ditto.
15924         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
15925         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
15926         rounding methods.
15927         * sched-rgn.c (compute_dom_prob_ps): Ditto.
15928         (compute_trg_info): Ditto.
15929         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
15930         (purge_dead_edges): Ditto.
15931         * loop-unswitch.c (unswitch_loop): Ditto.
15932         * cgraphclones.c (cgraph_clone_edge): Ditto.
15933         (cgraph_clone_node): Ditto.
15934         * tree-inline.c (copy_bb): Ditto.
15935         (copy_edges_for_bb): Ditto.
15936         (initialize_cfun): Ditto.
15937         (copy_cfg_body): Ditto.
15938         (expand_call_inline): Ditto.
15940 2013-04-08  Kai Tietz  <ktietz@redhat.com>
15942         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
15943         TARGET_CYGWIN64 by TARGET_64BIT.
15945 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
15947         * config/epiphany/epiphany.md (GPR_1): New constant.
15948         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
15949         * config/epiphany/epiphany.c (gen_compare_reg):
15950         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
15951         is already in place.
15952         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
15953         Don't require being called during rtl expansion; If y operlaps r0,
15954         return 0.
15955         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
15956         (epiphany_expand_epilogue): Likewise.
15958         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
15959         Don't use CC_FPmode for ORDERED / UNORDERED.
15960         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
15962         * config/epiphany/constraints.md (CnL): New constraint.
15963         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
15964         * config/epiphany/predicates.md (add_operand): Allow 1024.
15966         * config/epiphany/epiphany.md (logical_op): New code iterator.
15967         (op_mnc): New code attribute.
15968         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
15969         (mov_f+1, mov_f+2): New peephole2 patterns.
15971         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
15972         (cstoresi4): Also allow re-use of zero result when doing a NE
15973         comparison to a non-zero operand.
15974         Use (clobber (scratch)) for first insn if the gpr output is not needed.
15976         * config/epiphany/epiphany.md (<insn_opname>v2si3):
15977         Use gen_addsi3_i / gen_subsi3_i.
15979 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
15981         PR c++/34949
15982         PR c++/50243
15983         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
15984         contain anything but clobbers, at most one __builtin_stack_restore,
15985         optionally debug stmts and final resx, and if it has at least one
15986         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
15987         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
15988         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
15989         which isn't defaut definition, remove them.
15990         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
15991         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
15992         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
15993         with MEM_REF LHS with SSA_NAME address.
15995 2013-04-08  Jeff Law  <law@redhat.com>
15997         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
15999 2013-04-08  Richard Biener  <rguenther@suse.de>
16001         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
16002         extra newline.
16003         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
16004         determined vector type.
16005         (vect_analyze_data_refs): Likewise.
16006         (vect_get_new_vect_var): Adjust.
16007         (vect_create_destination_var): Preserve SSA name versions.
16008         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
16009         not dump anything here.
16011 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
16013         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
16014         Add member lr_slot_known.
16015         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
16016         if necessary.
16017         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
16018         Remove code that sets lr_slot_offset according to what a previous
16019         version of epiphany_emit_save_restore used to do.
16020         (epiphany_emit_save_restore): When doing an lr save or restore,
16021         set/verify lr_slot_known and lr_slot_offset.
16023 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
16025         PR target/54338
16026         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
16027         in ALL_REGS.
16029 2013-04-08  Richard Biener  <rguenther@suse.de>
16031         * alias.c (find_base_term): Fix thinko in previous change.
16033 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
16035         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
16036         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
16037         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
16038         if possible to compute val.
16039         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
16040         For QImode integers don't require anything about precision.  Use
16041         const_with_all_bytes_same to find out if the constant doesn't have
16042         repeated bytes in it.
16044 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16046         * config/s390/s390.c (s390_expand_insv): Only accept insertions
16047         within mode size.
16049 2013-04-08  Marek Polacek  <polacek@redhat.com>
16051         PR rtl-optimization/48182
16052         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
16053         value to 1.
16055 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16057         PR target/55487
16058         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
16059         nuses, make sure we have a label.
16061 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16063         PR target/56843
16064         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
16065         (rs6000_emit_swdiv_low_precision): Remove.
16066         (rs6000_emit_swdiv): Rewrite to handle between one and four
16067         iterations of Newton-Raphson generally; modify required number of
16068         iterations for some cases.
16069         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
16071 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
16073         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
16074         set-but-unused variable.
16076         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
16077         basic blocks of released function bodies garbage-collectable.
16079         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
16080         (struct rtl_opt_pass): Add TODO_df_finish.
16082         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
16084 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16086         * config/arm/constraints.md (q): New constraint.
16087         * config/arm/ldrdstrd.md: New file.
16088         * config/arm/arm.md (ldrdstrd.md) New include.
16089         (arm_movdi): Use "q" instead of "r" constraint
16090         for double-word memory access.
16091         (movdf_soft_insn): Likewise.
16092         * config/arm/vfp.md (movdi_vfp): Likewise.
16093         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
16094         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
16095         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
16096         (mem_ok_for_ldrd_strd): Likewise.
16097         (output_move_double): Update assertion.
16099 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16101         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
16103 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16105         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
16106         define_insn_and_split.
16107         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
16109 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16111         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
16112         define_insn_and_split.
16113         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
16114         (shiftsi3_compare): New pattern.
16115         (rrx): New pattern.
16116         * config/arm/unspecs.md (UNSPEC_RRX): New.
16118 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16120         * config/arm/arm.md (negdi_extendsidi): New pattern.
16121         (negdi_zero_extendsidi): Likewise.
16123 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16125         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
16126         define_insn_and_split.
16127         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
16128         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
16130 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16132         * config/arm/arm.md (arm_subdi3): Convert define_insn into
16133         define_insn_and_split.
16134         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
16135         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
16137 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16139         * config/arm/arm.md (subsi3_carryin): New pattern.
16140         (subsi3_carryin_const): Likewise.
16141         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
16142         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
16144 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16146         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
16148 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16150         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
16151         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
16153 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16155         * config/arm/arm.c (arm_expand_builtin): Change fcode
16156         type to unsigned int.
16158 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16160         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
16162 2013-04-04  Ian Lance Taylor  <iant@google.com>
16164         * doc/standards.texi (Standards): The Go frontend supports the Go 1
16165         language standard.
16167 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
16169         PR middle-end/56729
16170         * df-scan.c (df_insn_delete): Disable failing assert.
16172 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16174         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
16175         New function prototype.
16176         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
16177         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
16178         (arm_builtin_vectorized_function): New function.
16180 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16182         * config/arm/arm_neon_builtins.def: New file.
16183         * config/arm/arm.c (neon_builtin_data): Move contents to
16184         arm_neon_builtins.def.
16185         (enum arm_builtins): Include neon builtin definitions.
16186         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
16187         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
16189 2013-04-04  Marek Polacek  <polacek@redhat.com>
16191         PR tree-optimization/48186
16192         * predict.c (maybe_hot_frequency_p): Return false if
16193         HOT_BB_FREQUENCY_FRACTION is 0.
16194         (cgraph_maybe_hot_edge_p): Likewise.
16196 2013-04-04  Richard Biener  <rguenther@suse.de>
16198         PR tree-optimization/56826
16199         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
16200         more accurately.
16202 2013-04-04  Richard Biener  <rguenther@suse.de>
16204         PR tree-optimization/56213
16205         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
16206         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
16208 2013-04-04  Richard Biener  <rguenther@suse.de>
16210         PR tree-optimization/56837
16211         * tree-loop-distribution.c (classify_partition): For non-zero
16212         values require that the value has the same precision as its
16213         mode to be useful as memset value.
16215 2013-04-03  Nick Clifton  <nickc@redhat.com>
16217         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
16218         (fmssf4): Use fmsf.s on E3V5 architectures.
16219         (fnmasf4): Use fnmaf.s on E3V5 architectures.
16220         (fnmssf4): Use fnmsf.s on E3V5 architectures.
16222 2013-04-03  Jeff Law  <law@redhat.com>
16224         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
16225         (lra-eliminations.o): Likewise.
16227 2013-04-03  Teresa Johnson  <tejohnson@google.com>
16229         * gcov-io.c (compute_working_sets): Moved most of body of old
16230         compute_working_sets here from profile.c.
16231         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
16232         (gcov_working_set_t): Moved typedef here from basic-block.h
16233         (compute_working_set): Declare.
16234         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
16235         (get_working_sets): Renamed from compute_working_set,
16236         replace most of body with call to new compute_working_sets.
16237         (get_exec_counts): Replace call to compute_working_sets
16238         to get_working_sets.
16239         * profile.h (get_working_sets): Renamed from compute_working_set.
16240         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
16241         to get_working_sets.
16242         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
16243         * gcov-dump.c (dump_working_sets): New function.
16245 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
16247         * hwint.c (sext_hwi, zext_hwi): New functions.
16248         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
16249         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
16250         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
16251         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
16252         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
16253         (sext_hwi, zext_hwi): New functions.
16255 2013-04-03  Jeff Law  <law@redhat.com>
16257         PR tree-optimization/56799
16258         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
16259         back test for widening conversion erroneously dropped in prior change.
16261 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16263         PR target/56809
16264         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
16265         instead of next_real_insn.
16267 2013-04-03  Marek Polacek  <polacek@redhat.com>
16269         PR sanitizer/55702
16270         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
16272 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16274         PR target/56809
16275         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
16276         next_real_insn.
16277         (thumb1_output_casesi): Likewise.
16278         (thumb2_output_casesi): Likewise.
16280 2013-04-03  Richard Biener  <rguenther@suse.de>
16282         PR tree-optimization/56817
16283         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
16284         Split out ...
16285         (tree_unroll_loops_completely_1): ... new function to manually
16286         walk the loop tree, properly defering outer loops of unrolled
16287         loops to later iterations.
16289 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
16291         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
16292         (vectorizable_load): Likewise.
16293         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
16294         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
16296 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
16298         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
16299         BIT_FIELD_REF.
16301 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
16303         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
16305 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
16307         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
16309 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
16311         PR tree-optimization/56790
16312         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
16313         folding.
16315 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
16317         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
16318         Handle VEC_MERGE.
16319         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
16320         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
16321         equal arguments.
16323 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
16325         PR c/19449
16326         * tree.h (force_folding_builtin_constant_p): New decl.
16327         * builtins.c (force_folding_builtin_constant_p): New variable.
16328         (fold_builtin_constant_p): Fold immediately also if
16329         force_folding_builtin_constant_p.
16331 2013-04-03  Richard Biener  <rguenther@suse.de>
16333         PR tree-optimization/56812
16334         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
16335         DRs of the same interleaving chain are independent.
16337 2013-04-02  Jason Merrill  <jason@redhat.com>
16339         * gdbinit.in (pbb): Use debug fn.
16341 2013-04-02  Lawrence Crowl  <crowl@google.com>
16343         * sese.h (struct ivtype_map_elt_s): Remove unused.
16344         (extern debug_ivtype_map): Remove unused.
16345         (extern eq_ivtype_map_elts): Remove unused.
16346         * sese.c (debug_ivtype_map): Removed unused.
16347         (debug_ivtype_map_1): Removed unused.
16348         (debug_ivtype_elt): Remove unused.
16349         (eq_ivtype_map_elts): Remove unused.
16352 2013-04-02  Kai Tietz  <ktietz@redhat.com>
16354         PR target/52790
16355         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
16356         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
16357         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
16358         function.
16359         (legitimize_pe_coff_symbol): Likewise.
16360         (is_imported_p): New helper-function.
16361         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
16362         for Windows x64 targets.
16363         (ix86_expand_prologue): Optimize for pe-coff targets.
16364         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
16365         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
16366         medium/large code-model.
16367         (legitimize_pic_address): Likewise.
16368         (legitimize_tls_address): Likewise.
16369         (ix86_expand_call): Likewise.
16370         (x86_output_mi_thunk): Likewise.
16371         (get_dllimport_decl): Add new beimport argument.
16372         (construct_plt_address): Don't assert for x64 pe-coff targets.
16373         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
16374         targets.
16375         (SYMBOL_FLAG_STUBVAR): New macro.
16376         (SYMBOL_REF_STUBVAR_P): Likewise.
16377         * config/i386/winnt.c (stub_list): New structure.
16378         (stub_head): New local variable.
16379         (i386_pe_record_stub): New function.
16380         (i386_pe_file_end): Emit refptr-stubs.
16382 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
16384         PR rtl-optimization/56745
16385         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
16386         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
16388         PR c++/34949
16389         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
16390         and both of them are MEM_REFs, just compare first argument for
16391         equality and attempt to deal even with differing offsets.
16393         PR c++/34949
16394         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
16395         of gimple_clobber_p to be MEM_REF.
16396         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
16397         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
16398         after gimplification.
16399         * asan.c (get_mem_ref_of_assignment): Don't instrument
16400         gimple_clobber_p stmts.
16401         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
16402         gimple_clobber_p stmt if they have MEM_REF lhs and
16403         are dead because of another gimple_clobber_p stmt.
16404         * tree-ssa-live.c (clear_unused_block_pointer): Treat
16405         gimple_clobber_p stmts like debug stmts.
16406         (remove_unused_locals): Remove clobbers with MEM_REF lhs
16407         that refer to unused VAR_DECLs or uninitialized values.
16408         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
16409         gimple_clobber_p stmts if they refer to removed parameters.
16410         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
16411         formatting.
16413 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
16415         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
16416         using SWI48 mode attribute.
16418 2013-04-02  Wei Mi  <wmi@google.com>
16420         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
16421         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
16422         *<rotate_insn><mode>3_mask in i386.md.
16424 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
16426         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
16428 2013-04-02  Richard Biener  <rguenther@suse.de>
16430         PR tree-optimization/56778
16431         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
16432         Runtime alias tests are not supported for gather loads.
16433         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
16434         stmts referenced from SSA operands before updating SSA form.
16436 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
16437             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16439         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
16440         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
16441         * config/arm/cortex-a53.md: New file.
16442         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
16443         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
16444         * config/arm/arm.c (arm_issue_rate): Likewise.
16445         * config/arm/arm-tune.md: Regenerate
16446         * config/arm/arm-tables.opt: Regenerate.
16447         * config/arm/arm-cores.def: Add cortex-a53.
16449 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
16451         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
16452         non-static link.
16454 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
16456         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
16457         scalar load/store operations using B/H registers.
16458         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
16460 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
16462         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
16463         scalar move.
16464         * config/aarch64/aarch64.c
16465         (aarch64_simd_scalar_immediate_valid_for_move): New.
16466         * config/aarch64/aarch64-protos.h
16467         (aarch64_simd_scalar_immediate_valid_for_move): New.
16468         * config/aarch64/constraints.md (Dh, Dq): New.
16469         * config/aarch64/iterators.md (hq): New.
16471 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
16473         * reorg.c (get_branch_condition): Deal with conditional returns.
16474         (fill_simple_delay_slots): Remove dead code dealing with jumps.
16476 2013-04-01  Wei Mi  <wmi@google.com>
16478         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
16479         Truncate operand 2 using %b asm operand modifier.
16480         (*<shift_insn><mode>3_mask): Ditto.
16481         (*<rotate_insn><mode>3_mask): Ditto.
16483 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
16485         PR middle-end/56798
16486         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
16488 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
16490         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
16491         of next_real_insn.
16492         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
16494 2013-03-30  Lawrence Crowl  <crowl@google.com>
16496         * dse.c (clear_alias_sets): Remove never set.
16497         (disqualified_clear_alias_sets): Remove never set.
16498         (clear_alias_mode_pool): Remove never set.
16499         (dse_step0): Remove condition that is never true.
16500         (canon_address): Remove condition that is never true.
16501         (dse_step7): Remove condition that is never true.
16502         (rest_of_handle_dse): Remove condition that is never true.
16503         (rest_of_handle_dse::did_global): Remove never read from above.
16504         (dse_step2_spill): Remove never called from above.
16505         (dse_step5_spill): Remove never called from above.
16507 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
16509         * doc/md.texi (Standard Names) <casesi>: Update documentation for
16510         JUMP_TABLE_DATA changes.
16511         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
16512         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
16513         (Insns) <jump_table_data>: New entry.
16514         * doc/tm.texi: Regenerate.
16516         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
16518         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
16519         for table jump at the end of a basic block using tablejump_p.
16520         * targhooks.c (default_invalid_within_doloop): Likewise.
16521         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
16522         target hook implementation that is identical to the default hook.
16523         (rs6000_invalid_within_doloop): Remove.
16525         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
16526         unused variable from tablejump_p call.
16528         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
16529         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
16530         (INSN_DELETED_P): Likewise.
16531         (emit_jump_table_data): New prototype.
16532         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
16533         after 4th as unused.
16534         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
16535         * sched-vis.c (print_insn): Likewise.
16536         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
16537         insn for compatibility with back ends that use next_active_insn to
16538         identify jump table data.
16539         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
16540         (remove_insn): Likewise.
16541         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
16542         to be emitted.
16543         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
16544         (emit_jump_table_data): New function.
16546         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
16547         basic block, a JUMP_TABLE_DATA never is.
16548         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
16549         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
16550         off from code handling real insns.
16551         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
16552         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
16553         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
16554         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
16555         is not a NONDEBUG_INSN_P.
16556         * ira-costs.c (scan_one_insn): Likewise.
16557         * jump.c (mark_all_labels): Likewise.
16558         (mark_jump_label_1): Likewise.
16559         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
16560         * lra.c (get_insn_freq): Expect all insns reaching here to be in
16561         a basic block.
16562         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
16563         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
16564         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
16565         JUMP_TABLE_DATA_P insns.
16566         (calculate_elim_costs_all_insns): Likewise.
16567         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
16568         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
16569         (delete_output_reload): Code style fixups.
16570         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
16571         insn flags on this non-insn.
16572         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
16573         as scheduling barriers, for pre-change compatibility.
16574         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
16575         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
16577         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
16578         redundant JUMP_TABLE_DATA_P test.
16579         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
16580         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
16581         (frv_for_each_packet): Likewise.
16582         * config/i386/i386.c (min_insn_size): Likewise.
16583         (ix86_avoid_jump_mispredicts): Likewise.
16584         * config/m32r/m32r.c (m32r_is_insn): Likewise.
16585         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
16586         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
16587         (mips16_insn_length): Robustify.
16588         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
16589         (mips16_split_long_branches): Likewise.
16590         * config/pa/pa.c (pa_combine_instructions): Likewise.
16591         * config/rs6000/rs6000.c (get_next_active_insn): Treat
16592         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
16593         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
16594         as contributing to pool range lengths.
16595         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
16596         Remove redundant JUMP_TABLE_DATA_P test.
16597         (sh_loop_align): Likewise.
16598         (split_branches): Likewise.
16599         (sh_insn_length_adjustment): Likewise.
16600         * config/spu/spu.c (get_branch_target): Likewise.
16602 2013-03-29  Jan Hubicka  <jh@suse.cz>
16604         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
16605         gcov streaming; stream hot bb threshold to ltrans.
16606         * predict.c (get_hot_bb_threshold): Break out from ....
16607         (maybe_hot_count_p): ... here.
16608         (set_hot_bb_threshold): New function.
16609         * lto-section-in.c (lto_section_name): Add profile.
16610         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
16611         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
16612         and data-streamer.h
16613         (histogram_entry): New structure.
16614         (histogram, histogram_pool): New global vars.
16615         (histogram_hash): New structure.
16616         (histogram_hash::hash): New method.
16617         (histogram_hash::equal): Likewise.
16618         (account_time_size): New function.
16619         (cmp_counts): New function.
16620         (dump_histogram): New function.
16621         (ipa_profile_generate_summary): New function.
16622         (ipa_profile_write_summary): New function.
16623         (ipa_profile_read_summary): New function.
16624         (ipa_profile): Decide on threshold.
16625         (pass_ipa_profile): Add ipa_profile_write_summary and
16626         ipa_profile_read_summary.
16627         * Makefile.in (ipa.o): Update dependencies.
16628         * lto-streamer.h (LTO_section_ipa_profile): New section.
16630 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
16632         * tree.h (VAR_P): New.
16634 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
16636         PR lto/56777
16637         * doc/invoke.texi ([-fwhole-program]): Fix typo.
16639 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
16641         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
16642         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
16643         (control_flow_insn_p): Likewise.
16644         * cfgrtl.c (duplicate_insn_chain): Likewise.
16645         * final.c (get_attr_length_1): Likewise.
16646         (shorten_branches): Likewise.
16647         (final_scan_insn): Likewise.
16648         * function.c (instantiate_virtual_regs): Likewise.
16649         * gcse.c (insert_insn_end_basic_block): Likewise.
16650         * ira-costs.c (scan_one_insn): Likewise.
16651         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
16652         * lra.c (check_rtl): Likewise.
16653         * reload1.c (elimination_costs_in_insn): Likewise.
16654         * reorg.c (follow_jumps): Likewise.
16656         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
16657         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
16658         (thumb_far_jump_used_p): Likewise.
16659         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
16660         (workaround_speculation): Likewise.
16661         (add_sched_insns_for_speculation): Likewise.
16662         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
16663         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
16664         (frv_for_each_packet): Likewise.
16665         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
16666         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
16667         (final_emit_insn_group_barriers): Likewise.
16668         * config/m32r/m32r.c (m32r_is_insn): Likewise.
16669         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
16670         (mips16_insn_length): Likewise.
16671         * config/pa/pa.c (pa_reorg): Likewise.
16672         (pa_combine_instructions): Likewise.
16673         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
16674         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
16675         (sh_reorg): Likewise.
16676         (split_branches): Likewise.
16677         * config/spu/spu.c (get_branch_target): Likewise.
16679         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
16680         JUMP_TABLE_DATA_P.
16682 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
16684         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
16685         Fix declaration name.
16687 2013-03-28  Lawrence Crowl  <crowl@google.com>
16689         * graphds.h (struct graph.indicies): Remove unused.
16690         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
16691         (SCOP_ORIGINAL_PDDRS): Remove unused.
16692         * sese.h (extern insert_loop_close_phis): Removed unused.
16693         (extern insert_guard_phis): Removed unused.
16694         (extern ivtype_map_elt_info): Removed unused.
16695         (new_ivtype_map_elt): Removed unused.
16696         * sese.c (ivtype_map_elt_info): Removed unused.
16698 2013-03-28  Lawrence Crowl  <crowl@google.com>
16700         * Makefile.in: Add several missing include dependences.
16701         (DUMPFILE_H): New.
16702         (test-dump.o): New.  This object is not added to any executable,
16703         but is present for ad-hoc testing.
16704         * bitmap.c
16705         (debug (const bitmap_head_def &)): New.
16706         (debug (const bitmap_head_def *)): New.
16707         * bitmap.h
16708         (extern debug (const bitmap_head_def &)): New.
16709         (extern debug (const bitmap_head_def *)): New.
16710         * cfg.c
16711         (debug (edge_def &)): New.
16712         (debug (edge_def *)): New.
16713         * cfghooks.c
16714         (debug (basic_block_def &)): New.
16715         (debug (basic_block_def *)): New.
16716         * dumpfile.h
16717         (dump_node (const_tree, int, FILE *)): Correct source file.
16718         * dwarf2out.c
16719         (debug (die_struct &)): New.
16720         (debug (die_struct *)): New.
16721         * dwarf2out.h
16722         (extern debug (die_struct &)): New.
16723         (extern debug (die_struct *)): New.
16724         * gimple-pretty-print.c
16725         (debug (gimple_statement_d &)): New.
16726         (debug (gimple_statement_d *)): New.
16727         * gimple-pretty-print.h
16728         (extern debug (gimple_statement_d &)): New.
16729         (extern debug (gimple_statement_d *)): New.
16730         * ira-build.c
16731         (debug (ira_allocno_copy &)): New.
16732         (debug (ira_allocno_copy *)): New.
16733         (debug (ira_allocno &)): New.
16734         (debug (ira_allocno *)): New.
16735         * ira-int.h
16736         (extern debug (ira_allocno_copy &)): New.
16737         (extern debug (ira_allocno_copy *)): New.
16738         (extern debug (ira_allocno &)): New.
16739         (extern debug (ira_allocno *)): New.
16740         * ira-lives.c
16741         (debug (live_range &)): New.
16742         (debug (live_range *)): New.
16743         * lra-int.h
16744         (debug (lra_live_range &)): New.
16745         (debug (lra_live_range *)): New.
16746         * lra-lives.c
16747         (debug (lra_live_range &)): New.
16748         (debug (lra_live_range *)): New.
16749         * omega.c
16750         (debug (omega_pb_d &)): New.
16751         (debug (omega_pb_d *)): New.
16752         * omega.h
16753         (extern debug (omega_pb_d &)): New.
16754         (extern debug (omega_pb_d *)): New.
16755         * print-rtl.c
16756         (debug (const rtx_def &)): New.
16757         (debug (const rtx_def *)): New.
16758         * print-tree.c
16759         (debug_tree (tree): Move within file.
16760         (debug_raw (const tree_node &)): New.
16761         (debug_raw (const tree_node *)): New.
16762         (dump_tree_via_hooks (const tree_node *, int)): New.
16763         (debug (const tree_node &)): New.
16764         (debug (const tree_node *)): New.
16765         (debug_verbose (const tree_node &)): New.
16766         (debug_verbose (const tree_node *)): New.
16767         (debug_head (const tree_node &)): New.
16768         (debug_head (const tree_node *)): New.
16769         (debug_body (const tree_node &)): New.
16770         (debug_body (const tree_node *)): New.
16771         (debug_vec_tree (tree): Move and reimplement in terms of dump.
16772         (debug (vec<tree, va_gc> &)): New.
16773         (debug (vec<tree, va_gc> *)): New.
16774         * rtl.h
16775         (extern debug (const rtx_def &)): New.
16776         (extern debug (const rtx_def *)): New.
16777         * sbitmap.c
16778         (debug_raw (simple_bitmap_def &)): New.
16779         (debug_raw (simple_bitmap_def *)): New.
16780         (debug (simple_bitmap_def &)): New.
16781         (debug (simple_bitmap_def *)): New.
16782         * sbitmap.h
16783         (extern debug (simple_bitmap_def &)): New.
16784         (extern debug (simple_bitmap_def *)): New.
16785         (extern debug_raw (simple_bitmap_def &)): New.
16786         (extern debug_raw (simple_bitmap_def *)): New.
16787         * sel-sched-dump.c
16788         (debug (vinsn_def &)): New.
16789         (debug (vinsn_def *)): New.
16790         (debug_verbose (vinsn_def &)): New.
16791         (debug_verbose (vinsn_def *)): New.
16792         (debug (expr_def &)): New.
16793         (debug (expr_def *)): New.
16794         (debug_verbose (expr_def &)): New.
16795         (debug_verbose (expr_def *)): New.
16796         (debug (vec<rtx> &)): New.
16797         (debug (vec<rtx> *)): New.
16798         * sel-sched-dump.h
16799         (extern debug (vinsn_def &)): New.
16800         (extern debug (vinsn_def *)): New.
16801         (extern debug_verbose (vinsn_def &)): New.
16802         (extern debug_verbose (vinsn_def *)): New.
16803         (extern debug (expr_def &)): New.
16804         (extern debug (expr_def *)): New.
16805         (extern debug_verbose (expr_def &)): New.
16806         (extern debug_verbose (expr_def *)): New.
16807         (extern debug (vec<rtx> &)): New.
16808         (extern debug (vec<rtx> *)): New.
16809         * sel-sched-ir.h
16810         (_list_iter_cond_expr): Make inline instead of static.
16811         * sreal.c
16812         (debug (sreal &)): New.
16813         (debug (sreal *)): New.
16814         * sreal.h
16815         (extern debug (sreal &)): New.
16816         (extern debug (sreal *)): New.
16817         * tree.h
16818         (extern debug_raw (const tree_node &)): New.
16819         (extern debug_raw (const tree_node *)): New.
16820         (extern debug (const tree_node &)): New.
16821         (extern debug (const tree_node *)): New.
16822         (extern debug_verbose (const tree_node &)): New.
16823         (extern debug_verbose (const tree_node *)): New.
16824         (extern debug_head (const tree_node &)): New.
16825         (extern debug_head (const tree_node *)): New.
16826         (extern debug_body (const tree_node &)): New.
16827         (extern debug_body (const tree_node *)): New.
16828         (extern debug (vec<tree, va_gc> &)): New.
16829         (extern debug (vec<tree, va_gc> *)): New.
16830         * tree-cfg.c
16831         (debug (struct loop &)): New.
16832         (debug (struct loop *)): New.
16833         (debug_verbose (struct loop &)): New.
16834         (debug_verbose (struct loop *)): New.
16835         * tree-dump.c: Add header dependence.
16836         * tree-flow.h
16837         (extern debug (struct loop &)): New.
16838         (extern debug (struct loop *)): New.
16839         (extern debug_verbose (struct loop &)): New.
16840         (extern debug_verbose (struct loop *)): New.
16841         * tree-data-ref.c
16842         (debug (data_reference &)): New.
16843         (debug (data_reference *)): New.
16844         (debug (vec<data_reference_p> &)): New.
16845         (debug (vec<data_reference_p> *)): New.
16846         (debug (vec<ddr_p> &)): New.
16847         (debug (vec<ddr_p> *)): New.
16848         * tree-data-ref.h
16849         (extern debug (data_reference &)): New.
16850         (extern debug (data_reference *)): New.
16851         (extern debug (vec<data_reference_p> &)): New.
16852         (extern debug (vec<data_reference_p> *)): New.
16853         (extern debug (vec<ddr_p> &)): New.
16854         (extern debug (vec<ddr_p> *)): New.
16855         * tree-ssa-alias.c
16856         (debug (pt_solution &)): New.
16857         (debug (pt_solution *)): New.
16858         * tree-ssa-alias.h
16859         (extern debug (pt_solution &)): New.
16860         (extern debug (pt_solution *)): New.
16861         * tree-ssa-alias.c
16862         (debug (_var_map &)): New.
16863         (debug (_var_map *)): New.
16864         (debug (tree_live_info_d &)): New.
16865         (debug (tree_live_info_d *)): New.
16866         * tree-ssa-alias.h
16867         (extern debug (_var_map &)): New.
16868         (extern debug (_var_map *)): New.
16869         (extern debug (tree_live_info_d &)): New.
16870         (extern debug (tree_live_info_d *)): New.
16872 2013-03-28  Jan Hubicka  <jh@suse.cz>
16874         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
16876 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
16878         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
16879         record only when desired or required.
16881 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
16883         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
16884         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
16886 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16888         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
16889         (*andsi3_compare0_uxtw): New pattern.
16890         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
16891         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
16893 2013-03-28  Jan Hubicka  <jh@suse.cz>
16895         * data-streamer-in.c (streamer_read_gcov_count): New function.
16896         * gimple-streamer-out.c: Include value-prof.h.
16897         (output_gimple_stmt): Output histogram.
16898         (output_bb): Use streamer_write_gcov_count.
16899         * value-prof.c: Include data-streamer.h
16900         (dump_histogram_value): Add HIST_TYPE_MAX.
16901         (stream_out_histogram_value): New function.
16902         (stream_in_histogram_value): New function.
16903         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
16904         (stream_out_histogram_value, stream_in_histogram_value): Declare.
16905         * data-streamer-out.c (streamer_write_gcov_count): New function.
16906         (streamer_write_gcov_count_stream): New function.
16907         * lto-cgraph.c (lto_output_edge): Update counter streaming.
16908         (lto_output_node): Likewise.
16909         (input_node, input_edge): Likewise.
16910         * lto-streamer-out.c (output_cfg): Update streaming.
16911         * lto-streamer-in.c (input_cfg): Likewise.
16912         * data-streamer.h (streamer_write_gcov_count,
16913         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
16914         * gimple-streamer-in.c: Include value-prof.h
16915         (input_gimple_stmt): Input histograms.
16916         (input_bb): Update profile streaming.
16918 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
16920         * genmodes.c (emit_max_int): New function.
16921         (emit_insn_modes_h): Added call to emit_max_function.
16922         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
16923         Added doc.
16924         * machmode.def: Fixed comment.
16926 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
16928         * combine.c (try_combine): Removed useless assert.
16929         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
16931 2013-03-28  Marek Polacek  <polacek@redhat.com>
16932             Richard Biener  <rguenther@suse.de>
16934         PR tree-optimization/56695
16935         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
16936         build signed result of a vector comparison.
16937         * tree-cfg.c (verify_gimple_comparison): Check that a result
16938         of a vector comparison has signed type.
16940 2013-03-28  Richard Biener  <rguenther@suse.de>
16942         PR tree-optimization/37021
16943         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
16944         do not restrict gaps between groups.
16945         * tree-vect-stmts.c (vectorizable_load): Properly account for
16946         a gap between groups.
16948 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
16950         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
16951         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
16952         is not enabled.
16954 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
16956         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
16957         * doc/extend.texi (Named Address Spaces): Ditto.
16958         (Variable Attributes): Ditto.
16960 2013-03-27  Kai Tietz  <ktietz@redhat.com>
16962         * config.build: Add support for cygwin x64 target.
16963         * config.gcc: Likewise.
16964         * config.host: Likewise.
16965         * configure.ac: Likewise
16966         * configure: Regenerated.
16968 2013-03-27  Kai Tietz  <ktietz@redhat.com>
16970         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
16971         * config/i386/t-cygwin-w64: New file.
16972         * config/i386/cygwin-w64.h: New file.
16973         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
16974         and add support for x64-cygwin target.
16975         (CPP_SPEC): Likewise.
16976         (CXX_WRAP_SPEC_LIST): Undefine before define.
16977         (LIBGCJ_SONAME): Use 15 as version.
16979 2013-03-27  Richard Biener  <rguenther@suse.de>
16981         PR tree-optimization/56716
16982         * tree-ssa-structalias.c (perform_var_substitution): Adjust
16983         dumping for ref nodes.
16985 2013-03-27  Martin Jambor  <mjambor@suse.cz>
16987         PR tree-optimization/55334
16988         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
16989         restricted pointers to arrays.
16991 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
16993         * Makefile.in (.SUFFIXES): Add .cc.
16994         (.c.o): Apply same recipe for implicit rule .cc.o.
16996 2013-03-27  Richard Biener  <rguenther@suse.de>
16998         PR tree-optimization/37021
16999         * tree-vect-data-refs.c (vect_check_strided_load): Allow
17000         REALPART/IMAGPART_EXPRs around the supported refs.
17001         * tree-ssa-structalias.c (find_func_aliases): Assume that
17002         floating-point values are not used to transfer pointers.
17004 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
17006         * target.def (TARGET_HAS_IFUNC_P): New target hook.
17007         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
17008         * doc/tm.texi: Regenerate.
17009         * targhooks.h (default_has_ifunc_p): New.
17010         * targhooks.c (default_has_ifunc_p): Ditto.
17011         * config/linux-protos.h: New file.
17012         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
17013         hook for linux which disables support of indirect functions in android.
17014         * config/linux-android.c: New file.
17015         * config/t-linux-android.c: Ditto.
17016         * config.gcc: Added new object file linux-android.o.
17017         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
17018         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
17019         * varasm.c (do_assemble_alias): Likewise.
17020         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
17021         doesn't support indirect functions.
17022         * configure: Regenerate.
17024 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
17026         PR target/56102
17027         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
17028         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
17029         mult-word mode.
17031 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17033         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
17035 2013-03-27  Terry Guo  <terry.guo@arm.com>
17037         * config/arm/arm-cores.def: Added core cortex-r7.
17038         * config/arm/arm-tune.md: Regenerated.
17039         * config/arm/arm-tables.opt: Regenerated.
17040         * doc/invoke.texi: Added entry for core cortex-r7.
17042 2013-03-27  Walter Lee  <walt@tilera.com>
17044         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
17045         double-decrement of next_scratch_regno.
17047 2013-03-27  Walter Lee  <walt@tilera.com>
17049         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
17050         input operands.
17051         (insn_v1mulus): Ditto.
17052         (insn_v2muls): Ditto.
17054 2013-03-27  Walter Lee  <walt@tilera.com>
17056         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
17057         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
17059 2013-03-27  Walter Lee  <walt@tilera.com>
17061         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
17062         (*sibcall_value): Ditto.
17064 2013-03-27  Walter Lee  <walt@tilera.com>
17066         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
17067         (insn_mnz_v8qi): ... this ...
17068         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
17069         vector equivalent.
17070         (insn_v<n>mnz): Replaced by ...
17071         (insn_v1mnz): ... this ...
17072         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
17073         equivalent.
17074         (insn_mz_<mode>): Replaced by ...
17075         (insn_mz_v8qi): ... this ...
17076         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
17077         vector equivalent.
17078         (insn_v<n>mz): Replaced by ...
17079         (insn_v1mz): ... this ...
17080         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
17081         equivalent.
17083 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
17085         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
17087 2013-03-26  Roland McGrath  <mcgrathr@google.com>
17089         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
17090         than fprintf with a non-constant, non-format string.
17092 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
17094         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
17095         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
17096         operand 0 predicate.
17097         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
17098         attribute.  Use general_x64nomem_operand as operand 1 predicate.
17099         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
17100         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
17101         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
17102         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
17103         (mov<mode>_insv_1): Remove expander.  Merge insn with
17104         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
17105         Use general_x64nomem_operand as operand 1 predicate.
17106         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
17107         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
17108         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
17109         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
17110         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
17111         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
17112         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
17113         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
17114         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
17115         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
17116         (general_x64nomem_operand): Ditto.
17118 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
17120         * config/rtems.opt: Add -pthread option.
17122 2013-03-26  Richard Biener  <rguenther@suse.de>
17124         * alias.c (find_base_term): Avoid redundant and not used recursion.
17125         (base_alias_check): Get the initial base term from the caller.
17126         (true_dependence_1): Compute and pass base terms to base_alias_check.
17127         (write_dependence_p): Likewise.
17128         (may_alias_p): Likewise.
17130 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
17132         * config/aarch64/aarch64.c (aarch64_classify_address): Support
17133         PC-relative load in SI modes and above only.
17135 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
17137         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
17138         * config/arm/iwmmxt.md (WCGR0): Update.
17139         (WCGR1, WCGR2, WCGR3): Likewise.
17141 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
17143         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
17144         Use x64 and nox64 isa attributes.
17146 2013-03-26  Richard Biener  <rguenther@suse.de>
17148         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
17149         alignment computations and rely on get_object_alignment_1
17150         for the !TYPE_P case.
17151         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
17153 2013-03-26  Walter Lee  <walt@tilera.com>
17155         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
17156         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
17158 2013-03-25  Jeff Law  <law@redhat.com>
17160         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
17161         check for INTEGRAL_TYPE_P that was missing due to checking in
17162         wrong version of prior patch.
17164 2013-03-25  Walter Lee  <walt@tilera.com>
17166         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
17167         TILEGX_INSN_SHUFFLEBYTES1.
17168         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
17169         shufflebytes1.
17170         (tilegx_builtins): Ditto.
17171         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
17173 2013-03-25  Walter Lee  <walt@tilera.com>
17175         * config/tilegx/tilegx.md (floatsisf2): New pattern.
17176         (floatunssisf2): New pattern.
17177         (floatsidf2): New pattern.
17178         (floatunssidf2): New pattern.
17180 2013-03-25  Walter Lee  <walt@tilera.com>
17182         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
17183         tests for constraint J, K, N, P.
17185 2013-03-25  Walter Lee  <walt@tilera.com>
17187         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
17188         Use indirect/pcrel encoding.
17189         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
17190         Ditto.
17192 2013-03-25  Steve Ellcey  <sellcey@mips.com>
17194         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
17195         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
17196         * config/mips/mips.c (mips_option_override): Set IMADD default.
17197         * config/mips/mips.h (PTF_AVOID_IMADD): New.
17198         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
17199         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
17200         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
17201         * doc/invoke.texi (-mimadd/-mno-imadd): New.
17203 2013-03-25  Jeff Law  <law@redhat.com>
17205         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
17206         slightly to avoid creating and folding useless trees.  Simplify
17207         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
17209 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
17211         * config/i386/i386.md (*zero_extendsidi2): Merge with
17212         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
17213         * config/i386/predicates.md (x86_64_zext_operand): Rename from
17214         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
17215         targets.  Clarify comment.
17217 2013-03-25  Martin Jambor  <mjambor@suse.cz>
17219         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
17220         pass-through jump functions differently.
17221         (ipa_read_jump_function): Likewise.  Also use setter functions to set
17222         up jump functions.
17224 2013-03-25  Martin Jambor  <mjambor@suse.cz>
17226         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
17227         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
17228         process it.
17229         (ipa_get_indirect_edge_target): New function.
17230         (devirtualization_time_bonus): New parameter known_aggs, pass it to
17231         ipa_get_indirect_edge_target.  Update all callers.
17232         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
17233         ipa_get_indirect_edge_target_1 instead of calling
17234         ipa_get_indirect_edge_target.
17235         (create_specialized_node): Pass aggvlas to
17236         ipcp_discover_new_direct_edges.
17238 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17240         * config/arm/arm.md (f_sels, f_seld): New types.
17241         (*cmov<mode>): New pattern.
17242         * config/arm/predicates.md (arm_vsel_comparison_operator): New
17243         predicate.
17245 2013-03-25  Kai Tietz  <ktietz@redhat.com>
17247         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
17248         POSIX-printf for mingw-hosted  builds.
17250 2013-03-25  Richard Biener  <rguenther@suse.de>
17252         PR middle-end/56694
17253         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
17254         must-not-throw stmt location.
17256 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17258         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
17259         Emit load-acquire versions when acq is true.
17260         (arm_emit_store_exclusive): Add rel parameter.
17261         Emit store-release versions when rel is true.
17262         (arm_split_compare_and_swap): Use acquire-release instructions
17263         instead.
17264         of barriers when appropriate.
17265         (arm_split_atomic_op): Likewise.
17266         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
17267         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
17268         (VUNSPEC_SLX): Likewise.
17269         (VUNSPEC_LDA): Likewise.
17270         (VUNSPEC_STL): Likewise.
17271         * config/arm/sync.md (atomic_load<mode>): New pattern.
17272         (atomic_store<mode>): Likewise.
17273         (arm_load_acquire_exclusive<mode>): Likewise.
17274         (arm_load_acquire_exclusivesi): Likewise.
17275         (arm_load_acquire_exclusivedi): Likewise.
17276         (arm_store_release_exclusive<mode>): Likewise.
17278 2013-03-25  Catherine Moore  <clm@codesourcery.com>
17280         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
17281         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
17282         * config/mip/predicates.md (lwsp_swsp_operand,
17283         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
17284         sb16_operand, db4_operand, db7_operand, ib3_operand,
17285         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
17286         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
17287         andi16_operand): New predicates.
17288         * config/mips/mips.md (compression): New attribute.
17289         (enabled): New attribute.
17290         (length): Consider compression in computing length.
17291         (shift_compression): New code attribute.
17292         (*add<mode>3): New operands. Record compression.
17293         (sub<mode>3): Likewise.
17294         (one_cmpl<mode>2): Likewise.
17295         (*and<mode>3): Likewise.
17296         (*ior<mode>3): Likewise.
17297         (unnamed pattern for xor): Likewise.
17298         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
17299         (*<optab><mode>3): Likewise.
17300         (*mov<mode>_internal: Likewise.
17301         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
17302         (mips_unsigned_immediate_p): New.
17303         (umips_lwsp_swsp_address_p): New.
17304         (m16_based_address_p): New.
17305         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
17306         (mips_unsigned_immediate_p): New prototype.
17307         (lwsp_swsp_address_p): New prototype.
17308         (m16_based_address_p): New prototype.
17309         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
17310         (mips_signed_immediate_p): New function.
17311         (m16_based_address_p): New function.
17312         (lwsp_swsp_address_p): New function.
17313         (mips_print_operand_punctuation): Recognize short delay slot insns
17314         for microMIPS.add<mode>3"
17316 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17318         PR target/56720
17319         * config/arm/iterators.md (v_cmp_result): New mode attribute.
17320         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
17322 2013-03-25  Richard Biener  <rguenther@suse.de>
17324         PR tree-optimization/56689
17325         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
17326         any edge.
17328 2013-03-25  Richard Biener  <rguenther@suse.de>
17330         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
17331         of bitmap.
17332         (memory_references): Likewise.
17333         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
17334         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
17335         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
17336         (gather_mem_refs_in_loops): Fold into ...
17337         (analyze_memory_references): ... this.  Move initialization
17338         to tree_ssa_lim_initialize.
17339         (fill_always_executed_in): Rename to ...
17340         (fill_always_executed_in_1): ... this.
17341         (fill_always_executed_in): Move contains_call computation to
17342         this new function from ...
17343         (tree_ssa_lim_initialize): ... here.
17344         (tree_ssa_lim): Call fill_always_executed_in.
17346 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
17348         * postreload.c (reload_combine): Fix code detecting returns.
17350 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
17352         * function.c (emit_use_return_register_into_block): On cc0 targets,
17353         do not emit the sequence between cc0 setter and user.
17355 2013-03-25  Kai Tietz  <ktietz@redhat.com>
17357         * config/i386/predicates.md (local_symbolic_operand): Interpret
17358         dll-imported symbols as none-local.
17360 2013-03-25  Richard Biener  <rguenther@suse.de>
17362         * tree-ssa-loop-im.c (struct depend): Remove.
17363         (struct lim_aux_data): Make depends a vec of gimples.
17364         (free_lim_aux_data): Adjust.
17365         (add_dependency): Likewise.
17366         (set_level): Likewise.
17368 2013-03-25  Richard Biener  <rguenther@suse.de>
17370         PR middle-end/56434
17371         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
17372         the pointer returned by calls with ECF_MALLOC set.
17374 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
17376         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
17378 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
17380         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
17381         using MMXMODE mode iterator.
17382         (*move<mode>_internal): Merge with *movv2sf_internal and
17383         *movv2sf_internal_rex64 using MMXMODE mode iterator.
17385 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
17387         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
17388         (record_last_mem_set_info): Likewise.
17390         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
17391         of XNEWVEC followed by memset.
17392         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
17394 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
17396         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
17397         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
17398         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
17399         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
17400         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
17401         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
17402         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
17403         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
17404         BARRIER_P instead of GET_CODE.
17406 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
17408         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
17409         inaccuracy in the probing code.
17411         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
17412         (ctrapdi4): Likewise.
17414 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
17416         * calls.c (expand_call): Add missing guard to code handling return
17417         of non-BLKmode structures in MSB.
17418         * function.c (expand_function_end): Likewise.
17420 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
17422         * combine.c (try_combine): Adjust comment.  Do not add the set of
17423         insn #0 if the destination indirectly is set or dies in insn #2.
17424         Tidy up code to distribute a new note.
17426 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
17428         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
17429         also for alternatives 16 and 17.
17431 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
17433         * config/i386/sse.md (*mov<mode>_internal): Merge with
17434         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
17435         Emit insn template depending on type attribute.  Use
17436         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
17437         movd instead of movq mnemonic for interunit moves.  Rewrite mode
17438         attribute calculation.  Remove unit attribute calculation.
17439         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
17440         Set prefix_data16 attribute for DImode ssemov types.
17441         Use Ym instead of y for SSE-MMX conversion alternatives.
17442         Reorder operand constraints.
17444 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
17446         * df.h (df_insn_delete): Adjust prototype.
17447         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
17448         and let it decide whether mark the basic block dirty.
17449         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
17450         * df-scan.c (df_insn_info_delete): New helper function, split
17451         off from df_insn_delete.
17452         (df_scan_free_bb_info): Use it.
17453         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
17454         Likewise.
17455         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
17456         that the insn is actually an insn and it has a non-NULL basic block.
17457         Do not mark basic block dirty if only deleting a DEBUG_INSN.
17459 2013-03-22  Richard Biener  <rguenther@suse.de>
17461         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
17462         dep_ref members.
17463         (mem_ref_alloc): Do not allocate them.
17464         (refs_independent_p): Do not query or maintain a cache.
17466 2013-03-22  Richard Biener  <rguenther@suse.de>
17468         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
17469         (gather_mem_refs_in_loops): Do not compute it.
17470         (analyze_memory_references): Do not allocate it.
17471         (tree_ssa_lim_finalize): Do not free it.
17472         (for_all_locs_in_loop): Do not query all_refs_in_loop.
17474 2013-03-22  Richard Biener  <rguenther@suse.de>
17476         * is-a.h (as_a): Use gcc_checking_assert.
17478 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
17480         * config/aarch64/aarch64.c (aarch64_print_operand): New
17481         format specifier for printing a constant in hex.
17482         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
17483         format specifier for printing second operand.
17485 2013-03-22  Richard Biener  <rguenther@suse.de>
17487         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
17488         bitmaps.
17489         (gather_mem_refs_in_loops): Perform store accumulation here.
17490         (create_vop_ref_mapping_loop): Remove.
17491         (create_vop_ref_mapping): Likewise.
17492         (analyze_memory_references): Initialize refs_stored_in_loop.
17493         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
17494         (record_indep_loop): Remove.
17495         (record_dep_loop): New function.
17496         (ref_indep_loop_p_1): Adjust to only walk over references
17497         in the loop, not its subloops.
17498         (ref_indep_loop_p): Rename to ...
17499         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
17500         maintaining a more fine-grained cache.
17501         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
17502         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
17504 2013-03-22  Richard Biener  <rguenther@suse.de>
17506         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
17507         (struct mem_ref): Make accesses_in_loop a vec of a vec of
17508         aggregate mem_ref_loc.
17509         (free_mem_ref_locs): Inline into ...
17510         (memref_free): ... this and adjust.
17511         (mem_ref_alloc): Adjust.
17512         (mem_ref_locs_alloc): Remove.
17513         (record_mem_ref_loc): Adjust.
17514         (get_all_locs_in_loop): Rewrite into ...
17515         (for_all_locs_in_loop): ... this iterator.
17516         (rewrite_mem_ref_loc): New functor.
17517         (rewrite_mem_refs): Use for_all_locs_in_loop.
17518         (sm_set_flag_if_changed): New functor.
17519         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
17520         (ref_always_accessed): New functor.
17521         (ref_always_accessed_p): Use for_all_locs_in_loop.
17523 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
17525         * tree-pass.h (PROP_gimple_lvec): New.
17526         * passes.c (dump_properties): Handle PROP_gimple_lvec.
17527         (init_optimization_passes): Move pass_lower_vector.
17528         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
17529         PROP_gimple_lvec.
17530         (pass_lower_vector): Provide PROP_gimple_lvec.
17531         (pass_lower_vector_ssa): Likewise.
17532         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
17534 2013-03-21  Mark Wielaard  <mjw@redhat.com>
17536         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
17538 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
17540         * config/i386/i386.md (*movdi_internal): Disparage slightly
17541         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
17542         conversion alternatives.
17544 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
17546         PR middle-end/48087
17547         * diagnostic.def (DK_WERROR): New kind.
17548         * diagnostic.h (werrorcount): Define.
17549         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
17550         promoted to DK_ERROR, increment DK_WERROR counter instead of
17551         DK_ERROR counter.
17552         * toplev.c (toplev_main): Call print_ignored_options even if
17553         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
17554         even if just werrorcount is non-zero.
17556         PR debug/55608
17557         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
17558         on failure.
17559         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
17560         (string_cst_pool_decl): New function.
17561         (optimize_one_addr_into_implicit_ptr): New function.
17562         (resolve_addr_in_expr): Optimize DWARF location expression
17563         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
17564         which doesn't live in memory, but has DW_AT_location or
17565         DW_AT_const_value, or refers to a string literal, into
17566         DW_OP_GNU_implicit_pointer.
17567         (optimize_location_into_implicit_ptr): New function.
17568         (resolve_addr): If removing DW_AT_location of a variable because
17569         it was DW_OP_addr of address of the variable, but the variable doesn't
17570         live in memory, try to emit const value attribute for the initializer.
17572 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
17574         * tree.h (VECTOR_TYPE_P): New macro.
17575         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
17576         TYPE_MODE): Use it.
17577         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
17578         VEC_COND_EXPR cannot be lvalues.
17579         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
17581 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
17583         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
17584         Restrict the transformation to equal modes.
17586 2013-03-21  Richard Biener  <rguenther@suse.de>
17588         PR tree-optimization/39326
17589         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
17590         (MEM_ANALYZABLE): Adjust.
17591         (record_mem_ref_loc): Move bitmap ops ...
17592         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
17593         unanalyzable refs, do not record locations for it.
17594         (analyze_memory_references): Allocate ref zero as shared
17595         unanalyzable ref.
17596         (refs_independent_p): Do not test for unanalyzed mems here.
17597         (ref_indep_loop_p_1): Special-case disambiguation against
17598         the unanalyzed ref.
17599         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
17601 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
17603         * config/arm/arm-protos.h (tune_params): Add
17604         prefer_neon_for_64bits field.
17605         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
17606         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
17607         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
17608         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
17609         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
17610         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
17611         (arm_option_override): Handle -mneon-for-64bits new option.
17612         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
17613         (prefer_neon_for_64bits): Declare new variable.
17614         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
17615         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
17616         (arch_enabled): Handle new arch types. Remove support for onlya8
17617         and nota8.
17618         (one_cmpldi2): Use new arch names.
17619         (zero_extend<mode>di2, extend<mode>di2): Ditto.
17620         * config/arm/arm.opt (mneon-for-64bits): Add option.
17621         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
17622         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
17623         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
17624         of onlya8.
17625         * doc/invoke.texi (-mneon-for-64bits): Document.
17627 2013-03-21  Richard Biener  <rguenther@suse.de>
17629         PR tree-optimization/39326
17630         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
17631         (sort_bbs_in_loop_postorder_cmp): New function.
17632         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
17634 2013-03-21  Richard Biener  <rguenther@suse.de>
17636         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
17637         (vect_insert_into_interleaving_chain): Likewise.
17638         (vect_drs_dependent_in_basic_block): Inline ...
17639         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
17640         split out from ...
17641         (vect_analyze_data_ref_dependence): ... here.  Simplify.
17642         (vect_check_interleaving): Simplify.
17643         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
17644         (vect_slp_analyze_data_ref_dependences): ... this new function.
17645         (dr_group_sort_cmp): New function.
17646         (vect_analyze_data_ref_accesses): Compute data-reference groups
17647         here instead of in vect_analyze_data_ref_dependence.  Use
17648         a more efficient algorithm.
17649         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
17650         vect_slp_analyze_data_ref_dependences.  Call
17651         vect_analyze_data_ref_accesses earlier.
17652         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
17653         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
17654         (vect_slp_analyze_data_ref_dependences): New prototype.
17656 2013-03-21  Richard Biener  <rguenther@suse.de>
17658         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
17659         ref is stored in the loop.
17660         (find_refs_for_sm): Walk only over all stores.
17661         (store_motion_loop): Allocate from lim_bitmap_obstack.
17662         (store_motion): Likewise.
17664 2013-03-21  Richard Biener  <rguenther@suse.de>
17666         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
17667         Update virtual SSA form.
17669 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17671         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
17672         * configure: Regenerate.
17673         * config.in: Regenerate.
17674         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
17675         if !HAVE_LD_EH_FRAME_CIEV3.
17677 2013-03-21  Richard Biener  <rguenther@suse.de>
17679         * tree-cfg.c (verify_expr_no_block): New function.
17680         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
17681         nor DECL_VALUE_EXPR have locations with associated blocks.
17682         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
17683         (clear_unused_block_pointer): Remove code dealing with
17684         blocks in DECL_DEBUG_EXPR locations.
17686 2013-03-21  Richard Biener  <rguenther@suse.de>
17688         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
17689         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
17690         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
17691         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
17692         instead of DECL_DEBUG_EXPR_IS_FROM.
17693         * gimplify.c (gimplify_modify_expr): Likewise.
17694         * tree-cfg.c (verify_expr_location_1): Likewise.
17695         * tree-complex.c (create_one_component_var): Likewise.
17696         * tree-sra.c (create_access_replacement): Likewise.
17697         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
17698         (clear_unused_block_pointer): Likewise.
17699         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
17700         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
17701         * var-tracking.c (var_debug_decl): Likewise.
17702         (track_expr_p): Likewise.
17703         * tree-inline.c (add_local_variables): Likewise.  Set
17704         DECL_HAS_DEBUG_EXPR_P after copying it.
17705         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
17706         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
17708 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
17710         PR bootstrap/56656
17711         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
17712         * configure: Regenerate.
17713         * config.in: Regenerate.
17714         * config/i386/i386.md (*movdf_internal): Use
17715         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
17716         movd instead of movq mnemonic for interunit moves.
17717         (*movdi_internal): Ditto.
17719 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17721         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
17722         (abd<mode>_3): New pattern.
17723         (aba<mode>_3): New pattern.
17724         (fabd<mode>_3): New pattern.
17726 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17728         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
17729         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
17730         occurrence of REGISTER_PREFIX as its empty string.
17732 2013-03-20  Jeff Law  <law@redhat.com>
17734         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
17735         addititional equivalences for equality comparisons between an SSA_NAME
17736         and a constant where the SSA_NAME was set from a widening conversion.
17738 2013-03-20  Walter Lee  <walt@tilera.com>
17740         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
17742 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
17744         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
17745         depending on type attribute.
17746         (*movti_internal): Ditto.
17747         (*movtf_internal): Ditto.
17748         (*movxf_internal): Ditto.
17749         (*movdf_internal): Ditto.
17750         (*movsf_internal): Ditto.
17752 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
17754         * config/i386/i386.md (*movti_internal): Set prefix attribute to
17755         maybe_vex for sselog1 and ssemov types.
17756         (*movdi_internal): Reorder operand constraints.
17757         (*movsi_internal): Ditto.  Set prefix attribute to
17758         maybe_vex for sselog1 and ssemov types.
17759         (*movtf_internal): Set prefix attribute to maybe_vex
17760         for sselog1 and ssemov types.
17761         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
17762         DImode ssemov types.  Reorder operand constraints.
17763         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
17764         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
17765         attribute for SImode ssemov types.  Reorder operand constraints.
17767 2013-03-20  Martin Jambor  <mjambor@suse.cz>
17769         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
17770         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
17772 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
17774         * config/rs6000/predicates.md (indexed_address, update_address_mem
17775         update_indexed_address_mem): New predicates.
17776         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
17777         attribute for load/store instructions.
17778         * config/rs6000/dfp.md (movsd_store): Likewise.
17779         (movsd_load): Likewise.
17780         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
17781         (unnamed HI->DI extend define_insn): Likewise.
17782         (unnamed SI->DI extend define_insn): Likewise.
17783         (unnamed QI->SI extend define_insn): Likewise.
17784         (unnamed QI->HI extend define_insn): Likewise.
17785         (unnamed HI->SI extend define_insn): Likewise.
17786         (unnamed HI->SI extend define_insn): Likewise.
17787         (extendsfdf2_fpr): Likewise.
17788         (movsi_internal1): Likewise.
17789         (movsi_internal1_single): Likewise.
17790         (movhi_internal): Likewise.
17791         (movqi_internal): Likewise.
17792         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
17793         attribute for load/store instructions.
17794         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
17795         instructions.
17796         (mov<mode>_softfloat): Likewise.
17797         (mov<mode>_hardfloat32): Likewise.
17798         (mov<mode>_hardfloat64): Likewise.
17799         (mov<mode>_softfloat64): Likewise.
17800         (movdi_internal32): Likewise.
17801         (movdi_internal64): Likewise.
17802         (probe_stack_<mode>): Likewise.
17804 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
17806         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
17807         floating point, and decimal floating point to reload iterator.
17809         * config/rs6000/constraints.md (wl constraint): New constraints to
17810         return FLOAT_REGS if certain options are used to reduce the number
17811         of separate patterns that exist in the file.
17812         (wx constraint): Likewise.
17813         (wz constraint): Likewise.
17815         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
17816         -mdebug=reg, print wg, wl, wx, and wz constraints.
17817         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
17818         Initialize the reload functions for 64-bit binary/decimal floating
17819         point types.
17820         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
17821         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
17822         create the buffer on the stack to overcome not having a 32-bit
17823         load and store.
17824         (rs6000_emit_move): Likewise.
17825         (rs6000_secondary_memory_needed_rtx): Likewise.
17826         (rs6000_alloc_sdmode_stack_slot): Likewise.
17827         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
17828         via xxlxor, just like DFmode 0.0.
17830         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
17831         define as 1 if we are running on a power7 or newer.
17832         (enum r6000_reg_class_enum): Add new constraints.
17834         * config/rs6000/dfp.md (movsd): Delete, combine with binary
17835         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
17836         with other moves by using conditional constraits (wg).  Use LFIWZX
17837         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
17838         (movsd splitter): Likewise.
17839         (movsd_hardfloat): Likewise.
17840         (movsd_softfloat): Likewise.
17842         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
17843         binary and decimal floating point moves.
17844         (fmove_ok): New attributes to combine binary and decimal floating
17845         point moves, and to combine power6x (mfpgpr) moves along normal
17846         floating moves.
17847         (real_value_to_target): Likewise.
17848         (f32_lr): Likewise.
17849         (f32_lm): Likewise.
17850         (f32_li): Likewise.
17851         (f32_sr): Likewise.
17852         (f32_sm): Likewise.
17853         (f32_si): Likewise.
17854         (movsf): Combine binary and decimal floating point moves.  Combine
17855         power6x (mfpgpr) moves with other moves by using conditional
17856         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
17857         (mov<mode> for SFmode/SDmode); Likewise.
17858         (SFmode/SDmode splitters): Likewise.
17859         (movsf_hardfloat): Likewise.
17860         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
17861         (movsf_softfloat): Likewise.
17862         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
17864         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
17865         wx and wz constraints.
17867         * config/rs6000/constraints.md (wg constraint): New constraint to
17868         return FLOAT_REGS if -mmfpgpr (power6x) was used.
17870         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
17871         constraint.
17873         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
17874         -mdebug=reg, print wg, wl, wx, and wz constraints.
17875         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
17876         Initialize the reload functions for 64-bit binary/decimal floating
17877         point types.
17878         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
17879         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
17880         create the buffer on the stack to overcome not having a 32-bit
17881         load and store.
17882         (rs6000_emit_move): Likewise.
17883         (rs6000_secondary_memory_needed_rtx): Likewise.
17884         (rs6000_alloc_sdmode_stack_slot): Likewise.
17885         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
17886         via xxlxor, just like DFmode 0.0.
17888         * config/rs6000/dfp.md (movdd): Delete, combine with binary
17889         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
17890         with other moves by using conditional constraits (wg).  Use LFIWZX
17891         and STFIWX for loading SDmode on power7.
17892         (movdd splitters): Likewise.
17893         (movdd_hardfloat32): Likewise.
17894         (movdd_softfloat32): Likewise.
17895         (movdd_hardfloat64_mfpgpr): Likewise.
17896         (movdd_hardfloat64): Likewise.
17897         (movdd_softfloat64): Likewise.
17899         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
17900         64-bit binary and decimal floating point moves.
17901         (FMOVE64X): Likewise.
17902         (movdf): Combine 64-bit binary and decimal floating point moves.
17903         Combine power6x (mfpgpr) moves with other moves by using
17904         conditional constraits (wg).
17905         (mov<mode> for DFmode/DDmode): Likewise.
17906         (DFmode/DDmode splitters): Likewise.
17907         (movdf_hardfloat32): Likewise.
17908         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
17909         (movdf_softfloat32): Likewise.
17910         (movdf_hardfloat64_mfpgpr): Likewise.
17911         (movdf_hardfloat64): Likewise.
17912         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
17913         (movdf_softfloat64): Likewise.
17914         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
17915         (reload_<mode>_load): Move to later in the file so they aren't in
17916         the middle of the floating point move insns.
17917         (reload_<mode>_store): Likewise.
17919         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
17920         constraint.
17922         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
17923         constraint if -mdebug=reg.
17924         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
17925         Enable using dd reload support if needed.
17927         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
17928         binary and decimal floating point moves in rs6000.md.
17929         (movtd_internal): Likewise.
17931         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
17932         decimal floating point moves.
17933         (movtf): Likewise.
17934         (movtf_internal): Likewise.
17935         (mov<mode>_internal, TDmode/TFmode): Likewise.
17936         (movtf_softfloat): Likewise.
17937         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
17939         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
17940         movdi_internal64, using wg constraint for move direct operations.
17941         (movdi_internal64): Likewise.
17943         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
17944         MODES_TIEABLE_P for selected modes.  Print the numerical value of
17945         the various virtual registers. Use GPR/FPR first/last values,
17946         instead of hard coding the register numbers.  Print which modes
17947         have reload functions registered.
17948         (rs6000_option_override_internal): If -mdebug=reg, trace the options
17949         settings before/after setting cpu, target and subtarget settings.
17950         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
17951         and for secondary reload failures in rs6000_secondary_reload_inner.
17952         (rs6000_secondary_reload_fail): Likewise.
17953         (rs6000_secondary_reload_inner): Likewise.
17955         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
17956         macros for first/last GPR and FPR registers.
17957         (LAST_GPR_REGNO): Likewise.
17958         (FIRST_FPR_REGNO): Likewise.
17959         (LAST_FPR_REGNO): Likewise.
17961         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
17962         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
17963         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
17964         (vcond<mode><mode>): Likewise.
17965         (vcondu<mode><mode>): Likewise.
17966         (vector_gtu<mode>): Likewise.
17967         (vector_gte<mode>): Likewise.
17968         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
17969         to prevent the compiler from converting DImode operations to TImode.
17970         (ior<mode>3): Likewise.
17971         (and<mode>3): Likewise.
17972         (one_cmpl<mode>2): Likewise.
17973         (nor<mode>3): Likewise.
17974         (andc<mode>3): Likewise.
17976         * config/rs6000/constraints.md (wt constraint): New constraint
17977         that returns VSX_REGS if TImode is allowed in VSX registers.
17979         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
17980         constant under VSX.
17982         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
17983         similar to TImode, but it is restricted to being in the GPRs.
17985         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
17986         TImode to occupy a single VSX register.
17988         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
17989         -mvsx-timode for power7/power8.
17990         (power7 cpu): Likewise.
17991         (power8 cpu): Likewise.
17993         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
17994         sure that TFmode/TDmode take up two registers if they are ever
17995         allowed in the upper VSX registers.
17996         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
17997         registers.
17998         (rs6000_init_hard_regno_mode_ok): Likewise.
17999         (rs6000_debug_reg_global): Add debugging for PTImode and wt
18000         constraint.  Print if LRA is turned on.
18001         (rs6000_option_override_internal): Give an error if -mvsx-timode
18002         and VSX is not enabled.
18003         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
18004         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
18005         to reg+offset addressing.  Use PTImode when checking offset
18006         addresses for validity.
18007         (reg_offset_addressing_ok_p): Likewise.
18008         (rs6000_legitimate_offset_address_p): Likewise.
18009         (rs6000_legitimize_address): Likewise.
18010         (rs6000_legitimize_reload_address): Likewise.
18011         (rs6000_legitimate_address_p): Likewise.
18012         (rs6000_eliminate_indexed_memrefs): Likewise.
18013         (rs6000_emit_move): Likewise.
18014         (rs6000_secondary_reload): Likewise.
18015         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
18016         reloads to fpr registers to continue to use reg+offset addressing,
18017         but 64-bit reloads to altivec registers need reg+reg addressing.
18018         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
18019         it.  Treat LO_SUM like a PLUS operation.
18020         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
18021         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
18022         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
18023         registers to share a register with a smaller sized type, since VSX
18024         puts scalars in the upper 64-bits.
18025         (print_operand): Add support for PTImode.
18026         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
18027         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
18028         registers, but don't have arithmetic support.
18029         (rs6000_memory_move_cost): Add test for VSX.
18030         (rs6000_opt_masks): Add -mvsx-timode.
18032         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
18033         for TImode.
18034         (VSs): Likewise.
18035         (VSr): Use wt constraint for TImode.
18036         (VSv): Drop TImode support.
18037         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
18038         (vsx_movti_64bit): Likewise.
18039         (vsx_movti_32bit): Likewise.
18040         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
18041         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
18042         one '?' on the appropriate output constraint.  Do not allow TImode
18043         logical operations on 32-bit systems.
18044         (vsx_ior<mode>3): Likewise.
18045         (vsx_xor<mode>3): Likewise.
18046         (vsx_one_cmpl<mode>2): Likewise.
18047         (vsx_nor<mode>3): Likewise.
18048         (vsx_andc<mode>3): Likewise.
18049         (vsx_concat_<mode>): Likewise.
18050         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
18052         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
18053         OPTION_MASK_VSX_TIMODE.
18054         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
18055         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
18057         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
18058         (TI2 iterator): New iterator for TImode, PTImode.
18059         (wd mode attribute): Add values for vector types.
18060         (movti_string): Replace TI move operations with operations for TImode
18061         and PTImode.  Add support for TImode being allowed in VSX registers.
18062         (mov<mode>_string, TImode/PTImode): Likewise.
18063         (movti_ppc64): Likewise.
18064         (mov<mode>_ppc64, TImode/PTImode): Likewise.
18065         (TI mode splitters): Likewise.
18067         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
18068         constraint.
18070 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
18072         PR tree-optimization/56355
18073         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
18074         Also handle integers with undefined overflow.
18076 2013-03-20  Catherine Moore  <clm@codesourcery.com>
18077             Maciej W. Rozycki  <macro@codesourcery.com>
18078             Tom de Vries  <tom@codesourcery.com>
18079             Nathan Sidwell  <nathan@codesourcery.com>
18080             Iain Sandoe  <iain@codesourcery.com>
18081             Nathan Froyd  <froydnj@codesourcery.com>
18082             Chao-ying Fu  <fu@mips.com>
18084         * doc/extend.texi: (micromips, nomicromips, nocompression):
18085         Document new function attributes.
18086         * doc/invoke.texi (minterlink-compressed, mmicromips,
18087         m14k, m14ke, m14kec): Document new options.
18088         (minterlink-mips16): Update documentation.
18089         * doc/md.texi (ZC, ZD): Document new constraints.
18090         * configure.ac (gcc_cv_as_micromips): Check if linker
18091         supports the .set micromips directive.
18092         * configure: Regenerate.
18093         * config.in: Regenerate.
18094         * config/mips/mips-tables.opt: Regenerate.
18095         * config/mips/micromips.md: New file.
18096         * constraints.md (ZC, ZD): New constraints.
18097         * config/mips/predicates.md (movep_src_register): New predicate.
18098         (movep_src_operand): New predicate.
18099         (non_volatile_mem_operand): New predicate.
18100         * config/mips/mips.md (multimem): New type.
18101         (length): Differentiate between 17-bit and 18-bit branch offsets.
18102         (MOVEP1, MOVEP2): New mode iterator.
18103         (mov_<load>l): Use ZC constraint.
18104         (mov_<load>r): Likewise.
18105         (mov_<store>l): Likewise.
18106         (mov_<store>r): Likewise.
18107         (*branch_equality<mode>_inverted): Add microMIPS support.
18108         (*branch_equality<mode>): Likewise.
18109         (*jump_absolute): Likewise.
18110         (indirect_jump_<mode>): Likewise.
18111         (tablejump_<mode>): Likewise.
18112         (<optab>_internal): Likewise.
18113         (sibcall_internal): Likewise.
18114         (sibcall_value_internal): Likewise.
18115         (prefetch): Use constraint ZD.
18116         * config/mips/mips.opt (minterlink-compressed): New option.
18117         (minterlink-mips16): Now an alias for minterlink-compressed.
18118         (mmicromips): New option.
18119         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
18120         (compare_and_swap_12): Likewise.
18121         (sync_add<mode>): Likewise.
18122         (sync_<optab>_12): Likewise.
18123         (sync_old_<optab>_12): Likewise.
18124         (sync_new_<optab>_12): Likewise.
18125         (sync_nand_12): Likewise.
18126         (sync_old_nand_12): Likewise.
18127         (sync_new_nand_12): Likewise.
18128         (sync_sub<mode>): Likewise.
18129         (sync_old_add<mode>): Likewise.
18130         (sync_old_sub<mode>): Likewise.
18131         (sync_new_add<mode>): Likewise.
18132         (sync_new_sub<mode>): Likewise.
18133         (sync_<optab><mode>): Likewise.
18134         (sync_old_<optab><mode>): Likewise.
18135         (sync_new_<optab><mode>): Likewise.
18136         (sync_nand<mode>): Likewise.
18137         (sync_old_nand<mode>): Likewise.
18138         (sync_new_nand<mode>): Likewise.
18139         (sync_lock_test_and_set<mode>): Likewise.
18140         (test_and_set_12): Likewise.
18141         (atomic_compare_and_swap<mode>): Likewise.
18142         (atomic_exchange<mode>_llsc): Likewise.
18143         (atomic_fetch_add<mode>_llsc): Likewise.
18144         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
18145         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
18146         (umips_save_restore_pattern_p): Likewise.
18147         (umips_load_store_pair_p): Likewise.
18148         (umips_output_load_store_pair): Likewise.
18149         (umips_movep_target_p): Likewise.
18150         (umips_12bit_offset_address_p): Likewise.
18151         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
18152         (mips_base_mips16): Rename this...
18153         (mips_base_compression_flags): ...to this. Update all uses.
18154         (mips_attribute_table): Add micromips, nomicromips and nocompression.
18155         (mips_mips16_decl_p): Delete.
18156         (mips_nomips16_decl_p): Delete.
18157         (mips_get_compress_on_flags): New function.
18158         (mips_get_compress_off_flags): New function.
18159         (mips_get_compress_mode): New function.
18160         (mips_get_compress_on_name): New function.
18161         (mips_get_compress_off_name): New function.
18162         (mips_insert_attributes): Support multiple compression types.
18163         (mips_merge_decl_attributes): Likewise.
18164         (umips_12bit_offset_address_p): New function.
18165         (mips_start_function_definition): Emit .set micromips directive.
18166         (mips_call_may_need_jalx_p): New function.
18167         (mips_function_ok_for_sibcall): Add microMIPS support.
18168         (mips_print_operand_punctuation): Support short delay slots and
18169         compact jumps.
18170         (umips_swm_mask, umips_swm_encoding): New.
18171         (umips_build_save_restore): New function.
18172         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
18173         (was_mips16_p): Remove.
18174         (old_compression_mode): New.
18175         (mips_set_compression_mode): New function.
18176         (mips_set_current_function): Add microMIPS support.
18177         (mips_option_override): Likewise.
18178         (umips_save_restore_pattern_p): New function.
18179         (umips_output_save_restore): New function.
18180         (umips_load_store_pair_p_1): New function.
18181         (umips_load_store_pair_p): New function.
18182         (umips_output_load_store_pair_1): New function.
18183         (umips_output_load_store_pair): New function.
18184         (umips_movep_target_p) New function.
18185         (mips_prepare_pch_save): Add microMIPS support.
18186         * config/mips/mips.h (TARGET_COMPRESSION): New.
18187         (TARGET_CPU_CPP_BUILTINS): Update macro
18188         to use new compression flags and to support microMIPS.
18189         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
18190         (MIPS_ARCH_FLOAT_SPEC): Likewise.
18191         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
18192         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
18193         (ASM_SPEC): Support mmicromips and mno-micromips.
18194         (M16STORE_REG_P): New macro.
18195         (MIPS_CALL): Support TARGET_MICROMIPS.
18196         (MICROMIPS_J): New macro.
18197         (mips_base_mips16): Rename this...
18198         (mips_base_compression_flags): ...to this.
18199         (UMIPS_12BIT_OFFSET_P): New macro.
18200         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
18201         (MULTILIB_DIRNAMES): Likewise.
18202 2013-03-20  Richard Biener  <rguenther@suse.de>
18204         PR tree-optimization/56661
18205         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
18206         the result does not have to be distinct.
18208 2013-03-20  Richard Biener  <rguenther@suse.de>
18210         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
18211         remap_gimple_op_r.
18213 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18214             Steven Bosscher  <steven@gcc.gnu.org>
18216         PR rtl-optimization/56605
18217         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
18219 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
18221         PR bootstrap/56656
18222         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
18223         that require movd instead of movq.
18225 2013-03-20  Richard Biener  <rguenther@suse.de>
18227         * tree-ssa-structalias.c (struct variable_info): Add pointer
18228         to the first field of an aggregate with sub-vars.  Make
18229         this and the pointer to the next subfield its ID.
18230         (vi_next): New function.
18231         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
18232         storedanything_id, integer_id): Increment by one.
18233         (new_var_info, get_call_vi, lookup_call_clobber_vi,
18234         get_call_clobber_vi): Adjust.
18235         (solution_set_expand): Simplify and speedup.
18236         (solution_set_add): Inline into ...
18237         (set_union_with_increment): ... this.  Adjust accordingly.
18238         (do_sd_constraint): Likewise.
18239         (do_ds_constraint): Likewise.
18240         (do_complex_constraint): Simplify.
18241         (build_pred_graph): Adjust.
18242         (solve_graph): Likewise.  Simplify and speedup.
18243         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
18244         get_constraint_for_component_ref, get_constraint_for_1,
18245         first_vi_for_offset, first_or_preceding_vi_for_offset,
18246         create_function_info_for, create_variable_info_for_1,
18247         create_variable_info_for, intra_create_variable_infos): Adjust.
18248         (init_base_vars): Push NULL for ID zero.
18249         (compute_points_to_sets): Adjust.
18251 2013-03-20  Richard Biener  <rguenther@suse.de>
18253         * cfgloop.c (verify_loop_structure): Streamline and avoid
18254         ICEing on corrupt loop tree.
18255         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
18256         loop tree.
18258 2013-03-20  Richard Biener  <rguenther@suse.de>
18260         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
18261         check whether an SSA update is needed.
18263 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
18265         * config/mips/constraints.md (T): Rename to...
18266         (Yf): ...this.
18267         (U): Rename to...
18268         (Yd): ...this.
18269         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
18270         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
18272 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
18274         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
18275         (*subsi3_carryin_uxtw): Likewise.
18277 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
18279         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
18280         (*rorsi3_insn_uxtw): Likewise.
18282 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
18284         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
18285         (*extrsi5_insn_uxtw): Likewise.
18287 2013-03-19  Richard Biener  <rguenther@suse.de>
18289         PR tree-optimization/56273
18290         * passes.c (init_optimization_passes): Move second VRP after DOM.
18292 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
18294         * config/i386/i386.md (*movti_internal): Merge from
18295         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
18296         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
18297         nox64 isa attributes.
18299 2013-03-18  Richard Biener  <rguenther@suse.de>
18301         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
18302         (unite): Likewise.
18303         (merge_node_constraints): Likewise.
18304         (build_succ_graph): Likewise.
18305         (valid_graph_edge): Inline into single caller.
18306         (unify_nodes): Likewise.  Use bitmap_set_bit return value
18307         and cache varinfo.
18308         (scc_visit): Fix formatting and variable use.
18309         (do_sd_constraint): Use gcc_checking_assert.
18310         (do_ds_constraint): Likewise.
18311         (do_complex_constraint): Likewise.
18312         (condense_visit): Likewise.  Cleanup.
18313         (dump_pred_graph): New function.
18314         (perform_var_substitution): Dump the pred-graph before
18315         variable substitution.
18316         (find_equivalent_node): Use gcc_checking_assert.
18317         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
18319 2013-03-18  Richard Biener  <rguenther@suse.de>
18321         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
18322         Remove cond_expr_stmt_list argument and do not gimplify the
18323         built expression.
18324         (vect_loop_versioning): Adjust.
18325         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
18326         Cleanup to use less temporaries.
18327         (vect_create_data_ref_ptr): Cleanup.
18329 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
18331         PR tree-optimization/56635
18332         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
18333         require types_compatible_p types.
18335 2013-03-18  Nick Clifton  <nickc@redhat.com>
18337         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
18338         spurious backslash.
18340         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
18341         Add missing line to comment describing function.
18343 2013-03-18  Richard Biener  <rguenther@suse.de>
18345         PR tree-optimization/56210
18346         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18347         Handle string / character search functions.
18348         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
18350 2013-03-18  Richard Biener  <rguenther@suse.de>
18352         PR middle-end/56483
18353         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
18354         and implement properly.
18355         * gimple.h (gimple_cond_single_var_p): Remove.
18357 2013-03-18  Richard Biener  <rguenther@suse.de>
18359         * tree-data-ref.h (find_data_references_in_loop): Declare.
18360         * tree-data-ref.c (get_references_in_stmt): Use a stack
18361         vector pre-allocated in the callers.
18362         (find_data_references_in_stmt): Adjust.
18363         (graphite_find_data_references_in_stmt): Likewise.
18364         (create_rdg_vertices): Likewise.
18365         (find_data_references_in_loop): Export.
18366         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
18367         Compute dependences here...
18368         (vect_analyze_data_refs): ...not here.  When we encounter
18369         a non-vectorizable data reference in basic-block vectorization
18370         truncate the data reference vector.  Do not bother to
18371         fixup data-dependence information for gather loads.
18372         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
18373         of data references, as reported.
18375 2013-03-18  Richard Biener  <rguenther@suse.de>
18377         PR tree-optimization/3713
18378         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
18379         has_constants and expr.
18380         (stmt_has_constants): Properly valueize SSA names when deciding
18381         whether the stmt has constants.
18383 2013-03-18  Richard Biener  <rguenther@suse.de>
18385         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
18386         whole function when there is nothing to do.
18387         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
18388         * tree-vectorizer.c (vectorize_loops): Update virtual and
18389         loop-closed SSA once.
18390         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
18392 2013-03-18  Richard Biener  <rguenther@suse.de>
18394         PR middle-end/56113
18395         * domwalk.c (bb_postorder): New global static.
18396         (cmp_bb_postorder): New function.
18397         (walk_dominator_tree): Replace scheme imposing an order for
18398         visiting dominator sons by one sorting them at the time they
18399         are pushed on the stack.
18401 2013-03-18  Richard Biener  <rguenther@suse.de>
18403         PR tree-optimization/39326
18404         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
18405         (struct mem_ref): Replace mem member with ao_ref typed member.
18406         (MEM_ANALYZABLE): Adjust.
18407         (memref_eq): Likewise.
18408         (mem_ref_alloc): Likewise.
18409         (gather_mem_refs_stmt): Likewise.
18410         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
18411         (execute_sm_if_changed_flag_set): Adjust.
18412         (execute_sm): Likewise.
18413         (ref_always_accessed_p): Likewise.
18414         (refs_independent_p): Likewise.
18415         (can_sm_ref_p): Likewise.
18417 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
18419         PR c/56566
18420         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
18421         return 1 even for !unsignedp.
18423 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
18425         * config/i386/i386.md (isa): Add x64 and nox64.
18426         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
18427         (*pushtf): Enable *roF alternative for x64 isa only.
18428         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
18429         mode attribute of integer alternatives to DImode for TARGET_64BIT.
18430         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
18431         (*movtf_internal): Merge from *movtf_internal_rex64 and
18432         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
18433         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
18434         nox64 isa attributes.
18435         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
18436         nox64 isa attributes.
18437         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
18439 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
18441         * config/alpha/alpha.c (TARGET_LRA_P): New define.
18443 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
18445         PR target/56640
18446         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
18447         class names.  Remove trailing comma after "ALL_REGS".
18449 2013-03-16  Jan Hubicka  <jh@suse.cz>
18451         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
18452         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
18453         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
18454         of cgraph_get_create_node.
18455         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
18457 2013-03-16  Jason Merrill  <jason@redhat.com>
18459         PR debug/49090
18460         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
18461         with DW_AT_default_value.
18463 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
18465         * BASE-VER: Set to 4.9.0.
18467 2013-03-14  Andi Kleen  <ak@linux.intel.com>
18469         PR target/56619
18470         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
18471         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
18472         Document _x* TSX intrinsics.
18474 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
18475             David Holsgrove  <david.holsgrove@xilinx.com>
18477         * configure.ac: Add MicroBlaze TLS support detection.
18478         * configure: Regenerate.
18479         * config/microblaze/microblaze-protos.h
18480         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
18481         symbol_mentioned_p, label_mentioned_p): Add prototypes.
18482         * config/microblaze/microblaze.c (microblaze_address_type): Add
18483         ADDRESS_TLS and tls_reloc address types.
18484         (microblaze_address_info): Add tls_reloc.
18485         (TARGET_HAVE_TLS): Define.
18486         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
18487          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
18488          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
18489         load_tls_operand,  microblaze_call_tls_get_addr,
18490         microblaze_legitimize_tls_address): New functions.
18491         (microblaze_classify_unspec): Handle UNSPEC_TLS.
18492         (get_base_reg): Use microblaze_tls_symbol_p.
18493         (microblaze_classify_address): Handle TLS.
18494         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
18495         label_mentioned_p and microblaze_tls_referenced_p.
18496         (microblaze_legitimize_address): Handle TLS.
18497         (microblaze_address_insns): Handle ADDRESS_TLS.
18498         (pic_address_needs_scratch): Handle TLS.
18499         (print_operand_address): Handle TLS.
18500         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
18501         (microblaze_expand_move): Handle TLS.
18502         (microblaze_legitimate_constant_p): Check
18503         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
18504         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
18505         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
18506         (PIC_OFFSET_TABLE_REGNUM): Set.
18507         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
18508         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
18509         (addsi3, movsi_internal2, movdf_internal): Update constraints
18510         * config/microblaze/predicates.md (arith_plus_operand): Define
18511         (move_operand): Redefine as move_src_operand,
18512         check microblaze_tls_referenced_p.
18514 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
18516         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
18517         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
18519 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
18521         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
18522         CC mode for AND.
18524 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
18526         PR tree-optimization/53265
18527         * common.opt (Waggressive-loop-optimizations): New option.
18528         * tree-ssa-loop-niter.c: Include tree-pass.h.
18529         (do_warn_aggressive_loop_optimizations): New function.
18530         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
18531         if number_of_latch_executions returned constant.
18532         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
18533         early.  If number_of_latch_executions returned constant, set
18534         nb_iterations_upper_bound back to it.
18535         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
18536         field.
18537         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
18538         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
18540         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
18541         (MULTILIB_OSDIRNAMES): Set.
18542         * genmultilib: If defaultosdirname doesn't start with :: , set
18543         defaultosdirname2 instead, clear it and emit two . multilib_raw
18544         entries instead of just one.
18546 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
18548         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
18549         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
18550         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
18551         (SUBTARGET_OVERRIDE_OPTIONS): New.
18553 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
18555         PR target/49880
18556         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
18557         (musermode): Convert to Var(TARGET_USERMODE).
18558         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
18559         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
18560         * config/sh/sh.c (sh_option_override): Use
18561         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
18562         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
18563         condition.
18564         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
18565         TARGET_SH4.
18566         (udivsi3_i4_single, divsi3_i4_single): Use
18567         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
18569 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
18571         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
18572         default setting.
18574 2013-03-13  Richard Biener  <rguenther@suse.de>
18576         PR tree-optimization/56608
18577         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
18578         calls when vectorizing basic-blocks.
18580 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
18582         PR plugins/45078
18583         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
18584         tm_file.
18586 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
18588         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
18590 2013-03-11  Jan Hubicka  <jh@suse.cz>
18592         PR lto/56557
18593         * lto-streamer-out.c (output_symbol_p): Skip references from
18594         constructors of external variables.
18596 2013-03-11  Jan Hubicka  <jh@suse.cz>
18598         PR middle-end/56571
18599         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
18600         from pseudos.
18601         * emit-rtl.c (verify_rtx_sharing): Likewise.
18602         (copy_insn_1): Likewise.
18603         * rtl.c (copy_rtx): Likewise.
18605 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
18607         PR target/56591
18608         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
18609         output_operand_lossage message.
18611 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
18613         PR target/56470
18614         * arm.c (shift_op): Validate RTL pattern on the fly.
18615         (arm_print_operand, case 'S'): Don't use shift_operator to validate
18616         the RTL.
18618 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18620         PR target/56347
18621         * config/pa/pa.md (call_value): Check for calls to powf and direct to
18622         new call patterns that clobber %fr12.
18623         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
18624         split and postreload patterns.
18625         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
18626         registers %fr12 and %fr12R as call used.
18628 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
18630         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
18631         (canon_address, record_store, replace_read, check_mem_read_rtx,
18632         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
18633         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
18634         rest_of_handle_dse): Likewise.
18636 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
18638         PR middle-end/56524
18639         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
18640         Add base_optabs.
18641         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
18642         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
18643         (save_optabs_if_changed): Replace with...
18644         (init_tree_optimization_optabs): ...this.
18645         * optabs.c (save_optabs_if_changed): Rename to...
18646         (init_tree_optimization_optabs): ...this.  Take the optimization node
18647         as argument.  Do nothing if the base optabs are already correct.
18648         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
18649         to recompute optabs.
18650         * function.h (function): Remove optabs field.
18651         * function.c (invoke_set_current_function_hook): Call
18652         init_tree_optimization_optabs.  Use the result to initialize
18653         this_fn_optabs.
18655 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
18657         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
18658         if GTMA_HAS_NO_INSTRUMENTATION.
18659         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
18660         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
18661         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
18662         * gimple-pretty-print.c (dump_gimple_transaction): Handle
18663         GTMA_HAS_NO_INSTRUMENTATION.
18665 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
18667         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
18668         libasan_preinit.o.
18670 2013-03-08  Marek Polacek  <polacek@redhat.com>
18671             Jakub Jelinek  <jakub@redhat.com>
18673         PR tree-optimization/56478
18674         * predict.c (is_comparison_with_loop_invariant_p): Change the
18675         type of loop_step to tree.
18676         (predict_loops): Adjust.
18677         (predict_iv_comparison): Perform the computations on double_ints.
18679 2013-03-08  Richard Biener  <rguenther@suse.de>
18681         PR tree-optimization/56570
18682         * tree-cfg.c (verify_expr_location_1): Verify locations for
18683         DECL_DEBUG_EXPR.
18684         * tree-sra.c (create_access_replacement): Strip locations
18685         from DECL_DEBUG_EXPRs.
18687 2013-03-08  Richard Biener  <rguenther@suse.de>
18689         * tree-inline.c (expand_call_inline): Do not associate
18690         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
18691         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
18693 2013-03-08  Richard Biener  <rguenther@suse.de>
18695         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
18696         or block changes with -Og.  Fix for location / block encoding
18697         changes and PHI arguments with locations.
18699 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
18701         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
18702         for all counters.
18703         (struct output_info): Likewise.
18704         (register_overhead): Remove bad gcc_assert.
18705         (bitmap_find_bit): If there is only a single bitmap element, do not
18706         count a miss as a search.
18707         (print_statistics): Update for counter type changes.
18708         (dump_bitmap_statistics): Likewise.  Print headers such that they
18709         are properly lined up with the printed counters.
18711 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
18713         PR tree-optimization/56559
18714         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
18715         check that it has only a single use.
18717 2013-03-07  Richard Biener  <rguenther@suse.de>
18719         * doc/invoke.texi (fwhole-program): Discourage use in combination
18720         with -flto.
18722 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
18724         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
18726         PR tree-optimization/56539
18727         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
18728         instead of GSI_CONTINUE_LINKING as last argument to
18729         force_gimple_operand_gsi.  Adjust function comment.
18731         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
18732         aarch64-cores.def.
18734         PR middle-end/56548
18735         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
18736         promoted mode, convert the result back to the original mode.
18738 2013-03-06  Richard Biener  <rguenther@suse.de>
18740         PR middle-end/56294
18741         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
18742         (insert_updated_phi_nodes_compare_uids): New function.
18743         (update_ssa): Sort symbols_to_rename after UID before
18744         traversing it to insert PHI nodes.
18746 2013-03-06  Richard Biener  <rguenther@suse.de>
18748         PR middle-end/50494
18749         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
18750         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
18752         Revert
18753         2013-02-13  Richard Biener  <rguenther@suse.de>
18755         PR lto/50494
18756         * varasm.c (output_constant_def_1): Get the decl representing
18757         the constant as argument.
18758         (output_constant_def): Wrap output_constant_def_1.
18759         (make_decl_rtl): Use output_constant_def_1 with the decl
18760         representing the constant.
18761         (build_constant_desc): Optionally re-use a decl already
18762         representing the constant.
18763         (tree_output_constant_def): Adjust.
18765 2013-03-06  Joey Ye  <joey.ye@arm.com>
18767         PR lto/50293
18768         * gcc.c (convert_white_space): New function.
18769         (main): Handles white space in function name.
18771 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
18773         PR target/56529
18774         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
18775         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
18776         to SH_DIV_CALL_TABLE for TARGET_SH2.
18777         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
18778         list.
18779         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
18780         call-table options.
18782 2013-03-05  Sterling Augustine  <saugustine@google.com>
18783             Cary Coutant  <ccoutant@google.com>
18785         PR debug/55364
18786         * dwarf2out.c (resolve_addr): Don't call
18787         remove_loc_list_addr_table_entries a second time for the same
18788         expression.
18790 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18792         PR debug/56510
18793         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
18794         (avoid_complex_debug_insns): New function.
18795         (expand_debug_locations): Call it.
18797         PR rtl-optimization/56484
18798         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
18799         lifetimes of hard registers on small register class machines.
18801 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
18803         * config/microblaze/microblaze-protos.h: Rename
18804         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
18805         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
18806         fast_interrupt.
18807         (microblaze_fast_interrupt_function_p): New function.
18808         (microblaze_is_interrupt_handler): Rename to
18809         microblaze_is_interrupt_variant and add fast_interrupt check.
18810         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
18811         (save_restore_insns): Likewise.
18812         (compute_frame_size): Likewise.
18813         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
18814         (microblaze_globalize_label): Likewise.
18815         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
18816         * config/microblaze/microblaze.md: Use wrapper
18817         microblaze_is_interrupt_variant.
18819 2013-03-05  Kai Tietz  <ktietz@redhat.com>
18821         * sdbout.c (sdbout_one_type): Switch to current function's section
18822         supporting cold/hot.
18824 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
18826         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
18827         -mxl-reorder.
18829 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18831         PR middle-end/56461
18832         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
18833         if VALGRIND_GET_VBITS is defined, temporarily make object
18834         memory all defined, and restore previous valgrind addressability
18835         and definability afterwards.  Free this_object at the end.
18837         PR middle-end/56461
18838         * lra.c (lra): Call lra_clear_live_ranges if live_p,
18839         right before calling lra_create_live_ranges, also call it
18840         when clearing live_p.  Only call lra_clear_live_ranges
18841         at the end if live_p.
18843         PR middle-end/56461
18844         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
18846 2013-03-05  Richard Biener  <rguenther@suse.de>
18848         PR tree-optimization/56521
18849         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
18850         value-id.
18852 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
18854         PR c++/55135
18855         * except.h (remove_unreachable_eh_regions): New prototype.
18856         * except.c (remove_eh_handler_splicer): New function, split out
18857         of remove_eh_handler.
18858         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
18859         warning about running it on many EH regions one at a time.
18860         (remove_unreachable_eh_regions_worker): New function, walk the
18861         EH tree in depth-first order and remove non-marked regions.
18862         (remove_unreachable_eh_regions): New function.
18863         * tree-eh.c (mark_reachable_handlers): New function, split out
18864         from remove_unreachable_handlers.
18865         (remove_unreachable_handlers): Use mark_reachable_handlers and
18866         remove_unreachable_eh_regions.
18867         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
18868         and remove_unreachable_eh_regions.
18870 2013-03-05  Richard Biener  <rguenther@suse.de>
18872         PR middle-end/56525
18873         * loop-init.c (fix_loop_structure): Remove loops in two stages,
18874         not freeing them until the end.
18876 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
18878         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
18880 2013-03-05  Richard Biener  <rguenther@suse.de>
18882         PR tree-optimization/56270
18883         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
18884         of loads after scheduling an SLP instance.
18886 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18888         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
18889         tic6x.exp.
18890         (check_gcc_parallelize): Run guality.exp as a separate job from
18891         vect.exp with unsorted.exp and $(dg_target_exps) separately from
18892         struct-layout-1.exp with stackalign.exp.
18894         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
18896         PR middle-end/56461
18897         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
18898         load_index sbitmap even if some bit in it isn't set.
18900         PR middle-end/56461
18901         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
18902         (discover_iteration_bound_by_body_walk): Change queues to
18903         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
18904         spelling in comment.  Call safe_push on queues[bound_index] directly.
18905         Release queues[queue_index] in every iteration unconditionally.
18906         Release bounds vector.
18908         PR middle-end/56461
18909         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
18910         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
18911         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
18912         inner_phis vector.
18914 2013-03-05  Richard Biener  <rguenther@suse.de>
18916         PR lto/56515
18917         * tree-inline.c (remap_blocks_to_null): New function.
18918         (expand_call_inline): When expanding a call stmt without
18919         an associated block inline remap all callee blocks to NULL.
18921 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18923         PR rtl-optimization/56494
18924         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
18925         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
18926         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
18928         PR middle-end/56461
18929         * sel-sched-ir.c (free_sched_pools): Release
18930         succs_info_pool.stack[succs_info_pool.max_top] vectors too
18931         if succs_info_pool.max_top isn't -1.
18933         PR bootstrap/56509
18934         * opts.c (opts_obstack, opts_concat): Moved to...
18935         * opts-common.c (opts_obstack, opts_concat): ... here.
18937 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
18939         PR middle-end/56461
18940         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
18942 2013-03-04  Martin Jambor  <mjambor@suse.cz>
18944         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
18945         all appropriate places.
18947 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
18949         PR tree-optimization/56424
18950         * ipa-split.c (split_function): Do not set the RSO flag if result is
18951         not by reference and its type is a register type.
18953 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
18955         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
18956         (microblaze_legitimate_pic_operand): Likewise
18957         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
18958         new function microblaze_legitimate_pic_operand
18959         * config/microblaze/microblaze-protos.h
18960         (microblaze_legitimate_pic_operand): Declare.
18962 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18964         * config/microblaze/predicates.md (call_insn_simple_operand):
18965         New predicate for supported rtx code types.
18966         * config/microblaze/microblaze.md (call_internal1): Use
18967         call_insn_simple_operand predicate.
18969 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
18971         PR middle-end/56461
18972         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
18973         partitions.ordered_remove.
18975         PR middle-end/56461
18976         * tree-vect-stmts.c (vectorizable_conversion): Don't call
18977         vec_oprnds0.create (1) for modifier == NONE.
18979         PR middle-end/56461
18980         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
18981         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
18982         vec_oprnds1 right before pushing anything to it for
18983         scalar_shift_arg.
18985         PR middle-end/56461
18986         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
18987         set nbbs to 0 instead of having separate code path.
18988         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
18989         instead of false as last argument if returning NULL.
18991 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
18993         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
18994         the attribute is now called "target" instead of "option".
18995         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
18996         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
18997         attribute/pragma name for TARGET_OPTION_VALID_P and
18998         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
18999         * doc/tm.texi: Regenerated.
19001 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
19003         * config/microblaze/microblaze.c:
19004         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
19005         * config/microblaze/microblaze.h: Add -mxl-reorder to
19006         DRIVER_SELF_SPECS.
19007         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
19008         instructions emitted if TARGET_REORDER.
19009         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
19010         or 0 for -m/-mno case, but initialises as 2 to detect default use case
19011         separately.
19013 2013-03-01  Xinliang David Li  <davidxl@google.com>
19015         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
19016         walk length.
19018 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
19020         PR middle-end/56461
19021         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
19022         vector even when returning true.  Fix up function comment formatting.
19024         PR middle-end/56461
19025         * ira-build.c (ira_loop_nodes_count): New variable.
19026         (create_loop_tree_nodes): Initialize it.
19027         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
19029         PR middle-end/56461
19030         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
19031         method on dr_chain and result_chain.
19032         * tree-vect-stmts.c (vectorizable_store): Only call
19033         result_chain.create if j == 0.
19035         PR middle-end/56461
19036         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
19037         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
19038         before overwriting it.
19040 2013-03-01  Tobias Burnus  <burnus@net-b.de>
19042         * doc/extended.texi (C Extensions): Change order in @menu
19043         to match @node.
19044         (Other MIPS Built-in Functions): Move last MIPS entry before
19045         "picoChip Built-in Functions".
19046         (SH Built-in Functions): Move after RX Built-in Functions.
19047         * doc/gcc.texi (Introduction): Change order in @menu
19048         to match @node.
19049         * doc/md.texi (Constraints): Ditto.
19050         * gty.texi (Type Information): Ditto.
19051         (User-provided marking routines for template types): Make
19052         subsection.
19053         * doc/invoke.texi (AArch64 Options): Move before
19054         "Adapteva Epiphany Options".
19056 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
19057             Jakub Jelinek  <jakub@redhat.com>
19059         PR sanitizer/56454
19060         * asan.c (gate_asan): Lookup no_sanitize_address instead of
19061         no_address_safety_analysis attribute.
19062         * doc/extend.texi (no_address_safety_attribute): Rename to
19063         no_sanitize_address attribute, mention no_address_safety_analysis
19064         attribute as deprecated alias.
19066 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
19068         PR middle-end/56461
19069         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
19070         type to vec<vec<tree> > *.
19071         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
19072         to be vec<tree> instead of vec<tree> *, set vec_defs
19073         to vNULL and call vec_defs.create (number_of_vects), adjust other
19074         uses of vec_defs.
19075         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
19076         vectorizable_condition): Adjust vect_get_slp_defs callers.
19078 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
19080         * config/aarch64/aarch64.c
19081         (aarch64_float_const_representable): Remove unused variable.
19083 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
19085         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
19087 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
19089         * config/aarch64/aarch64-builtins.c
19090         (aarch64_init_simd_builtins): Make static.
19092 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
19094         * config/aarch64/aarch64.c
19095         (aarch64_simd_make_constant): Make static.
19097 2013-02-28  Martin Jambor  <mjambor@suse.cz>
19099         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
19100         with no initialization to the RHS of debug statements.
19102 2013-02-28  Martin Jambor  <mjambor@suse.cz>
19104         PR tree-optimization/56294
19105         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
19106         Adjust dumping.
19107         (get_access_replacement): Do not call create_access_replacement.
19108         Assert a replacement exists.
19109         (get_repl_default_def_ssa_name): Create the replacement declaration
19110         itself.
19112 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
19114         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
19115         final_end_function.
19117 2013-02-28  Marek Polacek  <polacek@redhat.com>
19119         PR rtl-optimization/56466
19120         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
19121         if we're changing a loop.
19122         (peel_loops_completely): Likewise.
19124 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
19126         PR c++/55813
19127         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
19129 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
19131         PR target/56445
19132         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
19133         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
19134         INTX_FTYPE_FX, FX_FTYPE_INTX.
19135         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
19137 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
19139         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
19140         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
19141         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
19142         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
19143         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
19144         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
19145         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
19146         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
19147         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
19148         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
19149         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
19150         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
19151         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
19152         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
19153         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
19154         (avrxmega6): Increase max flash segments from 5 to 6.
19155         * config/avr/t-multilib: Regenerate.
19156         * config/avr/avr-tables.opt: Regenerate.
19157         * doc/avr-mmcu.texi: Regenerate.
19159 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
19161         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
19162         (avr_device_to_arch): Rename to avr_device_to_ld.
19163         (avr_device_to_as): New prototype.
19164         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
19165         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
19166         * config/avr/driver-avr.c (avr_device_to_as): New.
19167         (avr_device_to_arch): Rename to avr_device_to_ld.
19169 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
19171         PR middle-end/56461
19172         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
19173         method on dr_chain and result_chain.
19175         PR middle-end/56461
19176         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
19177         pointer_set_destroy on not_executed_last_iteration.
19179         PR middle-end/56461
19180         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
19182         PR middle-end/56461
19183         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
19184         FOR_EACH_DEFINED_FUNCTION when freeing state.
19186         PR middle-end/56461
19187         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
19188         pool_free.
19189         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
19190         overwriting it.
19192         PR middle-end/56461
19193         * ipa-cp.c (decide_whether_version_node): Call vec_free on
19194         known_aggs[i].items and release known_aggs vector.
19196         PR middle-end/56461
19197         * ipa-reference.c (propagate): Free node_info even for alias nodes.
19199 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19201         * config/microblaze/microblaze.c (microblaze_emit_compare):
19202         Use xor for EQ/NE comparisions.
19203         * config/microblaze/microblaze.md (cstoresf4): Add constraints
19204         (cbranchsf4): Adjust operator to comparison_operator.
19206 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
19208         PR middle-end/56461
19209         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
19210         vector.
19211         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
19212         vec_safe_push, always update *slot.
19213         (redirect_edge_var_map_clear): Use vec_free.
19214         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
19215         (free_var_map_entry): Use vec_free.
19216         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
19217         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
19219 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
19221         PR middle-end/45472
19222         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
19223         when the may_trap_p bit of the exprs being merged differs.
19224         Reorder tests for speculativeness in the logical and operator.
19226 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
19228         * incpath.c (add_standard_paths): Use reconcat instead of concat
19229         where appropriate and avoid leaking memory.
19231         * opts.h: Include obstack.h.
19232         (opts_concat): New prototype.
19233         (opts_obstack): New declaration.
19234         * opts.c (opts_concat): New function.
19235         (opts_obstack): New variable.
19236         (init_options_struct): Call gcc_init_obstack on opts_obstack.
19237         (finish_options): Use opts_concat instead of concat
19238         and XOBNEWVEC instead of XNEWVEC.
19239         * opts-common.c (generate_canonical_option, decode_cmdline_option,
19240         generate_option): Likewise.
19241         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
19242         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
19244         PR target/56455
19245         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
19246         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
19248 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
19250         PR middle-end/56461
19251         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
19253 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
19255         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
19256         (arm_block_move_unaligned_straight): Likewise.
19257         (arm_adjust_block_mem): Likewise.
19259 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
19261         PR target/48901
19262         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
19263         temp, cond and label.
19264         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
19266         PR target/52500
19267         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
19268         * config/c6x/c6x.h (dbx_register_map): Update declaration.
19270         PR target/52501
19271         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
19272         of prologue/epilogue functions.
19274         PR target/52550
19275         * config/tilegx/tilegx.c (tilegx_expand_prologue):
19276         Remove unused variable cfa_offset.
19277         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
19279         PR target/54639
19280         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
19281         type promotion to unsigned.
19283         PR target/54640
19284         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
19285         for HOST_WIDE_INT of 32 bit / same size as int.
19286         (arm_block_move_unaligned_straight): Likewise.
19287         (arm_adjust_block_mem): Likewise.
19289         PR target/54662
19290         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
19291         ALL_CFLAGS.
19293 2013-02-26  Marek Polacek  <polacek@redhat.com>
19295         PR tree-optimization/56426
19296         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
19298 2013-02-26  Richard Biener  <rguenther@suse.de>
19300         PR target/56444
19301         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
19302         unused variable loops.
19304 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
19306         PR tree-optimization/56448
19307         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
19308         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
19309         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
19310         later operands of the references, or even first operand for
19311         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
19313         PR tree-optimization/56443
19314         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
19315         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
19316         to type_for_mode langhook.
19318 2013-02-25  Matt Turner  <mattst88@gmail.com>
19320         * doc/invoke.texi: Document r4700.
19322 2013-02-25  Richard Biener  <rguenther@suse.de>
19324         PR tree-optimization/56175
19325         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
19326         split out from ...
19327         (simplify_bitwise_binary): ... here.  Also guard the conversion
19328         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
19330 2013-02-25  Catherine Moore  <clm@codesourcery.com>
19332         Revert:
19333         2013-02-24  Catherine Moore  <clm@codesourcery.com>
19334             Maciej W. Rozycki  <macro@codesourcery.com>
19335             Tom de Vries  <tom@codesourcery.com>
19336             Nathan Sidwell  <nathan@codesourcery.com>
19337             Iain Sandoe  <iain@codesourcery.com>
19338             Nathan Froyd  <froydnj@codesourcery.com>
19339             Chao-ying Fu  <fu@mips.com>
19341         * doc/extend.texi: (micromips, nomicromips, nocompression):
19342         Document new function attributes.
19343         * doc/invoke.texi (minterlink-compressed, mmicromips,
19344         m14k, m14ke, m14kec): Document new options.
19345         (minterlink-mips16): Update documentation.
19346         * doc/md.texi (ZC, ZD): Document new constraints.
19347         * configure.ac (gcc_cv_as_micromips): Check if linker
19348         supports the .set micromips directive.
19349         * configure: Regenerate.
19350         * config.in: Regenerate.
19351         * config/mips/mips-tables.opt: Regenerate.
19352         * config/mips/micromips.md: New file.
19353         * constraints.md (ZC, AD): New constraints.
19354         * config/mips/predicates.md (movep_src_register): New predicate.
19355         (movep_src_operand): New predicate.
19356         (non_volatile_mem_operand): New predicate.
19357         * config/mips/mips.md (multimem): New type.
19358         (length): Differentiate between 17-bit and 18-bit branch offsets.
19359         (MOVEP1, MOVEP2): New mode iterator.
19360         (mov_<load>l): Use ZC constraint.
19361         (mov_<load>r): Likewise.
19362         (mov_<store>l): Likewise.
19363         (mov_<store>r): Likewise.
19364         (*branch_equality<mode>_inverted): Add microMIPS support.
19365         (*branch_equality<mode>): Likewise.
19366         (*jump_absolute): Likewise.
19367         (indirect_jump_<mode>): Likewise.
19368         (tablejump_<mode>): Likewise.
19369         (<optab>_internal): Likewise.
19370         (sibcall_internal): Likewise.
19371         (sibcall_value_internal): Likewise.
19372         (prefetch): Use constraint ZD.
19373         * config/mips/mips.opt (minterlink-compressed): New option.
19374         (minterlink-mips16): Now an alias for minterlink-compressed.
19375         (mmicromips): New option.
19376         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
19377         (compare_and_swap_12): Likewise.
19378         (sync_add<mode>): Likewise.
19379         (sync_<optab>_12): Likewise.
19380         (sync_old_<optab>_12): Likewise.
19381         (sync_new_<optab>_12): Likewise.
19382         (sync_nand_12): Likewise.
19383         (sync_old_nand_12): Likewise.
19384         (sync_new_nand_12): Likewise.
19385         (sync_sub<mode>): Likewise.
19386         (sync_old_add<mode>): Likewise.
19387         (sync_old_sub<mode>): Likewise.
19388         (sync_new_add<mode>): Likewise.
19389         (sync_new_sub<mode>): Likewise.
19390         (sync_<optab><mode>): Likewise.
19391         (sync_old_<optab><mode>): Likewise.
19392         (sync_new_<optab><mode>): Likewise.
19393         (sync_nand<mode>): Likewise.
19394         (sync_old_nand<mode>): Likewise.
19395         (sync_new_nand<mode>): Likewise.
19396         (sync_lock_test_and_set<mode>): Likewise.
19397         (test_and_set_12): Likewise.
19398         (atomic_compare_and_swap<mode>): Likewise.
19399         (atomic_exchange<mode>_llsc): Likewise.
19400         (atomic_fetch_add<mode>_llsc): Likewise.
19401         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
19402         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
19403         (umips_save_restore_pattern_p): Likewise.
19404         (umips_load_store_pair_p): Likewise.
19405         (umips_output_load_store_pair): Likewise.
19406         (umips_movep_target_p): Likewise.
19407         (umips_12bit_offset_address_p): Likewise.
19408         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
19409         (mips_base_mips16): Rename this...
19410         (mips_base_compression_flags): ...to this. Update all uses.
19411         (mips_attribute_table): Add micromips, nomicromips and nocompression.
19412         (mips_mips16_decl_p): Delete.
19413         (mips_nomips16_decl_p): Delete.
19414         (mips_get_compress_on_flags): New function.
19415         (mips_get_compress_off_flags): New function.
19416         (mips_get_compress_mode): New function.
19417         (mips_get_compress_on_name): New function.
19418         (mips_get_compress_off_name): New function.
19419         (mips_insert_attributes): Support multiple compression types.
19420         (mips_merge_decl_attributes): Likewise.
19421         (umips_12bit_offset_address_p): New function.
19422         (mips_start_function_definition): Emit .set micromips directive.
19423         (mips_call_may_need_jalx_p): New function.
19424         (mips_function_ok_for_sibcall): Add microMIPS support.
19425         (mips_print_operand_punctuation): Support short delay slots and
19426         compact jumps.
19427         (umips_swm_mask, umips_swm_encoding): New.
19428         (umips_build_save_restore): New function.
19429         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
19430         (was_mips16_p): Remove.
19431         (old_compression_mode): New.
19432         (mips_set_compression_mode): New function.
19433         (mips_set_current_function): Add microMIPS support.
19434         (mips_option_override): Likewise.
19435         (umips_save_restore_pattern_p): New function.
19436         (umips_output_save_restore): New function.
19437         (umips_load_store_pair_p_1): New function.
19438         (umips_load_store_pair_p): New function.
19439         (umips_output_load_store_pair_1): New function.
19440         (umips_output_load_store_pair): New function.
19441         (umips_movep_target_p) New function.
19442         (mips_prepare_pch_save): Add microMIPS support.
19443         * config/mips/mips.h (TARGET_COMPRESSION): New.
19444         (TARGET_CPU_CPP_BUILTINS): Update macro
19445         to use new compression flags and to support microMIPS.
19446         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
19447         (MIPS_ARCH_FLOAT_SPEC): Likewise.
19448         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
19449         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
19450         (ASM_SPEC): Support mmicromips and mno-micromips.
19451         (M16STORE_REG_P): New macro.
19452         (MIPS_CALL): Support TARGET_MICROMIPS.
19453         (MICROMIPS_J): New macro.
19454         (mips_base_mips16): Rename this...
19455         (mips_base_compression_flags): ...to this.
19456         (UMIPS_12BIT_OFFSET_P): New macro.
19457         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
19458         (MULTILIB_DIRNAMES): Likewise.
19460 2013-02-25  Tom de Vries  <tom@codesourcery.com>
19462         PR rtl-optimization/56131
19463         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
19464         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
19465         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
19467 2013-02-25  Tobias Burnus  <burnus@net-b.de>
19469         * doc/invoke.texi (-fsanitize=): Move from optimization
19470         to debugging options.
19472 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
19474         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
19476 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
19477             Alexander Monakov  <amonakov@ispras.ru>
19479         PR middle-end/56077
19480         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
19481         flush pending lists also on non-jumps.  Adjust comment.
19483 2013-02-24  Catherine Moore  <clm@codesourcery.com>
19484             Maciej W. Rozycki  <macro@codesourcery.com>
19485             Tom de Vries  <tom@codesourcery.com>
19486             Nathan Sidwell  <nathan@codesourcery.com>
19487             Iain Sandoe  <iain@codesourcery.com>
19488             Nathan Froyd  <froydnj@codesourcery.com>
19489             Chao-ying Fu  <fu@mips.com>
19491         * doc/extend.texi: (micromips, nomicromips, nocompression):
19492         Document new function attributes.
19493         * doc/invoke.texi (minterlink-compressed, mmicromips,
19494         m14k, m14ke, m14kec): Document new options.
19495         (minterlink-mips16): Update documentation.
19496         * doc/md.texi (ZC, ZD): Document new constraints.
19497         * configure.ac (gcc_cv_as_micromips): Check if linker
19498         supports the .set micromips directive.
19499         * configure: Regenerate.
19500         * config.in: Regenerate.
19501         * config/mips/mips-tables.opt: Regenerate.
19502         * config/mips/micromips.md: New file.
19503         * constraints.md (ZC, AD): New constraints.
19504         * config/mips/predicates.md (movep_src_register): New predicate.
19505         (movep_src_operand): New predicate.
19506         (non_volatile_mem_operand): New predicate.
19507         * config/mips/mips.md (multimem): New type.
19508         (length): Differentiate between 17-bit and 18-bit branch offsets.
19509         (MOVEP1, MOVEP2): New mode iterator.
19510         (mov_<load>l): Use ZC constraint.
19511         (mov_<load>r): Likewise.
19512         (mov_<store>l): Likewise.
19513         (mov_<store>r): Likewise.
19514         (*branch_equality<mode>_inverted): Add microMIPS support.
19515         (*branch_equality<mode>): Likewise.
19516         (*jump_absolute): Likewise.
19517         (indirect_jump_<mode>): Likewise.
19518         (tablejump_<mode>): Likewise.
19519         (<optab>_internal): Likewise.
19520         (sibcall_internal): Likewise.
19521         (sibcall_value_internal): Likewise.
19522         (prefetch): Use constraint ZD.
19523         * config/mips/mips.opt (minterlink-compressed): New option.
19524         (minterlink-mips16): Now an alias for minterlink-compressed.
19525         (mmicromips): New option.
19526         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
19527         (compare_and_swap_12): Likewise.
19528         (sync_add<mode>): Likewise.
19529         (sync_<optab>_12): Likewise.
19530         (sync_old_<optab>_12): Likewise.
19531         (sync_new_<optab>_12): Likewise.
19532         (sync_nand_12): Likewise.
19533         (sync_old_nand_12): Likewise.
19534         (sync_new_nand_12): Likewise.
19535         (sync_sub<mode>): Likewise.
19536         (sync_old_add<mode>): Likewise.
19537         (sync_old_sub<mode>): Likewise.
19538         (sync_new_add<mode>): Likewise.
19539         (sync_new_sub<mode>): Likewise.
19540         (sync_<optab><mode>): Likewise.
19541         (sync_old_<optab><mode>): Likewise.
19542         (sync_new_<optab><mode>): Likewise.
19543         (sync_nand<mode>): Likewise.
19544         (sync_old_nand<mode>): Likewise.
19545         (sync_new_nand<mode>): Likewise.
19546         (sync_lock_test_and_set<mode>): Likewise.
19547         (test_and_set_12): Likewise.
19548         (atomic_compare_and_swap<mode>): Likewise.
19549         (atomic_exchange<mode>_llsc): Likewise.
19550         (atomic_fetch_add<mode>_llsc): Likewise.
19551         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
19552         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
19553         (umips_save_restore_pattern_p): Likewise.
19554         (umips_load_store_pair_p): Likewise.
19555         (umips_output_load_store_pair): Likewise.
19556         (umips_movep_target_p): Likewise.
19557         (umips_12bit_offset_address_p): Likewise.
19558         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
19559         (mips_base_mips16): Rename this...
19560         (mips_base_compression_flags): ...to this. Update all uses.
19561         (mips_attribute_table): Add micromips, nomicromips and nocompression.
19562         (mips_mips16_decl_p): Delete.
19563         (mips_nomips16_decl_p): Delete.
19564         (mips_get_compress_on_flags): New function.
19565         (mips_get_compress_off_flags): New function.
19566         (mips_get_compress_mode): New function.
19567         (mips_get_compress_on_name): New function.
19568         (mips_get_compress_off_name): New function.
19569         (mips_insert_attributes): Support multiple compression types.
19570         (mips_merge_decl_attributes): Likewise.
19571         (umips_12bit_offset_address_p): New function.
19572         (mips_start_function_definition): Emit .set micromips directive.
19573         (mips_call_may_need_jalx_p): New function.
19574         (mips_function_ok_for_sibcall): Add microMIPS support.
19575         (mips_print_operand_punctuation): Support short delay slots and
19576         compact jumps.
19577         (umips_swm_mask, umips_swm_encoding): New.
19578         (umips_build_save_restore): New function.
19579         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
19580         (was_mips16_p): Remove.
19581         (old_compression_mode): New.
19582         (mips_set_compression_mode): New function.
19583         (mips_set_current_function): Add microMIPS support.
19584         (mips_option_override): Likewise.
19585         (umips_save_restore_pattern_p): New function.
19586         (umips_output_save_restore): New function.
19587         (umips_load_store_pair_p_1): New function.
19588         (umips_load_store_pair_p): New function.
19589         (umips_output_load_store_pair_1): New function.
19590         (umips_output_load_store_pair): New function.
19591         (umips_movep_target_p) New function.
19592         (mips_prepare_pch_save): Add microMIPS support.
19593         * config/mips/mips.h (TARGET_COMPRESSION): New.
19594         (TARGET_CPU_CPP_BUILTINS): Update macro
19595         to use new compression flags and to support microMIPS.
19596         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
19597         (MIPS_ARCH_FLOAT_SPEC): Likewise.
19598         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
19599         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
19600         (ASM_SPEC): Support mmicromips and mno-micromips.
19601         (M16STORE_REG_P): New macro.
19602         (MIPS_CALL): Support TARGET_MICROMIPS.
19603         (MICROMIPS_J): New macro.
19604         (mips_base_mips16): Rename this...
19605         (mips_base_compression_flags): ...to this.
19606         (UMIPS_12BIT_OFFSET_P): New macro.
19607         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
19608         (MULTILIB_DIRNAMES): Likewise.
19610 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
19612         PR target/52555
19613         * target-globals.c (save_target_globals): For init_reg_sets and
19614         target_reinit remporarily set this_fn_optabs to this_target_optabs.
19616 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
19618         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
19619         * config/aarch64/t-aarch64
19620         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
19622 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
19624         PR inline-asm/56148
19625         * lra-constraints.c (process_alt_operands): Reload operand
19626         conflicting with earlier clobber only if no more other conflicting
19627         operands.
19629 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
19631         PR sanitizer/56393
19632         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
19633         if not linking a shared library.
19635 2013-02-22  Seth LaForge  <sethml@google.com>
19637         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
19639 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
19641         * config/arm/arm.md (split for extendsidi): Update condition.
19642         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
19643         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
19644         (qhs_zextenddi_cstr): Likewise.
19646 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
19648         PR middle-end/56420
19649         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
19650         avoid signed wrapping.
19651         (expand_mult): Handle properly multiplication by
19652         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
19653         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
19654         in the compiler if coeff is HOST_WIDE_INT_MIN.
19655         (expand_divmod): Don't make ext_op1 static, change it's type to
19656         uhwi.  Avoid undefined behavior in -INTVAL (op1).
19658         PR rtl-optimization/50339
19659         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
19660         field.
19661         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
19662         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
19663         into splitting_ashiftrt field.
19664         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
19665         ASHIFTRT.
19666         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
19667         choices.
19669 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
19671         PR middle-end/56108
19672         * trans-mem.c (execute_tm_mark): Do not expand transactions that
19673         are sure to go irrevocable.
19675 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
19677         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
19678         scalars are valid operands.
19680 2013-02-21  Martin Jambor  <mjambor@suse.cz>
19682         PR tree-optimization/56310
19683         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
19684         only matching indices and non-negative final offsets.
19685         (intersect_aggregates_with_edge): Pass src_idx to
19686         agg_replacements_to_vector.  Pass src_idx insstead of index to
19687         intersect_with_agg_replacements.
19689 2013-02-21  Martin Jambor  <mjambor@suse.cz>
19691         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
19692         instead of hard-wired defaults.
19694 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
19696         * doc/invoke.texi (MIPS Options): Update documentation of the
19697         floating-point multiply-accumulate instruction restrictions.
19699 2013-02-21  Kostya Serebryany  <kcc@google.com>
19701         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
19702         asan_shadow_offset on x86_64 linux.
19704 2013-02-21  Richard Biener  <rguenther@suse.de>
19706         PR tree-optimization/56415
19707         Revert
19708         2013-02-11  Richard Biener  <rguenther@suse.de>
19710         PR tree-optimization/56273
19711         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
19712         first VRP run.
19714 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
19716         PR bootstrap/56258
19717         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
19718         instead of @itemx.
19720         PR inline-asm/56405
19721         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
19722         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
19724 2013-02-20  Jan Hubicka  <jh@suse.cz>
19726         PR tree-optimization/56265
19727         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
19728         when target is referenced for first time.
19730 2013-02-20  Richard Biener  <rguenther@suse.de>
19732         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
19733         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
19734         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
19735         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
19736         not return anything.
19737         (rename_ssa_copies): Do not remove unused locals.
19738         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
19739         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
19740         * passes.c (execute_function_todo): Do not schedule unused locals
19741         removal if cleanup_tree_cfg did something.
19742         * tree-ssa-live.c (remove_unused_locals): Dump statistics
19743         about the number of removed locals.
19745 2013-02-20  Richard Biener  <rguenther@suse.de>
19747         PR tree-optimization/56398
19748         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
19750 2013-02-20  Martin Jambor  <mjambor@suse.cz>
19752         PR tree-optimization/55334
19753         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
19754         restricted pointers to arrays.
19756 2013-02-20  Richard Biener  <rguenther@suse.de>
19757             Jakub Jelinek  <jakub@redhat.com>
19759         PR tree-optimization/56396
19760         * tree-ssa-ccp.c (n_const_val): New static variable.
19761         (get_value): Return NULL for SSA names we don't have a lattice
19762         entry for.
19763         (ccp_initialize): Initialize n_const_val.
19764         * tree-ssa-copy.c (n_copy_of): New static variable.
19765         (init_copy_prop): Initialize n_copy_of.
19766         (get_value): Return NULL_TREE for SSA names we don't have a
19767         lattice entry for.
19769 2013-02-20  Martin Jambor  <mjambor@suse.cz>
19771         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
19773 2013-02-20  Richard Biener  <rguenther@suse.de>
19775         * genpreds.c (write_lookup_constraint): Do not compare first
19776         letter of the constraint again.
19778 2013-02-20  Richard Biener  <rguenther@suse.de>
19780         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
19781         and ceil_log2.
19782         (get_use_iv_cost): Terminate hashtable walk when coming across
19783         an empty entry.
19785 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
19787         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
19788         reassociation for avx2 targets.
19790 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19792         * config/microblaze/microblaze.c: microblaze_has_clz = 0
19793         Add version check for v8.10.a to enable microblaze_has_clz
19794         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
19795         version and TARGET_PATTERN_COMPARE check
19796         * config/microblaze/microblaze.md: New clzsi2 instruction
19798 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19800         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
19801         function before branching.
19803 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
19805         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
19806         DUMP_INSN_RTX_UID.
19807         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
19809 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
19811         PR middle-end/55889
19812         * sel-sched.c: Include ira.h.
19813         (implicit_clobber_conflict_p): New function.
19814         (moveup_expr): Use it.
19815         * Makefile.in (sel-sched.o): Depend on ira.h.
19817 2013-02-19  Richard Biener  <rguenther@suse.de>
19819         PR tree-optimization/56384
19820         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
19821         (vn_hash_type): Split out from ...
19822         (vn_hash_constant_with_type): ... here.
19823         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
19824         (vn_phi_eq): Compare types from vn_phi_s structure.
19825         (vn_phi_lookup): Populate vn_phi_s type.
19826         (vn_phi_insert): Likewise.
19828 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
19830         PR tree-optimization/56350
19831         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
19832         if haven't found reduction or nested cycle operand, rather than
19833         asserting we must find it.
19835         PR tree-optimization/56381
19836         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
19837         to fold_build3.
19839 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
19840             Jakub Jelinek  <jakub@redhat.com>
19842         PR target/52555
19843         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
19844         (swap_optab_enable): Same.
19845         (init_all_optabs): Use argument instead of global.
19846         * tree.h (struct tree_optimization_option): New field target_optabs.
19847         * expr.h (init_all_optabs): Add argument to prototype.
19848         (TREE_OPTIMIZATION_OPTABS): New.
19849         (save_optabs_if_changed): Protoize.
19850         * optabs.h: Declare this_fn_optabs.
19851         * optabs.c (save_optabs_if_changed): New.
19852         Declare this_fn_optabs.
19853         (init_optabs): Add argument to init_all_optabs() call.
19854         * function.c (invoke_set_current_function_hook): Handle per
19855         function optabs.
19856         * function.h (struct function): New field optabs.
19857         * config/mips/mips.c (mips_set_mips16_mode): Handle when
19858         optimization_current_node has changed.
19859         * target-globals.h (save_target_globals_default_opts): Protoize.
19860         * target-globals.c (save_target_globals_default_opts): New.
19862 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19864         PR target/56347
19865         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
19866         registers %fr12 and %fr12R as call used.
19868         PR target/56214
19869         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
19870         and HImode, require all displacements to be an integer multiple of
19871         their mode size.
19872         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
19873         only allow QImode and HImode when reload is in progress and strict is
19874         true.  Likewise for symbolic addresses.  Use base14_operand to check
19875         displacements in REG+BASE addresses.
19877 2013-02-18  Richard Biener  <rguenther@suse.de>
19879         PR tree-optimization/56366
19880         * tree-vect-loop.c (get_initial_def_for_induction): Properly
19881         handle sign-conversion of outer-loop initial induction value.
19883 2013-02-18  Richard Biener  <rguenther@suse.de>
19885         PR middle-end/56349
19886         * cfghooks.c (merge_blocks): If we merge a latch into another
19887         block adjust references to it.
19888         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
19889         (verify_loop_structure): Verify that a recorded latch is in fact
19890         a latch.
19892 2013-02-18  Richard Biener  <rguenther@suse.de>
19894         PR tree-optimization/56321
19895         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
19896         order SSA name release and virtual operand unlinking.
19898 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19900         * config/microblaze/microblaze.md (save_stack_block): Define.
19901         (restore_stack_block): Likewise.
19903 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19905         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
19906         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
19907         * config/microblaze/microblaze.c (microblaze_option_override):
19908         Bail out early for PIC modes when target does not support PIC.
19910 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19912         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
19913         Replace with a microblaze version.
19914         (microblaze_trampoline_init): Adapt for microblaze.
19915         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
19916         microblaze.
19918 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
19919             Dodji Seketeli  <dodji@redhat.com>
19921         PR asan/56330
19922         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
19923         (instrument_mem_region_access): Do not forget to always put
19924         instrumentation of the of 'base' and 'base + len' in a "if (len !=
19925         0) statement, even for cases where either 'base' or 'base + len'
19926         are not instrumented -- because they have been previously
19927         instrumented.  Simplify the logic by putting all the statements
19928         instrument 'base + len' inside a sequence, and then insert that
19929         sequence right before the current insertion point.  Then, to
19930         instrument 'base + len', just get an iterator on that statement.
19931         And do not forget to update the pointer to iterator the function
19932         received as argument.
19934 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
19936         PR rtl-optimization/56348
19937         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
19939 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
19941         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
19942         (clean_graph_dump_file): Pass base to start_graph_dump.
19944 2013-02-14  Richard Henderson  <rth@redhat.com>
19946         PR target/55941
19947         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
19949 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
19951         * collect2-aix.h: Define F_LOADONLY.
19953 2013-02-14  Richard Biener  <rguenther@suse.de>
19955         PR lto/50494
19956         * varasm.c (output_constant_def_1): Get the decl representing
19957         the constant as argument.
19958         (output_constant_def): Wrap output_constant_def_1.
19959         (make_decl_rtl): Use output_constant_def_1 with the decl
19960         representing the constant.
19961         (build_constant_desc): Optionally re-use a decl already
19962         representing the constant.
19963         (tree_output_constant_def): Adjust.
19965 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
19967         Fix an asan crash
19968         * asan.c (instrument_builtin_call):  Really put the length of the
19969         second source argument into src1_len.
19971 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
19973         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
19974         argument.  If it is false, don't create edge from then_bb to
19975         fallthru_bb.
19976         (insert_if_then_before_iter): Pass true to it.
19977         (build_check_stmt): Pass false to it.
19978         (transform_statements): Flush hash table only on extended basic
19979         block boundaries, rather than at the beginning of every bb.
19980         Don't flush hash table on nonfreeing_call_p calls.
19981         * tree-flow.h (nonfreeing_call_p): New prototype.
19982         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
19984 2013-02-13  David S. Miller  <davem@davemloft.net>
19986         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
19988 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
19990         PR target/56184
19991         * ira.c (max_regno_before_ira): Move from ...
19992         (ira): ... here.
19993         (fix_reg_equiv_init): Use max_regno_before_ira instead of
19994         vec_safe_length.
19996 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
19998         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
20000 2013-02-13  Richard Biener  <rguenther@suse.de>
20002         PR lto/56295
20003         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
20004         globals in MEM_REFs.
20006 2013-02-13  Richard Biener  <rguenther@suse.de>
20008         * loop-init.c (loop_optimizer_init): Clear loop state when
20009         re-initializing preserved loops.
20010         * loop-unswitch.c (unswitch_single_loop): Return whether
20011         we unswitched the loop.  Do not verify loop state here.
20012         (unswitch_loops): When we unswitched a loop discover new loops.
20014 2013-02-13  Kostya Serebryany  <kcc@google.com>
20016         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
20017         on x86_64 linux.
20018         * sanitizer.def: Rename __asan_init to __asan_init_v1.
20020 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
20022         Avoid instrumenting duplicated memory access in the same basic block
20023         * Makefile.in (asan.o): Add new dependency on hash-table.h
20024         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
20025         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
20026         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
20027         (free_mem_ref_resources, has_mem_ref_been_instrumented)
20028         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
20029         (get_mem_ref_of_assignment): New functions.
20030         (get_mem_refs_of_builtin_call): Extract from
20031         instrument_builtin_call and tweak a little bit to make it fit with
20032         the new signature.
20033         (instrument_builtin_call): Use the new
20034         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
20035         of is_gimple_builtin_call.
20036         (instrument_derefs, instrument_mem_region_access): Insert the
20037         instrumented memory reference into the hash table.
20038         (maybe_instrument_assignment): Renamed instrument_assignment into
20039         this, and change it to advance the iterator when instrumentation
20040         actually happened and return true in that case.  This makes it
20041         homogeneous with maybe_instrument_assignment, and thus give a
20042         chance to callers to be more 'regular'.
20043         (transform_statements): Clear the memory reference hash table
20044         whenever we enter a new BB, when we cross a function call, or when
20045         we are done transforming statements.  Use
20046         maybe_instrument_assignment instead of instrumentation.  No more
20047         need to special case maybe_instrument_assignment and advance the
20048         iterator after calling it; it's now handled just like
20049         maybe_instrument_call.  Update comment.
20051 2013-02-13  Richard Biener  <rguenther@suse.de>
20053         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
20054         Fix loop discovery code.
20056 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
20058         PR inline-asm/56148
20059         * lra-constraints.c (process_alt_operands): Match early clobber
20060         operand with itself.  Check conflicts with earlyclobber only if
20061         the operand is not reloaded.  Prefer to reload conflicting operand
20062         if earlyclobber and matching operands are the same.
20064 2013-02-12  Richard Biener  <rguenther@suse.de>
20066         PR lto/56297
20067         * lto-streamer-out.c (write_symbol): Do not output symbols
20068         for hard register variables.
20070 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
20072         PR target/54222
20073         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
20074         (umulsidi3_insn, mulsidi3_insn): New insns.
20076 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
20078         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
20079         (struct tune_params): Add vec_costs field.
20080         * config/arm/arm.c (arm_builtin_vectorization_cost)
20081         (arm_add_stmt_cost): New functions.
20082         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
20083         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
20084         (arm_default_vec_cost): New struct of type cpu_vec_costs.
20085         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
20086         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
20087         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
20088         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
20090 2013-02-12  Richard Biener  <rguenther@suse.de>
20092         PR lto/56295
20093         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
20094         decls again if possible.
20096 2013-02-12  Richard Biener  <rguenther@suse.de>
20098         PR middle-end/56288
20099         * tree-ssa.c (verify_ssa_name): Fix check, move
20100         SSA_NAME_IN_FREE_LIST check up.
20102 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
20103             Steven Bosscher   <steven@gcc.gnu.org>
20105         PR rtl-optimization/56151
20106         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
20107         equal to op0 or op1, and last_insn pattern is CODE operation
20108         with MEM dest and one of the operands matches that MEM.
20110 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
20112         * doc/extend.texi: Document Function Multiversioning and "default"
20113         parameter string to target attribute.
20114         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
20115         target attribute parameter is "default".
20116         (ix86_compare_version_priority): Remove checks for target attribute.
20117         (ix86_mangle_function_version_assembler_name): Change error to sorry.
20118         Remove check for target attribute equal to NULL. Add assert.
20119         (ix86_generate_version_dispatcher_body): Change error to sorry.
20121 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
20122             Jack Howarth  <howarth@bromo.med.uc.edu>
20123             Patrick Marlier  <patrick.marlier@gmail.com>
20125         PR libitm/55693
20126         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
20127         define ENDFILE_SPEC as TM_DESTRUCTOR.
20128         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
20130 2013-02-11  Alexander Potapenko  <glider@google.com>
20131             Jack Howarth  <howarth@bromo.med.uc.edu>
20132             Jakub Jelinek  <jakub@redhat.com>
20134         PR sanitizer/55617
20135         * config/darwin.c (cdtor_record): Rename ctor_record.
20136         (sort_cdtor_records): Rename sort_ctor_records.
20137         (finalize_dtors): New routine to sort destructors by
20138         priority before use in assemble_integer.
20139         (machopic_asm_out_destructor): Use finalize_dtors if needed.
20141 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
20143         PR rtl-optimization/56275
20144         * simplify-rtx.c (avoid_constant_pool_reference): Check that
20145         offset is non-negative and less than cmode size before
20146         calling simplify_subreg.
20148 2013-02-11  Richard Biener  <rguenther@suse.de>
20150         PR tree-optimization/56264
20151         * cfgloop.h (fix_loop_structure): Adjust prototype.
20152         * loop-init.c (fix_loop_structure): Return the number of
20153         newly discovered loops.
20154         * tree-cfgcleanup.c (repair_loop_structures): When new loops
20155         are discovered, do a full loop-closed SSA rewrite.
20157 2013-02-11  Richard Biener  <rguenther@suse.de>
20159         PR tree-optimization/56273
20160         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
20161         first VRP run.
20162         (check_array_ref): Fix missing newline in dumps.
20163         (search_for_addr_array): Likewise.
20165 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
20167         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
20169 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
20171         PR target/56256
20172         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
20174 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
20176         PR rtl-optimization/56246
20177         * lra-constraints.c (simplify_operand_subreg): Try to reuse
20178         reload pseudo.
20179         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
20180         constraints are satisfied.
20182 2013-02-08  Jeff Law  <law@redhat.com>
20184         PR debug/53948
20185         * emit-rtl.c (reg_is_parm_p): New function.
20186         * regs.h (reg_is_parm_p): New prototype.
20187         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
20188         callee-clobbered registers.
20190 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
20192         PR target/56043
20193         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
20194         If there is no implicit builtin declaration, just return NULL.
20196 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
20198         * config/i386/sse.md (FMAMODEM): New mode iterator.
20199         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
20200         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
20202 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
20204         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
20205         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
20206         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
20208 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20210         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
20211         (microblaze*-*-elf): Likewise.
20212         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
20213         LINK_SPEC.
20214         * config/microblaze/microblaze-c.c: Add builtin defines for
20215         _LITTLE_ENDIAN and _BIG_ENDIAN.
20216         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
20217         add to TARGET_DEFAULT flags.
20218         Expand ASM_SPEC and LINK_SPEC.
20219         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
20220         * config/microblaze/microblaze.md: Update extendsidi2 and
20221         movdi_internal instructions to use low-order / high-order reg
20222         print_operands.
20223         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
20224         options and inversemask / mask of LITTLE_ENDIAN.
20225         * config/microblaze/t-microblaze: Expand multilib options to
20226         include mlittle-endian (le) and update exceptions patterns.
20228 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
20230         PR rtl-optimization/56195
20231         * lra-constraints.c (get_reload_reg): Don't reuse regs
20232         if they have smaller mode than requested, if they have
20233         wider mode than requested, try to return a SUBREG.
20235         PR tree-optimization/56250
20236         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
20237         if type is unsigned and code isn't MULT_EXPR.
20239 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
20241         PR tree-optimization/56064
20242         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
20243         bits according to mode.
20244         * fixed-value.h (fixed_from_double_int)
20245         (const_fixed_from_double_int): Adjust comments.
20247 2013-02-08  Richard Biener  <rguenther@suse.de>
20249         PR lto/56231
20250         * lto-streamer.h (struct data_in): Remove current_file, current_line
20251         and current_col members.
20252         * lto-streamer-out.c (lto_output_location): Stream changed bits
20253         en-block for efficiency.
20254         * lto-streamer-in.c (clear_line_info): Remove.
20255         (lto_input_location): Cache current file, line and column
20256         globally via local statics.  Read changed bits en-block.
20257         (input_function): Do not call clear_line_info.
20258         (lto_read_body): Likewise.
20259         (lto_input_toplevel_asms): Likewise.
20261 2013-02-08  Michael Matz  <matz@suse.de>
20263         PR tree-optimization/52448
20264         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
20265         (nt_call_phase): New static.
20266         (add_or_mark_expr): Only mark accesses with newer phase than any
20267         call seen.
20268         (nonfreeing_call_p): New.
20269         (nt_init_block): Update nt_call_phase, mark blocks as visited.
20270         (nt_fini_block): Keep blocks marked as visited.
20271         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
20273 2013-02-08  Richard Biener  <rguenther@suse.de>
20275         * ira.c (ira): Free broken dominator information.
20277 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
20279         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
20281 2013-02-08  Marek Polacek  <polacek@redhat.com>
20283         * cfgloop.c (verify_loop_structure): Add more checking of headers.
20285 2013-02-08  Richard Biener  <rguenther@suse.de>
20287         PR middle-end/56181
20288         * cfgloop.h (flow_loops_find): Adjust.
20289         (bb_loop_header_p): Declare.
20290         * cfgloop.c (bb_loop_header_p): New function split out from ...
20291         (flow_loops_find): ... here.  Adjust function signature,
20292         support incremental loop structure update.
20293         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
20294         * cfgloopmanip.c (fix_loop_structure): Move ...
20295         * loop-init.c (fix_loop_structure): ... here.
20296         (apply_loop_flags): Split out from ...
20297         (loop_optimizer_init): ... here.
20298         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
20299         in incremental mode, only remove dead loops here.
20301 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
20303         PR target/54222
20304         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
20305         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
20306         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
20307         (*round<mode>3.libgcc): New insns for fixed-modes.
20308         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
20309         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
20310         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
20311         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
20312         implementations.  Define to __builtin_avr_absFX,
20313         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
20314         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
20315         __builtin_avr_countlsFX, respectively.
20316         * config/avr/avr-c.c (target.h): Include it.
20317         (enum avr_builtin_id): New enum.
20318         (avr_resolve_overloaded_builtin): New static function.
20319         (avr_register_target_pragmas): Use it to set
20320         targetm.resolve_overloaded_builtin.
20321         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
20322         tree nodes used by DEF_BUILTIN.
20323         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
20324         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
20325         <AVR_BUILTIN_xxBITS>: Same.
20327 2013-02-08  Richard Biener  <rguenther@suse.de>
20329         * cfgloop.c (verify_loop_structure): Properly handle
20330         a loop exiting to another loop header.
20331         * ira-int.h (ira_loops): Remove.
20332         * ira.c (ira_loops): Remove.
20333         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
20334         (do_reload): Use loop_optimizer_finalize.
20335         * ira-build.c (create_loop_tree_nodes): Use get_loops and
20336         number_of_loops to access the loop tree.
20337         (more_one_region_p): Likewise.
20338         (finish_loop_tree_nodes): Likewise.
20339         (rebuild_regno_allocno_maps): Likewise.
20340         (mark_loops_for_removal): Likewise.
20341         (mark_all_loops_for_removal): Likewise.
20342         (remove_unnecessary_regions): Likewise.
20343         (ira_build): Likewise.
20344         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
20346 2013-02-08  Richard Biener  <rguenther@suse.de>
20348         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
20349         * ipa-pure-const.c (analyze_function): Avoid calling
20350         mark_irreducible_loops twice.
20351         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
20353 2013-02-07  David S. Miller  <davem@davemloft.net>
20355         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
20356         on 'reg'.
20357         * var-tracking.c (vt_add_function_parameter): Test the presence of
20358         HAVE_window_save properly and do not remap argument registers when
20359         we have a leaf function.
20361 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
20363         PR bootstrap/56227
20364         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
20365         instead of "ll".
20366         * config/i386/i386.c (ix86_print_operand): Ditto.
20368 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
20370         * lra-constraints.c (process_alt_operands): Fix recently added comment.
20372 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
20374         PR rtl-optimization/56225
20375         * lra-constraints.c (process_alt_operands): Check that reload hard
20376         reg can hold value for strict_low_part.
20378 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
20380         PR debug/56154
20381         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
20382         dwarf2out_end_function.
20383         (in_first_function_p, maybe_at_text_label_p,
20384         first_loclabel_num_not_at_text_label): New variables.
20385         (dwarf2out_var_location): In the first function find out
20386         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
20387         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
20388         functions.
20390 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
20392         PR rtl-optimization/56178
20393         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
20394         SUBREG of a register.  Tidy up related block of code.
20395         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
20396         note if the source is a register or a SUBREG of a register.
20398 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
20400         PR target/56228
20401         * config/rs6000/rs6000.md (ptrm): New mode attr.
20402         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
20403         call_value_indirect_aix<pttrsize>,
20404         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
20405         m in constraints.
20407 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
20409         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
20410         if -bnortl. Convert to strcmp and strncmp.
20412 2013-02-07  Alan Modra  <amodra@gmail.com>
20414         PR target/54009
20415         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
20416         addresses won't wrap when offsetting.
20417         (rs6000_secondary_reload): Provide secondary reloads needed for
20418         wrapping LO_SUM addresses.
20420 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
20422         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
20423         MACH, just __MACH__.
20425 2013-02-06  Richard Biener  <rguenther@suse.de>
20427         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
20428         instead of calling fix_loop_structure.
20430 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
20432         PR middle-end/56217
20433         * omp-low.c (use_pointer_for_field): Return false if
20434         lower_send_shared_vars doesn't generate any copy-out code.
20436 2013-02-06  Tom de Vries  <tom@codesourcery.com>
20438         PR rtl-optimization/56131
20439         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
20440         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
20441         of the label is NULL.  Add comment.
20443 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
20445         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
20447         PR sanitizer/55374
20448         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
20449         (STATIC_LIBTSAN_LIBS): Likewise.
20450         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
20451         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
20452         is defined, don't add anything else beyond that.
20453         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
20454         (LINK_COMMAND_SPEC): Use them.
20456         PR tree-optimization/56205
20457         * tree-stdarg.c (check_all_va_list_escapes): Return true if
20458         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
20459         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
20461 2013-02-05  Richard Biener  <rguenther@suse.de>
20463         PR tree-optimization/53342
20464         PR tree-optimization/53185
20465         * tree-vectorizer.h (vect_check_strided_load): Remove.
20466         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
20467         not disallow peeling for vectorized strided loads.
20468         (vect_check_strided_load): Make static and simplify.
20469         (vect_analyze_data_refs): Adjust.
20470         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
20471         correctly when vectorizing strided loads.
20473 2013-02-05  Richard Biener  <rguenther@suse.de>
20475         * doc/install.texi: Refer to ISL, not PPL.
20477 2013-02-05  Jan Hubicka  <jh@suse.cz>
20479         PR tree-optimization/55789
20480         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
20482 2013-02-05  Jan Hubicka  <jh@suse.cz>
20484         PR tree-optimization/55789
20485         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
20486         the dead call anyway.
20488 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
20490         PR sanitizer/55374
20491         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
20493 2013-02-04  Alexander Potapenko  <glider@google.com>
20494             Jack Howarth  <howarth@bromo.med.uc.edu>
20495             Jakub Jelinek  <jakub@redhat.com>
20497         PR sanitizer/55617
20498         * config/darwin.c (sort_ctor_records): Stabilized qsort
20499         on constructor priority by using original position.
20500         (finalize_ctors): New routine to sort constructors by
20501         priority before use in assemble_integer.
20502         (machopic_asm_out_constructor): Use finalize_ctors if needed.
20504 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
20506         PR libstdc++/54314
20507         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
20508         about visibility on artificial decls.
20509         * config/sol2.c (solaris_assemble_visibility): Likewise.
20511 2013-02-04  Kai Tietz  <ktietz@redhat.com>
20513         PR target/56186
20514         * config/i386/i386.c (function_value_ms_64): Add additional valtype
20515         argument and improve checking of return-argument types for 16-byte
20516         modes.
20517         (ix86_function_value_1): Add additional valtype argument on call
20518         of function_value_64.
20519         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
20520         handling infunction_value_64 function.
20522 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
20524         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
20526 2013-02-04  Richard Biener  <rguenther@suse.de>
20528         PR tree-optimization/56188
20529         * tree-ssa-structalias.c (label_visit): Consider case with
20530         initially non-empty points-to set.
20531         (perform_var_substitution): Dump node mapping and clean up.
20533 2013-02-04  Richard Guenther  <rguenther@suse.de>
20535         PR lto/56168
20536         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
20537         node prevail as last resort.
20538         (lto_symtab_merge_decls): Remove guard on LTRANS here.
20539         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
20541 2013-02-04  Richard Biener  <rguenther@suse.de>
20543         PR tree-optimization/56113
20544         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
20545         Merge into ...
20546         (equiv_class_lookup_or_add): ... this.
20547         (label_visit): Adjust and fix error in previous patch.
20548         (perform_var_substitution): Adjust.
20550 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
20552         * config/sh/divtab.c: Fix formatting and comments throughout the file.
20553         * config/sh/sh4-300.md: Likewise.
20554         * config/sh/sh4a.md: Likewise.
20555         * config/sh/constraints.md: Likewise.
20556         * config/sh/sh.md: Likewise.
20557         * config/sh/netbsd-elf.h: Likewise.
20558         * config/sh/predicates.md: Likewise.
20559         * config/sh/sh-protos.h: Likewise.
20560         * config/sh/ushmedia.h: Likewise.
20561         * config/sh/linux.h: Likewise.
20562         * config/sh/sh.c: Likewise.
20563         * config/sh/superh.h: Likewise.
20564         * config/sh/elf.h: Likewise.
20565         * config/sh/sh4.md: Likewise.
20566         * config/sh/sh.h: Likewise.
20568 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20570         * config/pa/constraints.md: Adjust unused letters.  Change "T"
20571         constraint to match_test floating_point_store_memory_operand().
20572         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
20573         (base14_operand): New.
20574         (floating_point_store_memory_operand): New.
20575         (integer_store_memory_operand): Revise to use base14_operand and
20576         reg_plus_base_memory_operand.
20577         (move_dest_operand): Allow symbolic_memory_operands.
20578         (symbolic_memory_operand): Check for LO_SOM.
20579         (symbolic_operand): Change default case to break.
20580         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
20581         CONST_DOUBLE values to be reloaded by putting them into memory when
20582         the destination is a floating point register.
20583         (movdf): Remove code to handle CONST_DOUBLE.
20584         (movsf): Likewise.
20585         (reload_indf_r1): New.
20586         (reload_insf_r1): New.
20587         Consistently use "Q" and "T" constraints with integer and floating
20588         point move instructions, respectively.
20589         (movdi): Remove FAIL.
20590         Change predicate for source operand unamed DImode move from
20591         general_operand to move_src_operand.
20592         (umulsidi3): Change predicate for destination operand to
20593         register_operand.
20594         Likewise for similar unamed patterns.
20595         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
20596         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
20597         (hppa_legitimize_address): Simplify mask calculation.
20598         (pa_emit_move_sequence): Revised handling of secondary reloads from
20599         REG+D addresses for floating point loads and stores.  Directly handle
20600         loading CONST0_RTX (mode) to a floating point register.
20601         (pa_secondary_reload): Handle reloading DF and SFmode constant values
20602         to floating point registers.  Don't restrict secondary reloads to
20603         floating point registers to integer modes.  Revise some comments and
20604         cleanup some code.
20605         (TARGET_LEGITIMATE_ADDRESS_P): Define.
20606         (pa_legitimate_address_p): New.
20607         (pa_legitimize_reload_address): New.
20608         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
20609         (STRICT_REG_OK_FOR_BASE_P): New.
20610         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
20611         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
20613 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
20614             Andrew Dixie  <andrewd@gentrack.com>
20616         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
20617         flag set.
20619 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
20621         * expmed.c (extract_bit_field_1): Pass the full width of the
20622         structure to get_best_reg_extraction_insn.
20624 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
20626         PR target/54601
20627         * configure.ac (use_cxa_atexit): Add AIX.
20628         * configure: Regenerate.
20630         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
20632 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
20634         PR debug/54793
20635         * final.c (need_profile_function): New variable.
20636         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
20637         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
20638         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
20639         notes, targetm.asm_out.function_prologue doesn't emit anything,
20640         HAVE_prologue and profiler should be emitted before prologue,
20641         set need_profile_function instead of emitting it.
20642         (final_scan_insn): If need_profile_function, emit
20643         profile_function on the first NOTE_INSN_BASIC_BLOCK or
20644         NOTE_INSN_FUNCTION_BEG note.
20646 2013-02-01  Richard Henderson  <rth@redhat.com>
20648         * config/rs6000/rs6000.md (smulditi3): New.
20649         (umulditi3): New.
20651         * config/alpha/alpha.md (umulditi3): New.
20653 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
20655         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
20656         (ASM_OUTPUT_ALIGNED_LOCAL): New.
20658 2013-02-01  Richard Biener  <rguenther@suse.de>
20660         PR tree-optimization/56113
20661         * tree-ssa-structalias.c (label_visit): Reduce work for
20662         single-predecessor nodes.
20664 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
20666         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
20667         range isn't testing for zero.
20669 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
20671         PR middle-end/56113
20672         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
20674 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
20675             Nick Clifton  <nickc@redhat.com>
20677         * config/v850/constraints.md (Q): Define as a memory constraint.
20678         * config/v850/predicates.md (label_ref_operand): New predicate.
20679         (e3v5_shift_operand): New predicate.
20680         (ior_operator): New predicate.
20681         * config/v850/t-v850: Add e3v5 multilib.
20682         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
20683         (v850_gen_movdi): Prototype.
20684         * config/v850/v850.c: Add support for e3v5 architecture.
20685         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
20686         TARGET_V850E_UP.
20687         (construct_save_jarl): Add e3v5 long JARL support.
20688         (v850_adjust_insn_length): New function.  Adjust length of call
20689         insns when using e3v5 instructions.
20690         (v850_gen_movdi): New function: Generate instructions to move a
20691         DImode value.
20692         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
20693         (CPP_SPEC): Define __v850e3v5__ as appropriate.
20694         (TARGET_USE_FPU): Enable for e3v5.
20695         (CONST_OK_FOR_W): New macro.
20696         (ADJUST_INSN_LENGTH): Define.
20697         * config/v850/v850.md (UNSPEC_LOOP): Define.
20698         (attr cpu): Add v850e3v5.
20699         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
20700         (movdi): New pattern.
20701         (movdi_internal): New pattern.
20702         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
20703         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
20704         (cstoresf4): Likewise.
20705         (cstoredf4): Likewise.
20706         (insv): New pattern.
20707         (rotlso3_a): New pattern.
20708         (rotlsi3_b): New pattern
20709         (rotlsi3_v850e3v5): New pattern.
20710         (doloop_begin): New pattern.
20711         (fix_loop_counter): New pattern.
20712         (doloop_end): New pattern.
20713         (branch_normal): Add e3v5 long branch support.
20714         (branch_invert): Likewise.
20715         (branch_z_normal): Likewise.
20716         (branch_z_invert): Likewise.
20717         (branch_nz_normal): Likewise.
20718         (branch_nz_invert): Likewise.
20719         (call_internal_short): Add e3v5 register-indirect JARL support.
20720         (call_internal_long): Likewise.
20721         (call_value_internal_short): Likewise.
20722         (call_value_internal_long): Likewise.
20723         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
20724         (mloop): New option.
20725         * config.gcc: Add support for configuring v840e3v5 target.
20726         * doc/invoke.texi: Document new v850 specific command line options.
20728 2013-01-31  Paul Koning  <ni1d@arrl.net>
20730         PR debug/55059
20731         PR debug/54508
20732         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
20733         children if parent is a class.
20734         (prune_unused_types_prune): Don't add DW_AT_declaration.
20736 2013-01-31  Richard Biener  <rguenther@suse.de>
20738         PR tree-optimization/56157
20739         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
20740         match up operand with SLP child.
20742 2013-01-31  Jason Merrill  <jason@redhat.com>
20744         PR debug/54410
20745         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
20746         parameters the first time.
20747         (gen_scheduled_generic_parms_dies): Check completeness here.
20749 2013-01-31  Richard Biener  <rguenther@suse.de>
20751         PR middle-end/53073
20752         * common.opt (faggressive-loop-optimizations): New flag,
20753         enabled by default.
20754         * doc/invoke.texi (faggressive-loop-optimizations): Document.
20755         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
20756         infer_loop_bounds_from_undefined by it.
20758 2013-01-31  Richard Biener  <rguenther@suse.de>
20760         PR tree-optimization/56150
20761         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
20762         visit virtual operands.
20763         (find_uses_to_rename_bb): Likewise.
20765 2013-01-31  Richard Biener  <rguenther@suse.de>
20767         PR tree-optimization/56150
20768         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
20769         mixed store non-store stmts.
20771 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
20773         PR sanitizer/55374
20774         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
20775         LIBASAN_EARLY_SPEC is defined.
20776         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
20777         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
20778         before %o.
20779         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
20781         PR c++/55742
20782         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
20783         invalid args instead of ICEing on it.
20784         (ix86_valid_target_attribute_tree): Return error_mark_node if
20785         ix86_valid_target_attribute_inner_p failed.
20786         (ix86_valid_target_attribute_p): Return false only if
20787         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
20788         target("default") attribute.
20789         (sorted_attr_string): Change argument from const char * to tree,
20790         merge in all target attribute arguments rather than just one.
20791         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
20792         instead of free.  Avoid using strcat.
20793         (ix86_mangle_function_version_assembler_name): Mangle
20794         target("default") as if no target attribute is present.  Adjust
20795         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
20796         instead of xmalloc and XDELETEVEC instead of free.
20797         (ix86_function_versions): Don't return true if one of the decls
20798         doesn't have target attribute.  If they don't and one of the decls
20799         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
20800         sorted_attr_string caller.  Use XDELETEVEC instead of free.
20801         (ix86_supports_function_versions): Remove.
20802         (make_name): Fix up formatting.
20803         (make_dispatcher_decl): Remove resolver_name and its initialization.
20804         Avoid leaking memory.
20805         (is_function_default_version): Return true if there is
20806         target("default") attribute rather than no target attribute at all.
20807         (make_resolver_func): Avoid leaking memory.
20808         (ix86_generate_version_dispatcher_body): Likewise.
20809         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
20810         * target.def (supports_function_versions): Remove.
20811         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
20812         * doc/tm.texi: Regenerated.
20814 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
20816         PR rtl-optimization/56144
20817         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
20818         for values with side effects.
20820 2013-01-30  Richard Biener  <rguenther@suse.de>
20822         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
20823         (sparseset_pop): Likewise.
20824         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
20825         to be able to use quick_push in the worker loop.
20827 2013-01-30  Marek Polacek  <polacek@redhat.com>
20829         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
20831 2013-01-30  Richard Biener  <rguenther@suse.de>
20833         PR lto/56147
20834         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
20836 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
20838         PR tree-optimization/56064
20839         * fixed-value.c (fixed_from_double_int): New function.
20840         * fixed-value.h (fixed_from_double_int): New prototype.
20841         (const_fixed_from_double_int): New static inline function.
20842         * fold-const.c (native_interpret_fixed): New static function.
20843         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
20844         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
20845         (native_encode_fixed): New static function.
20846         (native_encode_expr) <FIXED_CST>: Use it.
20847         (native_interpret_int): Move double_int worker code to...
20848         * double-int.c (double_int::from_buffer): ...this new static method.
20849         * double-int.h (double_int::from_buffer): Prototype it.
20851 2013-01-30  Richard Biener  <rguenther@suse.de>
20853         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
20854         New pointer-map and obstack.
20855         (init_alias_vars): Allocate pointer-map and obstack.
20856         (delete_points_to_sets): Free them.
20857         (find_what_var_points_to): Cache result.
20858         (find_what_p_points_to): Adjust for changed interface of
20859         find_what_var_points_to.
20860         (compute_points_to_sets): Likewise.
20861         (ipa_pta_execute): Likewise.
20863 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20865         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
20866         * configure: Regenerate.
20867         * config.in: Regenerate.
20868         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
20869         #nobits/#progbits if supported.
20871 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
20873         PR target/56121
20874         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
20875         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
20876         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
20878 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20880         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
20881         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
20883 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20885         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
20886         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
20888 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20890         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
20891         declaration.
20892         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
20893         * config/arm/cortex-a7.md: New bypasses using
20894         arm_mac_accumulator_is_result.
20896 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20898         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
20899         (cortex_a7_neon_mla): Likewise.
20900         (cortex_a7_fpfmad): New reservation.
20901         (cortex_a7_fpmacs): Use ffmas and update required units.
20902         (cortex_a7_fpmuld): Update required units and latency.
20903         (cortex_a7_fpmacd): Likewise.
20904         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
20905         (cortex_a7_neon). Likewise.
20906         (bypass) Update participating units.
20908 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20910         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
20911         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
20912         from fmac to ffma.
20913         * config/arm/vfp11.md (vfp_farith): Use ffmas.
20914         (vfp_fmul): Use ffmad.
20915         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
20916         (cortex_r4_fmacd): Use ffmad.
20917         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
20918         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
20919         (cortex_a9_fmacd): Use ffmad.
20920         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
20921         (cortex_a8_vfp_macd): Use ffmad.
20922         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
20923         (cortex_a5_fpmacd): Use ffmad.
20924         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
20925         (cortex_a15_vfp_macd): Use ffmad.
20926         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
20928 2013-01-29  Jason Merrill  <jason@redhat.com>
20930         PR libstdc++/54314
20931         * varasm.c (default_assemble_visibility): Don't warn about
20932         visibility on artificial decls.
20934 2013-01-29  Richard Biener  <rguenther@suse.de>
20936         PR tree-optimization/56113
20937         * tree-ssa-structalias.c (equiv_class_lookup): Also return
20938         the bitmap leader.
20939         (label_visit): Free duplicate bitmaps and record the leader instead.
20940         (perform_var_substitution): Adjust.
20942 2013-01-29  Richard Biener  <rguenther@suse.de>
20944         PR tree-optimization/55270
20945         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
20946         the CFG, schedule loops for fixup.
20948 2013-01-29  Nick Clifton  <nickc@redhat.com>
20950         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
20951         SP_REG.
20953 2013-01-28  Leif Ekblad  <leif@rdos.net>
20955         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
20956         * config/i386/i386.h (TARGET_RDOS): New macro.
20957         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
20958         * config/i386/i386.c (ix86_option_override_internal): For 64bit
20959         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
20960         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
20961         DEFAULT_LARGE_SECTION_THRESHOLD.
20962         * config/i386/i386.md (R14_REG, R15_REG): New constants.
20963         * config/i386/rdos.h: New file.
20964         * config/i386/rdos64.h: New file.
20966 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
20968         PR other/54814
20969         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
20970         TEST_HARD_REG_BIT.
20972 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
20974         PR rtl-optimization/56117
20975         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
20976         call cselib_lookup_from_insn on the MEM before calling
20977         add_insn_mem_dependence.
20979 2013-01-28  Richard Biener  <rguenther@suse.de>
20981         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
20982         to a stmt that didn't have one.
20983         (copy_phis_for_bb): Likewise for PHI arguments.
20984         (copy_debug_stmt): Likewise for debug stmts.
20986 2013-01-28  Richard Biener  <rguenther@suse.de>
20988         PR tree-optimization/56034
20989         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
20990         (partition_builtin_p): Adjust.
20991         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
20992         it is the last partition.
20993         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
20994         up the vertex for the definition.
20995         (classify_partition): Classify whether a partition is a
20996         PKIND_REDUCTION, thus has uses outside of the loop.
20997         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
20998         Merge all PKIND_REDUCTION partitions into the last partition.
20999         (tree_loop_distribution): Seed partitions from reductions as well.
21001 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
21003         PR tree-optimization/56125
21004         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
21005         pow(x,c) into sqrt(x) * powi(x, n/2) or
21006         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
21007         optimizing for size.
21008         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
21009         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
21010         integer.
21012         PR tree-optimization/56094
21013         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
21014         to UNKNOWN_LOCATION while gimplifying expr.
21016 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
21018         PR target/56114
21019         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
21020         operand 0 in movabs insn template for -masm=intel asm alternative.
21021         (*movabs<mode>_2): Ditto for operand 1.
21023 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
21025         PR target/54663
21026         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
21027         of microblaze-c.o
21029 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
21031         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
21032         tm_file.
21034 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21036         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
21037         Undef to avoid warning.
21039 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
21041         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
21042         * configure: Regenerate.
21044 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
21046         PR tree-optimization/56098
21047         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
21048         for stmts with volatile ops.
21049         (cond_store_replacement): Don't optimize if assign has volatile ops.
21050         (cond_if_else_store_replacement_1): Don't optimize if either
21051         then_assign or else_assign have volatile ops.
21052         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
21053         volatile ops.
21055 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
21057         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
21059 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
21061         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
21062         missing ':' in asm example.
21064 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
21066         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
21067         entries into lane and laneq entries.
21068         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
21069         Remove AdvSIMD scalar modes.
21070         (aarch64_sq<r>dmulh_laneq<mode>): New.
21071         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
21072         modes.
21073         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
21074         builtin implementations to relfect changes in RTL in aarch64-simd.md.
21075         * config/aarch64/iterators.md (VCOND): New.
21076         (VCONQ): New.
21078 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
21080         PR target/54222
21081         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
21082         Add NULL LIBNAME argument to existing definitions.
21083         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
21084         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
21085         * config/avr/avr.c (DEF_BUILTIN): Same.
21086         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
21087         (avr_expand_builtin): Expand to a vanilla call if a libgcc
21088         implementation is available (DECL_ASSEMBLER_NAME is set).
21089         (avr_fold_absfx): New static function.
21090         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
21091         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
21092         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
21093         AVR_BUILTIN_ABSLLK.
21094         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
21095         (abshk, absk, abslk, absllk): Provide as static inline functions.
21097 2013-01-25  Marek Polacek  <polacek@redhat.com>
21099         PR tree-optimization/56035
21100         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
21102 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
21104         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
21105         (*movtf_internal_rex64): Add (!o,C) alternative
21106         (*movxf_internal_rex64): Ditto.
21107         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
21109 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
21111         * doc/invoke.texi: fix typo.
21112         * doc/objc.texi: fix typo.
21114 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
21116         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
21117         for the first two alternatives.
21119 2013-01-24  Diego Novillo  <dnovillo@google.com>
21121         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
21122         (ggc-zone.o): Remove.
21123         * configure.ac: Remove option --with-gc.
21124         * configure: Re-generate.
21125         * doc/install.texi: Remove documentation for --with-gc.
21126         * gengtype.c (write_enum_defn): Remove.  Update all users.
21127         (write_Types_process_field): Remove generation of gt_e_* argument.
21128         (output_type_enum): Remove.  Update all users.
21129         (write_enum_defn): Remove.  Update all users.
21130         (enum alloc_zone): Remove.  Update all users.
21131         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
21132         * ggc-common.c (ggc_splay_alloc): Remove first argument.
21133         Update all callers.
21134         (struct ptr_data): Remove field TYPE.  Update all users.
21135         (gt_pch_note_object): Remove argument TYPE.  Update all users.
21136         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
21137         Update all users.
21138         * ggc-none.c (ggc_alloc_typed_stat): Remove.
21139         (struct alloc_zone): Remove.
21140         (ggc_internal_alloc_zone_stat): Remove.
21141         (ggc_internal_cleared_alloc_zone_stat): Remove.
21142         * ggc-page.c (ggc_alloc_typed_stat): Remove.
21143         (ggc_pch_count_object): Remove last argument.  Update all users.
21144         (ggc_pch_alloc_object): Remove last argument.  Update all users.
21145         (struct alloc_zone): Remove.
21146         * ggc-zone.c: Remove.
21147         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
21148         (struct alloc_zone): Remove.
21149         (ggc_alloc_typed_stat): Remove.
21150         (ggc_alloc_typed): Remove.
21151         (ggc_splay_alloc): Remove first argument.
21152         (rtl_zone): Remove.  Update all users.
21153         (tree_zone): Remove.  Update all users.
21154         (tree_id_zone): Remove.  Update all users.
21155         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
21156         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
21157         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
21158         * tree-ssanames.c: Remove references to zone allocator in comments.
21160 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
21162         * config/avr/avr.c (avr_out_fract): Make register numbers that
21163         might be outside of source operand signed.
21165 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
21167         * config/i386/constraints.md (Yf): New constraint.
21168         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
21169         of f constraint to conditionaly disable x87 register preferences.
21170         (*movdf_internal): Ditto.
21171         (*movsf_internal): Ditto.
21173 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
21175         PR inline-asm/55934
21176         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
21177         that have operands with impossible constraints.
21178         Add a FIXME for a speed-up opportunity.
21179         * lra-constraints.c (process_alt_operands): Verify that a class
21180         selected from constraints on asms is valid for the operand mode.
21181         (curr_insn_transform): Remove incorrect comment.
21183 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
21185         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
21186         TOC operand is a valid symbol ref in the constant pool.
21188 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
21190         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
21192 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
21194         PR target/54222
21195         * config/avr/stdfix.h: New file.
21196         * t-avr (stdfix-gcc.h): New rule to build it.
21197         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
21199 2013-01-23  Kostya Serebryany  <kcc@google.com>
21201         * config/darwin.h: remove dependency on
21202         CoreFoundation (asan on Mac OS).
21204 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
21206         PR target/49069
21207         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
21208         instead of cmpdi_operand for first comparison operand.
21209         Don't assert that comparison operands aren't both constants.
21211 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
21213         * doc/install.texi (Downloading the Source): Update references to
21214         downloading separate components.
21216 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
21218         * doc/extend.texi (__int128): Improve grammar.
21220 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
21222         PR target/56028
21223         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
21224         alternative to (o,r).
21225         (*movdi_internal_rex64): Remove (!o,n) alternative.
21226         (DImode immediate->memory splitter): Remove.
21227         (DImode immediate->memory peephole2): Remove.
21228         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
21229         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
21230         alternative to (!o,*r).
21231         (*movtf_internal_sse): New pattern.
21232         (*movxf_internal_rex64): New pattern.
21233         (*movxf_internal): Disable for TARGET_64BIT.
21234         (*movdf_internal_rex64): Remove (!o,F) alternative.
21236 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
21238         PR middle-end/56074
21239         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
21240         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
21241         * tree-vect-loop-manip.c (find_loop_location): Also ignore
21242         stmt locations where LOCATION_LOCUS of the stmt location is
21243         UNKNOWN_LOCATION or BUILTINS_LOCATION.
21245         PR target/55686
21246         * config/i386/i386.md (UNSPEC_STOS): New.
21247         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
21248         *strsetqi_1): Add UNSPEC_STOS.
21250 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
21252         PR c++/56067
21253         * doc/invoke.texi: Remove left over -Wsynth example.
21255 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
21257         PR tree-optimization/56051
21258         * fold-const.c (fold_binary_loc): Don't fold
21259         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
21260         a narrowing conversion, or widening conversion from signed
21261         to unsigned.
21263 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
21265         PR rtl-optimization/56023
21266         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
21267         dependent on debug instruction.
21269 2013-01-21  Martin Jambor  <mjambor@suse.cz>
21271         PR middle-end/56022
21272         * function.c (allocate_struct_function): Call
21273         invoke_set_current_function_hook earlier.
21275 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
21277         * reload1.c (init_reload): Only initialize reload_obstack
21278         during the first call.
21280 2013-01-21  Marek Polacek  <polacek@redhat.com>
21282         * cfgloop.c (verify_loop_structure): Fix up grammar.
21284 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
21286         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
21287         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
21289 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
21291         PR target/56058
21292         * config/arm/marvell-pj4.md: Update copyright year.
21293         Fix up use of alu to alu_reg and simple_alu_imm.
21295 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
21297         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
21299 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
21301         PR target/55433
21302         * lra-constraints.c (curr_insn_transform): Don't reuse original
21303         insn for secondary memory move when memory mode should be different.
21305 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21307         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
21308         atomic_storedi_1): New patterns.
21310 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
21312         btver2 pipeline descriptions.
21313         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
21314         descriptions.
21315         * config/i386/i386.md (btver2_decode): New type attributes.
21316         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
21317         type attributes.
21318         * config/i386/btver2.md: New file describing btver2 pipelines.
21320 2013-01-19  Andrew Pinski  <apinski@cavium.com>
21322         PR tree-optimization/52631
21323         * tree-ssa-sccvn (visit_use): Before looking up the original
21324         statement, try looking up the simplified expression.
21326 2013-01-19  Anthony Green  <green@moxielogic.com>
21328         * config/moxie/moxie.c (moxie_expand_prologue): Set
21329         current_function_static_stack_size.
21331 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
21333         PR tree-optimization/56029
21334         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
21335         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
21337 2013-01-18  Sharad Singhai  <singhai@google.com>
21339         PR tree-optimization/55995
21340         * dumpfile.c (dump_loc): Print location only if available.
21341         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
21343 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
21345         PR target/55433
21346         * lra-constraints.c (curr_insn_transform): Reuse original insn for
21347         secondary memory move.
21348         (inherit_reload_reg): Use rclass instead of cl for
21349         check_secondary_memory_needed_p.
21351 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
21353         PR middle-end/56015
21354         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
21355         the case where writing real complex part of target modifies op1.
21357 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
21359         * config/aarch64/aarch64-simd.md
21360         (aarch64_vcond_internal<mode>): Handle unordered cases.
21361         * config/aarch64/iterators.md (v_cmp_result): New.
21363 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
21364             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
21366         * config/arm/marvell-pj4.md: New file.
21367         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
21368         * config/arm/arm.md (generic_sched): Add marvell_pj4.
21369         (generic_vfp): Likewise.
21370         * config/arm/arm-cores.def: Add marvell-pj4.
21371         * config/arm/arm-tune.md: Regenerate.
21372         * config/arm/arm-tables.opt: Regenerate.
21373         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
21374         * doc/invoke.texi: Document marvell-pj4.
21376 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
21378         * config/aarch64/arm_neon.h: Map scalar types to standard types.
21380 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
21382         PR debug/54114
21383         PR debug/54402
21384         PR debug/49888
21385         * var-tracking.c (negative_power_of_two_p): New.
21386         (global_get_addr_cache, local_get_addr_cache): New.
21387         (get_addr_from_global_cache, get_addr_from_local_cache): New.
21388         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
21389         heading comment.
21390         (vt_stack_offset_p): Remove.
21391         (vt_canon_true_dep): Always canonicalize loc's address.
21392         (clobber_overlapping_mems): Make sure we have a MEM.
21393         (local_get_addr_clear_given_value): New.
21394         (val_reset): Clear local cached entries.
21395         (compute_bb_dataflow): Create and release the local cache.
21396         Disable duplicate MEMs clobbering.
21397         (emit_notes_in_bb): Clobber MEMs likewise.
21398         (vt_emit_notes): Create and release the local cache.
21399         (vt_initialize, vt_finalize): Create and release the global
21400         cache, respectively.
21401         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
21403 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
21405         PR libmudflap/53359
21406         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
21407         not found in the symtab.
21409 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
21411         PR debug/56006
21412         PR rtl-optimization/55547
21413         PR rtl-optimization/53827
21414         PR debug/53671
21415         PR debug/49888
21416         * alias.c (offset_overlap_p): New, factored out of...
21417         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
21418         the conservative special case for symbolic constants.  Don't
21419         adjust zero sizes on alignment.
21421 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
21423         PR rtl-optimization/52573
21424         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
21425         REG_UNUSED for the same register.
21427 2013-01-17  Richard Biener  <rguenther@suse.de>
21428             Marek Polacek  <polacek@redhat.com>
21430         PR rtl-optimization/55833
21431         * loop-unswitch.c (unswitch_loops): Move loop verification...
21432         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
21433         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
21434         Set it to true when we're removing a loop from hierarchy tree in
21435         an irreducible region.
21436         (fix_bb_placements): Adjust caller.
21437         (fix_loop_placements): Likewise.
21439 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
21441         * config/avr/builtins.def (DEF_BUILTIN): Factor out
21442         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
21443         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
21444         Remove ID.  Adjust comments.
21445         * config/avr/avr-c.c (avr_builtin_name): Remove.
21446         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
21447         * config/avr/avr.c (avr_tolower): New static function.
21448         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
21449         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
21450         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
21451         default expansion.
21453 2013-01-17  Jan Hubicka  <jh@suse.cz>
21455         PR tree-optimization/55273
21456         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
21458 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
21460         PR target/55981
21461         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
21462         store through atomic_store<mode>_1.
21463         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
21465 2013-01-17  Martin Jambor  <mjambor@suse.cz>
21467         PR tree-optimizations/55264
21468         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
21469         for virtual methods.
21470         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
21471         virtual methods before inlining is over.
21472         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
21473         virtual functions.
21474         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
21475         non-virtual.
21477 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
21479         PR rtl-optimization/56005
21480         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
21481         pending reads for prefetch.
21483 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
21485         * config/aarch64/aarch64.md
21486         (*cstoresi_neg_uxtw): New pattern.
21487         (*cmovsi_insn_uxtw): New pattern.
21488         (*<optab>si3_uxtw): New pattern.
21489         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
21490         (*<optab>si3_insn_uxtw): New pattern.
21491         (*bswapsi2_uxtw): New pattern.
21493 2013-01-16  Richard Biener  <rguenther@suse.de>
21495         * tree-inline.c (tree_function_versioning): Remove set but
21496         never used variable.
21498 2013-01-16  Richard Biener  <rguenther@suse.de>
21500         PR tree-optimization/55964
21501         * tree-flow.h (rename_variables_in_loop): Remove.
21502         (rename_variables_in_bb): Likewise.
21503         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
21504         (copy_loop_before): Adjust and delete update-ssa status.
21505         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
21506         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
21507         (rename_variables_in_loop): Remove.
21508         (slpeel_update_phis_for_duplicate_loop): Likewise.
21509         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
21510         use available cfg machinery instead of duplicating it.
21511         Update PHI nodes and perform poor-mans SSA update here.
21512         (slpeel_tree_peel_loop_to_edge): Adjust.
21514 2013-01-16  Richard Biener  <rguenther@suse.de>
21516         PR tree-optimization/54767
21517         PR tree-optimization/53465
21518         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
21519         (vrp_visit_phi_node): For PHI arguments coming via backedges
21520         drop all symbolical range information.
21521         (execute_vrp): Compute backedges.
21523 2013-01-16  Richard Biener  <rguenther@suse.de>
21525         * doc/install.texi: Update CLooG and ISL requirements to
21526         0.18.0 and 0.11.1.
21528 2013-01-16  Christian Bruel  <christian.bruel@st.com>
21530         PR target/55301
21531         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
21532         (broken_move): Handle UNSPECV_SP_SWITCH_B.
21533         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
21535 2013-01-16  DJ Delorie  <dj@redhat.com>
21537         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
21538         (UNSPECV_SP_SWITCH_E): New.
21539         (sp_switch_1): Change to an unspec.
21540         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
21541         replace $r15.
21543 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
21545         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
21546         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
21547         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
21548         (expand_mem_thread_fence): Ditto.
21549         (expand_mem_signal_fence): Ditto.
21550         (expand_atomic_load): Ditto.
21551         (expand_atomic_store): Ditto.
21553 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
21555         PR rtl-optimization/55547
21556         PR rtl-optimization/53827
21557         PR debug/53671
21558         PR debug/49888
21559         * alias.c (memrefs_conflict_p): Set sizes to negative after
21560         AND adjustments.
21562 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
21564         PR target/55940
21565         * function.c (thread_prologue_and_epilogue_insns): Always
21566         add crtl->drap_reg to set_up_by_prologue.set, even if
21567         stack_realign_drap is false.
21569 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
21571         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
21572         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
21573         *call): Fix indention.
21575 2013-01-15  Tom de Vries  <tom@codesourcery.com>
21577         PR target/55876
21578         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
21579         Update comment.
21581 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
21583         PR rtl-optimization/55153
21584         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
21586 2013-01-15  Martin Jambor  <mjambor@suse.cz>
21588         PR tree-optimization/55920
21589         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
21590         accesses as grp_to_be_debug_replaced.
21592 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
21594         PR tree-optimization/55920
21595         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
21596         there is non-useless type conversion needed from debug rhs to lhs,
21597         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
21599 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
21600             Mikael Pettersson  <mikpe@it.uu.se>
21602         PR target/43961
21603         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
21604         Thumb.
21605         (ASM_OUTPUT_CASE_LABEL): Remove.
21606         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
21607         * final.c (shorten_branches): Update alignment of labels before
21608         jump tables if CASE_VECTOR_SHORTEN_MODE.
21610 2013-01-15  Richard Biener  <rguenther@suse.de>
21612         PR bootstrap/55961
21613         * system.h: Do not include gmp.h for building host tools.
21615 2013-01-15  Richard Biener  <rguenther@suse.de>
21617         PR middle-end/55882
21618         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
21619         account for bitpos when computing alignment.
21621 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
21623         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
21624         (ix86_target_macros_internal): Likewise.
21626         * config/i386/i386.c (m_CORE2I7): Removed.
21627         (m_CORE_HASWELL): New macro.
21628         (m_CORE_ALL): Likewise.
21629         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
21630         (initial_ix86_arch_features): Likewise.
21631         (processor_target_table): Initializations for Core avx2.
21632         (cpu_names): New names "core-avx2".
21633         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
21634         PROCESSOR_CORE_HASWELL.
21635         (ix86_issue_rate): New case.
21636         (ia32_multipass_dfa_lookahead): Likewise.
21637         (ix86_sched_init_global): Likewise.
21639         * config/i386/i386.h (TARGET_HASWELL): New macro.
21640         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
21641         (processor_type): New PROCESSOR_HASWELL.
21643 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
21645         PR tree-optimization/55955
21646         * tree-vect-loop.c (vectorizable_reduction): Give up early on
21647         *SHIFT_EXPR and *ROTATE_EXPR codes.
21649         PR tree-optimization/48766
21650         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
21651         -ftrapv disable -fwrapv.
21653 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
21655         PR target/55974
21656         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
21657         etc. to 1 and not to __flash.
21658         Use LL suffix for __INT24_MAX__ with -mint8.
21659         Use ULL suffix for __UINT24_MAX__ with -mint8.
21661 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
21663         * config/avr/avr-arch.h
21664         (struct base_arch_s): Use typedef avr_arch_t instead.
21665         (struct arch_info_s): Use typedef avr_arch_info_t instead.
21666         (struct mcu_type_s): Use typedef avr_mcu_t instead.
21667         * config/avr/avr.c: Same.
21668         * config/avr/avr-devices.c: Same.
21669         * config/avr/driver-avr.c: Same.
21670         * config/avr/gen-avr-mmcu-texi.c: Same.
21671         * config/avr/avr-mcus.def: Adjust comment.
21673 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
21675         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
21676         * config/aarch64/iterators.md (VALLDI): New.
21678 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
21679             Andi Kleen  <ak@linux.intel.com>
21681         PR target/55948
21682         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
21683         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
21684         memmodel flag.
21686 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
21688         * config/avr/avr-stdint.h: Remove trailing blanks.
21689         * config/avr/avr-log.h: Same.
21690         * config/avr/avr-arch.h: Same.
21691         * config/avr/avr-devices.c: Same.
21692         * config/avr/avr-dimode.md: Same.
21693         * config/avr/predicates.md: Same.
21694         * config/avr/avr-c.c: Same.  And fix typo.
21696         * config/avr/avr-protos.h: Same.  And:
21697         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
21698         (init_cumulative_args): Rename to avr_init_cumulative_args.
21699         (expand_prologue): Rename to avr_expand_prologue.
21700         (expand_epilogue): Rename to avr_expand_epilogue.
21701         (adjust_insn_length): Rename to avr_adjust_insn_length.
21702         (notice_update_cc): Rename to avr_notice_update_cc.
21703         (final_prescan_insn): Rename to avr_final_prescan_insn.
21704         * config/avr/avr.c: Same.
21705         * config/avr/avr.h: Same.
21706         * config/avr/avr.md: Remove trailing blanks.
21707         (prologue): Use avr_expand_prologue.
21708         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
21710 2013-01-14  Richard Biener  <rguenther@suse.de>
21712         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
21713         verify_location, collect_subblocks): New functions.
21714         (verify_gimple_in_cfg): Verify that locations only reference
21715         BLOCKs in the functions BLOCK tree.
21717 2013-01-14  Richard Biener  <rguenther@suse.de>
21719         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
21720         PHI argument.
21721         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
21722         unshare reference.
21723         (insert_out_of_ssa_copy_on_edge): Likewise.
21724         (rewrite_close_phi_out_of_ssa): Likewise.
21725         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
21726         debug expressions.
21727         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
21728         propagated constants.
21729         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
21730         can not be shared.
21732 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
21734         * config/avr/avr-modes.def: Add GPL copyright notice.
21736 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
21738         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
21739         MEMMODEL_MASK to determine memory model.
21740         (atomic_store<mode>): Ditto from operands[2].
21741         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
21743 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
21745         PR fortran/55935
21746         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
21747         (fold_gimple_assign): Don't call unshare_expr here.
21748         (fold_ctor_reference): Call unshare_expr.
21750 2013-01-13  Terry Guo  <terry.guo@arm.com>
21752         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
21753         * doc/fragments.texi: Document MULTILIB_REUSE.
21754         * gcc.c (multilib_reuse): New internal spec.
21755         (set_multilib_dir): Also search multilib from multilib_reuse.
21756         * genmultilib (tmpmultilib3): Refactor code.
21757         (tmpmultilib4): Ditto.
21758         (multilib_reuse): New multilib argument.
21760 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
21762         * Makefile.in: Update copyright.
21764 2013-01-12  Tom de Vries  <tom@codesourcery.com>
21766         PR middle-end/55890
21767         * calls.c (expand_call): Check if arg_nr is valid.
21769 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
21771         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
21772         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
21773         documentation.  Add missing '__' in front of
21774         __builtin_ia32_packssdw256.
21776 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21778         PR target/55719
21779         * config/s390/s390.c (s390_preferred_reload_class): Do not return
21780         NO_REGS for larl operands.
21781         (s390_reload_larl_operand): Use s390_load_address instead of
21782         emit_move_insn.
21784 2013-01-11  Richard Biener  <rguenther@suse.de>
21786         * tree-cfg.c (verify_node_sharing_1): Split out from ...
21787         (verify_node_sharing): ... here.
21788         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
21790 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
21792         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
21793         Substitute TREECHECKING.
21794         * configure: Regenerate.
21795         * Makefile.in (TREECHECKING): New.
21797 2013-01-11  Richard Guenther  <rguenther@suse.de>
21799         PR tree-optimization/44061
21800         * tree-vrp.c (extract_range_basic): Compute zero as
21801         value-range for __builtin_constant_p of function parameters.
21803 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
21805         Update copyright years.
21807 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
21809         PR rtl-optimization/55672
21810         * lra-eliminations.c (mark_not_eliminable): Permit addition with
21811         const to be eliminable.
21813 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
21815         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
21816         * configure: Regenerate.
21818 2013-01-10  Richard Biener  <rguenther@suse.de>
21820         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
21822 2013-01-10  Richard Biener  <rguenther@suse.de>
21824         PR bootstrap/55792
21825         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
21826         locations for virtual PHI arguments.
21827         (rewrite_update_phi_arguments): Likewise.
21829 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
21831         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
21832         on to assembler.
21834 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
21836         PR tree-optimization/55921
21837         * tree-complex.c (expand_complex_asm): New function.
21838         (expand_complex_operations_1): Call it for GIMPLE_ASM.
21840 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21842         PR target/55718
21843         * config/s390/s390.c (s390_symref_operand_p)
21844         (s390_loadrelative_operand_p): Merge the two functions.
21845         (s390_check_qrst_address, print_operand_address): Add parameters
21846         to s390_loadrelative_operand_p invokation.
21847         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
21848         (s390_reload_larl_operand, s390_secondary_reload): Use
21849         s390_loadrelative_operand_p instead of s390_symref_operand_p.
21850         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
21852 2013-01-09  Mike Stump  <mikestump@comcast.net>
21854         * dse.c (record_store): Remove unnecessary assert.
21856 2013-01-09  Jan Hubicka  <jh@suse.cz>
21858         PR tree-optimization/55569
21859         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
21860         * cfgloop.h (scale_loop_profile): Likewise.
21862 2013-01-09  Jan Hubicka  <jh@suse.cz>
21864         PR lto/45375
21865         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
21866         functions.
21867         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
21869 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
21871         PR middle-end/55114
21872         * expr.h (maybe_emit_group_store): Declare.
21873         * expr.c (maybe_emit_group_store): New function.
21874         * builtins.c (expand_builtin_int_roundingfn): Call it.
21875         (expand_builtin_int_roundingfn_2): Likewise.
21877 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
21879         PR rtl-optimization/55829
21880         * lra-constraints.c (match_reload): Add code for absent output.
21881         (curr_insn_transform): Add code for reloads of matched inputs
21882         without output.
21884 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
21886         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
21887         attribute of movddup insn to DF.
21888         (*vec_interleave_lowv2df): Ditto.
21889         (vec_dupv2df): Ditto.
21891 2013-01-09  Jan Hubicka  <jh@suse.cz>
21893         PR tree-optimiation/55875
21894         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
21895         EVERY_ITERATION parameter.
21896         (number_of_iterations_exit): Check if exit is executed every iteration.
21897         (idx_infer_loop_bounds): Similarly here.
21898         (n_of_executions_at_most): Simplify
21899         to only test for cases where statement is dominated by the
21900         particular bound; handle correctly the "postdominance" test.
21901         (scev_probably_wraps_p): Use max loop iterations info
21902         as a global bound first.
21904 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
21905             Nick Clifton  <nickc@redhat.com>
21907         * config/v850/v850.md (cbranchsf4): New pattern.
21908         (cstoresf4): New pattern.
21909         (cbranchdf4): New pattern.
21910         (cstoredf4): New pattern.
21911         (movsicc): Disallow floating point comparisons.
21912         (cmpsf_le_insn): Fix order of operators.
21913         (cmpsf_lt_insn): Likewise.
21914         (cmpsf_eq_insn): Likewise.
21915         (cmpdf_le_insn): Likewise.
21916         (cmpdf_lt_insn): Likewise.
21917         (cmpdf_eq_insn): Likewise.
21918         (cmpsf_ge_insn): Use LE comparison.
21919         (cmpdf_ge_insn): Likewise.
21920         (cmpsf_gt_insn): Use LT comparison.
21921         (cmpdf_gt_insn): Likewise.
21922         (cmpsf_ne_insn): Delete pattern.
21923         (cmpdf_ne_insn): Delete pattern.
21924         * config/v850/v850.c (v850_gen_float_compare): Use
21925         gen_cmpdf_eq_insn for NE comparison.
21926         (v850_float_z_comparison_operator)
21927         (v850_float_nz_comparison_operator): Move from here ...
21928         * config/v850/predicates.md: ... to here.  Move GT and GE
21929         comparisons into v850_float_z_comparison_operator.
21930         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
21931         Delete prototype.
21932         (v850_float_nz_comparison_operator): Likewise.
21934 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21936         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
21937         with calls to gen_insvsi/gen_insvdi.
21939 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
21941         * config/i386/i386.c (initial_ix86_tune_features): Set up
21942         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
21944 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
21945             Jakub Jelinek  <jakub@redhat.com>
21947         PR tree-optimization/48189
21948         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
21949         If nitercst is 0, don't predict the exit edge.
21951 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21953         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
21954         in asm_fprintf with reg_names.
21955         (aarch64_print_operand_address): Likewise.
21956         (aarch64_return_addr): Likewise.
21957         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
21959 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21961         * config/pa/pa.h (VAL_U6_BITS_P): Define.
21962         (INT_U6_BITS): Likewise.
21963         * config/pa/predicates.md (uint6_operand): New predicate.
21964         (shift5_operand, shift6_operand): Likewise.
21965         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
21966         arith32_operand.
21967         (lshrdi3): Use shift6_operand.
21968         (shrpsi4, shrpdi4): New insn patterns.
21969         (extzv): Delete expander.
21970         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
21971         predicates in unamed zero extract patterns.  Tighten common constraint.
21972         (extv): Delete expander.
21973         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
21974         predicates in unamed sign extract patterns.  Tighten common constraint.
21975         (insv): Delete expander.
21976         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
21977         predicates in unamed insert patterns.  Tighten common constraint.
21978         Change uint32_operand predicate to uint6_operand predicate in unamed
21979         DImode pattern to insert constant values of type 1...1xxxx.
21981 2013-01-04  Jan Hubicka  <jh@suse.cz>
21983         PR tree-optimization/55823
21984         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
21985         issue.
21987 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
21988             Uros Bizjak  <ubizjak@gmail.com>
21990         PR rtl-optimization/55845
21991         * df-problems.c (can_move_insns_across): Stop scanning at
21992         volatile_insn_p source instruction or give up if
21993         across_from .. across_to range contains any volatile_insn_p
21994         instructions.
21996 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
21998         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
21999         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
22000         Declare.
22001         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
22002         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
22004 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
22006         PR fortran/55341
22007         * asan.c (asan_clear_shadow): New function.
22008         (asan_emit_stack_protection): Use it.
22010 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
22012         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
22013         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
22014         with tab instead of space.
22016 2013-01-08  Nick Clifton  <nickc@redhat.com>
22018         * config/rl78/rl78.c (rl78_expand_prologue): Always select
22019         register bank 0 at the start of an interrupt handler.
22020         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
22021         MDBH registers.
22023 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
22025         * config/aarch64/aarch64-simd.md
22026         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
22027         (aarch64_simd_bsl): Likewise.
22028         (aarch64_vcond_internal<mode>): Likewise.
22029         (vcond<mode><mode>): Likewise.
22030         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
22031         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
22033 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
22035         * config/aarch64/aarch64-builtins.c
22036         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
22038 2013-01-08  Martin Jambor  <mjambor@suse.cz>
22040         PR debug/55579
22041         * tree-sra.c (analyze_access_subtree): Return true also after
22042         potentially creating a debug-only replacement.
22044 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
22046         PR middle-end/55890
22047         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
22049         PR tree-optimization/54120
22050         * tree-vrp.c (range_fits_type_p): Don't allow
22051         src_precision < precision from signed vr to unsigned_p
22052         if vr->min or vr->max is negative.
22053         (simplify_float_conversion_using_ranges): Test can_float_p
22054         against CODE_FOR_nothing.
22056 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
22057             Richard Biener  <rguenther@suse.de>
22059         PR middle-end/55851
22060         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
22061         types instead of just INTEGER_TYPE types.
22063 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
22065         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
22066         TF_SIZE): Define.
22068 2013-01-07  Steve Ellcey  <sellcey@mips.com>
22070         PR target/42661
22071         * config/mips/mips.opt: Change mad to mmad to match documentation.
22073 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
22075         PR target/55897
22076         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
22077         .progmemx.data now.
22079 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
22081         PR target/55897
22082         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
22083         (avr_addrspace_t): Add .section_name field.
22084         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
22085         array size.
22086         (avr_addrspace): Same.  Initialize .section_name.  Remove last
22087         NULL entry.  Put __memx into .progmemx.data.
22088         (progmem_section_prefix): Remove.
22089         (avr_asm_init_sections): No need to initialize progmem_section.
22090         (avr_asm_named_section): Use avr_addrspace[].section_name to get
22091         section name prefix.
22092         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
22093         retrieve the progmem section.
22094         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
22095         boundary to run over avr_addrspace[].
22096         (avr_register_target_pragmas): Ditto.
22098 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
22100         * varasm.c (output_constant_def_contents): For asan_protect_global
22101         protected strings, adjust DECL_ALIGN if needed, before testing for
22102         anchored symbols.
22103         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
22104         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
22105         normal decls.
22106         (output_object_block): For asan protected decls, emit asan padding
22107         after their contents.
22108         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
22109         (asan_finish_file): Test it here instead.
22111 2013-01-07  Nick Clifton  <nickc@redhat.com>
22112             Matthias Klose  <doko@debian.org>
22113             Doug Kwan  <dougkwan@google.com>
22114             H.J. Lu  <hongjiu.lu@intel.com>
22116         PR driver/55470
22117         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
22119         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
22121         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
22123         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
22125         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
22127 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
22129         PR target/54461
22130         * doc/install.texi (Cross-Compiler-Specific Options): Document
22131         --with-avrlibc.
22133 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
22135         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
22136         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
22137         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
22138         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
22139         vqmovun_high_s64): Fix source operand number and update copyright.
22141 2013-01-07  Richard Biener  <rguenther@suse.de>
22143         PR middle-end/55890
22144         * gimple.h (gimple_call_builtin_p): New overload.
22145         * gimple.c (validate_call): New function.
22146         (gimple_call_builtin_p): Likewise.
22147         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
22148         Use gimple_call_builtin_p.
22149         (find_func_clobbers): Likewise.
22150         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
22151         (strlen_optimize_stmt): Likewise.
22153 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
22155         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
22156         (vld1q_dup_*): Likewise.
22157         (vld1_*): Likewise.
22158         (vld1q_*): Likewise.
22159         (vld1_lane_*): Likewise.
22160         (vld1q_lane_*): Likewise.
22162 2013-01-07  Richard Biener  <rguenther@suse.de>
22164         * lto-streamer.h (LTO_minor_version): Bump to 2.
22166 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
22168         * config/aarch64/aarch64-protos.h
22169         (aarch64_const_double_zero_rtx_p): Rename to...
22170         (aarch64_float_const_zero_rtx_p): ...this.
22171         (aarch64_float_const_representable_p): New.
22172         (aarch64_output_simd_mov_immediate): Likewise.
22173         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
22174         move immediate case.
22175         * config/aarch64/aarch64.c
22176         (aarch64_const_double_zero_rtx_p): Rename to...
22177         (aarch64_float_const_zero_rtx_p): ...this.
22178         (aarch64_print_operand): Allow printing of new constants.
22179         (aarch64_valid_floating_const): New.
22180         (aarch64_legitimate_constant_p): Check for valid floating-point
22181         constants.
22182         (aarch64_simd_valid_immediate): Likewise.
22183         (aarch64_vect_float_const_representable_p): New.
22184         (aarch64_float_const_representable_p): Likewise.
22185         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
22186         (aarch64_output_simd_mov_immediate): New.
22187         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
22188         (*movdf_aarch64): Likewise.
22189         * config/aarch64/constraints.md (Ufc): New.
22190         (Y): call aarch64_float_const_zero_rtx.
22191         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
22193 2013-01-07  Richard Biener  <rguenther@suse.de>
22195         PR tree-optimization/55888
22196         PR tree-optimization/55862
22197         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
22198         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
22199         not if it is contained therein.
22201 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
22203         * config/avr/t-avr: Typo.
22205 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
22207         PR55243
22208         * config/avr/t-avr: Don't automatically rebuild
22209         $(srcdir)/config/avr/t-multilib
22210         $(srcdir)/config/avr/avr-tables.opt
22211         $(srcdir)/doc/avr-mmcu.texi
22212         (avr-mcus): New phony target to build them on request.
22213         (s-avr-mlib, s-avr-mmcu-texi): Remove.
22214         * avr/avr-mcus.def: Adjust comments.
22216 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
22218         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
22220 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
22222         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
22224 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
22226         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
22228 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
22230         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
22231         to generate profiling.
22232         * config/rs6000/aix64.h (LIB_SPEC): Same.
22234 2013-01-04  Andrew Pinski  <apinski@cavium.com>
22236         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
22237         New function.
22238         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
22240 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
22242         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
22243         unconditionally.
22244         (ix86_expand_move): Ditto.
22245         (ix86_zero_extend_to_Pmode): Ditto.
22246         (ix86_expand_call): Ditto.
22247         (ix86_expand_special_args_builtin): Ditto.
22248         (ix86_expand_builtin): Ditto.
22250 2013-01-04  Richard Biener  <rguenther@suse.de>
22252         PR tree-optimization/55862
22253         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
22254         translating them through PHI nodes.
22256 2013-01-04  Martin Jambor  <mjambor@suse.cz>
22258         PR tree-optimization/55755
22259         * tree-sra.c (sra_modify_assign): Do not check that an access has no
22260         children when trying to avoid producing a VIEW_CONVERT_EXPR.
22262 2013-01-04  Marek Polacek  <polacek@redhat.com>
22264         PR middle-end/55859
22265         * opts.c (default_options_optimization): Clarify error message.
22267 2013-01-04  Richard Biener  <rguenther@suse.de>
22269         PR middle-end/55863
22270         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
22271         reassociation.
22273 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
22275         PR target/53789
22276         * config/pa/pa.md (movsi): Revert previous change.
22277         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
22278         references.
22280 2013-01-03  Richard Henderson  <rth@redhat.com>
22282         * config/i386/i386.c (ix86_expand_move): Always assign to op1
22283         after eliminating TLS symbols.
22285 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
22287         PR bootstrap/50167
22288         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
22289         * graphite-poly.c (debug_gmp_value): Likewise.
22291 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
22293         PR target/55712
22294         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
22295         selected code model, define __code_mode_small__, __code_model_medium__,
22296         __code_model_large__, __code_model_32__ or __code_model_kernel__.
22297         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
22298         xchg temporary register with %k.  Declare temporary register as
22299         early clobbered.
22300         [__x86_64__]: For medium and large code models, preserve %rbx register.
22302 2013-01-03  Richard Biener  <rguenther@suse.de>
22304         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
22305         (dump_subscript): Adjust.
22306         (finalize_ddr_dependent): Do not dump redundant info.
22307         (analyze_siv_subscript): Adjust.
22308         (subscript_dependence_tester): Likewise.
22309         (compute_affine_dependence): Likewise.
22311 2013-01-03  Richard Biener  <rguenther@suse.de>
22313         Revert
22314         2013-01-03  Richard Biener  <rguenther@suse.de>
22316         PR tree-optimization/55857
22317         * tree-vect-stmts.c (vectorizable_load): Do not setup
22318         re-alignment for invariant loads.
22320         2013-01-02  Richard Biener  <rguenther@suse.de>
22322         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
22323         invariant load do not generate a vector load from the scalar location.
22325 2013-01-03  Richard Biener  <rguenther@suse.de>
22327         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
22328         for not vectorizing.
22329         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
22330         not build INDIRECT_REFs, call get_name once only.
22331         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
22332         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
22334 2013-01-03  Richard Biener  <rguenther@suse.de>
22336         PR tree-optimization/55857
22337         * tree-vect-stmts.c (vectorizable_load): Do not setup
22338         re-alignment for invariant loads.
22340 2013-01-03  Richard Biener  <rguenther@suse.de>
22342         PR lto/55848
22343         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
22344         prefer a built-in decl.
22346 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
22348         * gcc.c (process_command): Update copyright notice dates.
22349         * gcov.c (print_version): Likewise.
22350         * gcov-dump.c (print_version): Likewise.
22352         PR rtl-optimization/55838
22353         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
22354         iv0.step, iv1.step and step.
22356 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
22357             Marc Glisse  <marc.glisse@inria.fr>
22359         PR tree-optimization/55832
22360         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
22361         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
22362         integer_{one,zero}_node.
22364 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
22366         PR debug/54402
22367         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
22368         * var-tracking.c (reverse_op): Don't add reverse ops to
22369         VALUEs that have already
22370         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
22372 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
22374         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
22376 2013-01-02  Teresa Johnson  <tejohnson@google.com>
22378         * dumpfile.c (dump_loc): Print filename with location.
22379         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
22380         new location_t parameter to emit complete unroll message with
22381         new dump framework.
22382         (canonicalize_loop_induction_variables): Compute loops location
22383         and pass to try_unroll_loop_completely.
22384         * loop-unroll.c (report_unroll_peel): New function.
22385         (peel_loops_completely): Use new dump format with location
22386         for main dumpfile message, and invoke report_unroll_peel on success.
22387         (decide_unrolling_and_peeling): Ditto.
22388         (decide_peel_once_rolling): Remove old dumpfile message subsumed
22389         by report_unroll_peel.
22390         (decide_peel_completely): Ditto.
22391         (decide_unroll_constant_iterations): Ditto.
22392         (decide_unroll_runtime_iterations): Ditto.
22393         (decide_peel_simple): Ditto.
22394         (decide_unroll_stupid): Ditto.
22395         * cfgloop.c (get_loop_location): New function.
22396         * cfgloop.h (get_loop_location): Declare.
22398 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
22400         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
22401         NULL.
22403 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
22405         PR middle-end/55198
22406         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
22407         BLKmode objects when EXPAND_MEMORY is specified.
22409 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
22411         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
22412         in loop predicate.
22413         (fold_builtin_cpu): Do not share cpu model decls across statements.
22415 2013-01-02  Jason Merrill  <jason@redhat.com>
22417         PR c++/55804
22418         * tree.c (build_array_type_1): Revert earlier change.
22420 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
22422         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
22423         "cortex-a57".
22424         * config/aarch64/aarch64-tune.md: Re-generate.
22426 2013-01-02  Richard Biener  <rguenther@suse.de>
22428         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
22429         invariant load do not generate a vector load from the scalar location.
22431 2013-01-02  Richard Biener  <rguenther@suse.de>
22433         PR bootstrap/55784
22434         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
22435         * configure: Regenerate.
22437 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
22439         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
22440         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
22441         (expand_builtin_int_roundingfn_2): Keep the original target around
22442         for the fallback case.
22444 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
22446         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
22447         to be clear for sign changes.
22449 2013-01-01  Jan Hubicka  <jh@suse.cz>
22451         * ipa-inline-analysis.c: Fix formatting.
22453 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
22455         PR tree-optimization/55831
22456         * tree-vect-loop.c (get_initial_def_for_induction): Use
22457         gsi_after_labels instead of gsi_start_bb.
22459 Copyright (C) 2013 Free Software Foundation, Inc.
22461 Copying and distribution of this file, with or without modification,
22462 are permitted in any medium without royalty provided the copyright
22463 notice and this notice are preserved.