[AArch64] Use plus_constant.
[official-gcc.git] / gcc / ChangeLog
blob68a0f2eb51d094d732da8b314c586f678c800805
1 2013-09-20  Renlin Li  <renlin.li@arm.com>
3         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
4         (aarch64_expand_epilogue): Likewise.
5         (aarch64_legitimize_reload_address): Likewise.
7 2013-09-20  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9         PR middle-end/57748
10         * expr.c (expand_assignment): Remove misalignp code path.
12 2013-09-20  Marek Polacek  <polacek@redhat.com>
14         PR sanitizer/58413
15         * ubsan.c (get_ubsan_type_info_for_type): Use TYPE_SIZE instead of
16         TYPE_PRECISION.  Add asserts.
18 2013-09-20  Richard Biener  <rguenther@suse.de>
20         PR tree-optimization/58453
21         * tree-loop-distribution.c (distribute_loop): Apply the cost
22         model for -ftree-loop-distribute-patterns, too.
24 2013-09-20  Richard Biener  <rguenther@suse.de>
26         PR middle-end/58473
27         * tree-chrec.h (build_polynomial_chrec): Use gcc_checking_assert,
28         make type comparison less strict.
30 2013-09-20  Alan Modra  <amodra@gmail.com>
32         * configure: Regenerate.
33         * aclocal.m4: Regenerate.
35 2013-09-20  Marek Polacek  <polacek@redhat.com>
37         PR other/58467
38         * doc/extend.texi: Document that attribute used is meant to be used
39         on variables with static storage duration.
41 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
43         PR tree-optimization/58472
44         * tree-vect-stmts.c (vectorizable_store, vectorizable_load): For
45         simd_lane_access set inv_p = false.
46         * omp-low.c (lower_rec_input_clauses): Set TREE_NO_WARNING on
47         the simduid magic VAR_DECL.
49 2013-09-19  Jan Hubicka  <jh@suse.cz>
51         * i386.c (generic_memcpy, generic_memset): Fix 32bit template.
53 2013-09-17  Jeff Law  <law@redhat.com>
55         * tree-ssa-dom.c (record_temporary_equivalences): New function
56         split out of dom_opt_dom_walker::after_dom_children.
57         (dom_opt_dom_walker::thread_across_edge): Move common code
58         in here from dom_opt_dom_walker::after_dom_children.
59         (dom_opt_dom_walker::after_dom_children): Corresponding simplifictions.
61 2013-09-19  Jan Hubicka  <jh@suse.cz>
63         * i386.h (TARGET_GENERIC32, TARGET_GENERIC64): Remove.
64         (TARGET_GENERIC): Use PROCESOR_GENERIC
65         (enum processor_type): Unify generic32 and 64.
66         * i386.md (cpu): Likewise.
67         * x86-tune.def (use_leave): Enable for generic32.
68         (avoid_vector_decode, slow_imul_imm32_mem, slow_imul_imm8): Likewise.
69         * athlon.md: Change generic64 to generic in all occurences.
70         * i386-c.c (ix86_target_macros_internal): Unify generic64 and 32.
71         (ix86_target_macros_internal): Likewise.
72         * driver-i386.c (host_detect_local_cpu): Likewise.
73         * i386.c (generic64_memcpy, generic64_memset, generic64_cost): Rename to ..
74         (generic_memcpy, generic_memset, generic_cost): This one.
75         (generic32_memcpy, generic32_memset, generic32_cost): Remove.
76         (m_GENERIC32, m_GENERIC64): Remove.
77         (m_GENERIC): Turn into one flag.
78         (processor_target): Unify generic tunnings.
79         (ix86_option_override_internal): Replace generic32/64 by generic.
80         (ix86_issue_rate): Likewise.
81         (ix86_adjust_cost): Likewise.
83 2013-09-19  Jan Hubicka  <jh@suse.cz>
85         * cgraph.c (cgraph_create_edge_1): Avoid uninitialized read
86         of speculative flag.
88 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
90         * omp-low.c (expand_omp_sections): Always pass len - 1 to
91         GOMP_sections_start, even if !exit_reachable.
93 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
95         * lra-constraints.c (need_for_all_save_p): Use macro
96         HARD_REGNO_CALL_PART_CLOBBERED.
97         * lra-lives.c (check_pseudos_live_through_calls): Use the macro to
98         set up pseudo conflict hard regs.
100 2013-09-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
102         PR target/58452
103         * config/rs6000/paired.md (movmisalignv2sf): Fix to allow memory
104         operands.
106 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
108         PR rtl-optimization/58438
109         * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
110         * lra-constraints.c (undo_optional_reloads): Keep optional reloads
111         from previous subpasses.
113 2013-09-18  Richard Earnshaw  <rearnsha@arm.com>
115         * arm.c (arm_get_frame_offsets): Validate architecture supports
116         LDRD/STRD before accepting the tuning preference.
117         (arm_expand_prologue): Likewise.
118         (arm_expand_epilogue): Likewise.
120 2013-09-18  Richard Biener  <rguenther@suse.de>
122         PR tree-optimization/58417
123         * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
124         have chrecs with symbols defined in the loop as operands.
125         (chrec_fold_multiply): Likewise.
126         * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
127         parameters before folding binary operations.
128         (struct instantiate_cache_entry_hasher): Remove.
129         (struct instantiate_cache_type): Use a pointer-map.
130         (instantiate_cache_type::instantiate_cache_type): New function.
131         (instantiate_cache_type::get): Likewise.
132         (instantiate_cache_type::set): Likewise.
133         (instantiate_cache_type::~instantiate_cache_type): Adjust.
134         (get_instantiated_value_entry): Likewise.
135         (global_cache): New global.
136         (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
137         instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
138         instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
139         (instantiate_scev_name): Adjust.
140         (instantiate_scev): Construct global instead of local cache.
141         (resolve_mixers): Likewise.
143 2013-09-18  Daniel Morris  <danielm@ecoscentric.com>
144             Paolo Carlini  <paolo.carlini@oracle.com>
146         PR c++/58458
147         * doc/implement-cxx.texi: Fix references to the C++ standards.
149 2013-09-18  Jakub Jelinek  <jakub@redhat.com>
151         * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
152         * tree-vect-data-refs.c (vect_analyze_data_refs): For
153         simd_lane_access drs, update also DR_ALIGNED_TO.
155 2013-09-18  Marek Polacek  <polacek@redhat.com>
157         PR sanitizer/58411
158         * doc/extend.texi: Document no_sanitize_undefined attribute.
159         * builtins.c (fold_builtin_0): Don't sanitize function if it has the
160         no_sanitize_undefined attribute.
162 2013-09-18  Nick Clifton  <nickc@redhat.com>
164         * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
165         (ASM_DECLARE_FUNCTION_NAME): Define.
167 2013-09-17  Trevor Saunders  <tsaunders@mozilla.com>
169         * compare-elim.c (find_comparison_dom_walker): New class
170         (find_comparisons_in_bb): Rename to
171         find_comparison_dom_walker::before_dom_children
172         (find_comparisons): Adjust
173         * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
174         adjust.
175         (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
176         * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
177         (init_walk_dominator_tree): Remove declaration.
178         (fini_walk_dominator_tree): Remove declaration.
179         * fwprop.c (single_def_use_dom_walker): New class
180         (single_def_use_enter_block): Convert to
181         single_def_use_dom_walker::before_dom_children.
182         (single_def_use_leave_block): Convert to
183         single_def_use_dom_walker::after_dom_children.
184         (build_single_def_use_links): Adjust.
185         * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
186         class.
187         (find_candidates_in_block): Convert to
188         find_candidates_dom_walker::before_dom_children.
189         (execute_strength_reduction): Adjust.
190         * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
191         (sese_dom_walker): New class.
192         (sese_dom_walker::sese_dom_walker): New constructor.
193         (sese_dom_walker::~sese_dom_walker): New destructor.
194         (build_sese_conditions_before): Convert to
195         sese_dom_walker::before_dom_children.
196         (build_sese_conditions_after): Convert to
197         sese_dom_walker::after_dom_children.
198         (build_poly_scop): Adjust
199         * tree-into-ssa.c (rewrite_dom_walker): New class
200         (rewrite_enter_block): Convert to
201         rewrite_dom_walker::before_dom_children.
202         (rewrite_leave_block): Convert to
203         rewrite_dom_walker::after_dom_children.
204         (rewrite_update_dom_walker): New class.
205         (rewrite_update_enter_block): Convert to
206         rewrite_update_dom_walker::before_dom_children.
207         (rewrite_update_leave_block): Convert to
208         rewrite_update_dom_walker::after_dom_children.
209         (rewrite_blocks, rewrite_into_ssa): Adjust.
210         (mark_def_dom_walker): New class.
211         (mark_def_dom_walker::mark_def_dom_walker): New constructor.
212         (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
213         (mark_def_sites_blocks): Convert to
214         mark_def_dom_walker::before_dom_children.
215         (mark_def_site_blocks): Remove.
216         * tree-ssa-dom.c (dom_opt_dom_walker): New class.
217         (tree_ssa_dominator_optimize): Adjust.
218         (dom_thread_across_edge): Convert to method
219         dom_opt_dom_walker::thread_across_edge.
220         (dom_opt_enter_block): Convert to member function
221         dom_opt_dom_walker::before_dom_children.
222         (dom_opt_leave_block): Convert to member function
223         dom_opt_dom_walker::after_dom_children.
224         * tree-ssa-dse.c (dse_dom_walker): New class.
225         (dse_enter_block): Convert to member function
226         dse_dom_walker::before_dom_children.
227         (tree_ssa_dse): Adjust.
228         * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
229         (determine_invariantness_stmt): Convert to method
230         invariantness_dom_walker::before_dom_children.
231         (determine_invariantness): Remove
232         (move_computations_dom_walker): New class.
233         (move_computations_stmt): Convert to method
234         move_computations_dom_walker::before_dom_children.
235         (move_computations, tree_ssa_lim): Adjust.
236         * tree-ssa-phiopt.c (nontrapping_dom_walker): new class
237         (nt_init_block): Make method
238         notrappping_dom_walker::before_dom_children.
239         (nt_fini_block): Make
240         method nontrapping_dom_walker::after_dom_children.
241         (get_non_trapping): Adjust.
242         * tree-ssa-pre.c (eliminate_dom_walker): New class.
243         (eliminate_bb): Make method eliminate_dom_walker::before_dom_children.
244         (eliminate_leave_block): Make method.
245         eliminate_dom_walker::after_dom_children.
246         (eliminate): Adjust
247         * tree-ssa-strlen.c (strlen_dom_walker): New class.
248         (strlen_enter_block): Make method
249         strlen_dom_walker::before_dom_children.
250         (strlen_leave_block): Make
251         method strlen_dom_walker::after_dom_children.
252         (tree_ssa_strlen): Adjust.
253         * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
254         (tree_ssa_uncprop): Adjust.
255         (uncprop_leave_block): Make method
256         uncprop_dom_walker::after_dom_children.
257         (uncprop_leave_block): Make method
258         uncprop_dom_walker::before_dom_children.
260 2013-09-18  Bin Cheng  <bin.cheng@arm.com>
262         * config/arm/arm.c (thumb1_reorg): Search for flag setting insn
263         before branch in same basic block.  Check both src and dest of
264         the move insn.
266 2013-09-17  Nick Clifton  <nickc@redhat.com>
268         * config/rl78/rl78-real.md (bf): New pattern.
269         (bt): New pattern.
270         * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
271         (rl78_print_operand): Do not put a # before a %B.
272         * config/rl78/rl78.opt: Tweak doc strings.
274 2013-09-17  DJ Delorie  <dj@redhat.com>
276         * config/rl78/constraints.md (Wcv): Allow up to $r31.
277         * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
278         (rl78_option_override): Likewise, if -mallregs.
279         (is_virtual_register): Likewise.
280         * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
281         (REGNO_OK_FOR_BASE_P): Likewise.
282         * config/rl78/rl78.opt (-mallregs): New.
284 2013-09-17  Nick Clifton  <nickc@redhat.com>
286         * config/rl78/rl78.c (need_to_save): Change return type to bool.
287         For interrupt functions: save all call clobbered registers if the
288         interrupt handler is not a leaf function.
289         (rl78_expand_prologue): Always recompute the frame information.
290         For interrupt functions: only select bank 0 if one of the bank 0
291         registers is going to be psuhed.
293 2013-09-17  DJ Delorie  <dj@redhat.com>
295         * config/rl78/constraints.md: For each W* constraint, rename to C*
296         and create a W* constraint that checks for an optional ES: prefix
297         pattern also.
298         * config/rl78/rl78.md (UNS_ES_ADDR): New.
299         (es_addr): New.  Used to wrap far addresses.
300         * config/rl78/rl78-protos.h (rl78_es_addr): New.
301         (rl78_es_base): New.
302         * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
303         wrapped far addresses.
304         (rl78_print_operand_1): Unwrap far addresses before processing.
305         (rl78_lo16): Wrap far addresses in unspecs.
306         (rl78_es_addr): New.
307         (rl78_es_base): New.
308         (insn_ok_now): Check for not-yet-wrapped far addresses.
309         (transcode_memory_rtx): Properly re-wrap far addresses.
311 2013-09-17  Sebastian Huber  <sebastian.huber@embedded-brains.de>
313         * config/sparc/t-rtems: Add leon3 multilibs.
315 2013-09-17  Cong Hou  <congh@google.com>
317         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
318         when checking the dot production pattern. The type of rhs operand
319         of multiply is now checked correctly.
321 2013-09-17  Jeff Law  <law@redhat.com>
323         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
324         edge implied equivalences into successor phis.
325         * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
326         here from tree-ssa-threadedge.c.
327         (mark_threaded_blocks): When threading through a joiner, if both
328         successors of the joiner's clone reach the same block, verify the
329         PHI arguments are equal.  If not, cancel the jump threading request.
330         * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
331         tree-ssa-threadupdate.c
332         (thread_across_edge): Don't check PHI argument equality when
333         threading through joiner block here.
334         
335 2013-09-17  Andrew MacLeod <amacleod@redhat.com>
337         * tree-flow.h (ssa_undefined_value_p): Remove prototype.
338         * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
339         (warn_uninit, warn_uninitialized_vars, execute_early_warn_uninitialized,
340         make_pass_early_warn_uninitialized): Move to tree-ssa-uninit.c.
341         * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c
342         (has_undefined_value_p): New.  Pass dependant parts of 
343         ssa_undefined_value_p.
344         (uninit_undefined_value_p): Use has_undefined_value_p.
345         (warn_uninit, warn_uninitialized_vars, execute_early_warn_uninitialized,
346         make_pass_early_warn_uninitialized): Move from tree-ssa.c
347         * tree-ssa.h: Adjust prototypes
349 2013-09-17  Jan Hubicka  <jh@suse.cz>
351         PR middle-end/58332
352         * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
353         * ipa-inline.c (can_inline_edge_p): Do not downgrade
354         FUNCTION_NOT_OPTIMIZED.
355         * ipa-inline-analysis.c (compute_inline_parameters): Function
356         not optimized is not inlinable unless it is alwaysinline.
357         (inline_analyze_function): Force calls in not optimized
358         function not inlinable.
360 2013-09-17  Jan Hubicka  <jh@suse.cz>
362         PR middle-end/58329
363         * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
364         to return NULL.
365         * ipa.c (function_and_variable_visibility): Likewise.
366         * ipa-profile.c (ipa_profile): Likewise.
368 2013-09-17  Bernd Edlinger  <bernd.edlinger@hotmail.de>
370         PR ipa/58398
371         * cgraph.c (cgraph_function_body_availability): Check for ifunc
372         attribute, and don't inline the resolver in this case.
374 2013-09-17  Teresa Johnson  <tejohnson@google.com>
376         * coverage.c (get_coverage_counts): Add missing newline.
378 2013-09-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
380         PR tree-optimization/58088
381         * fold-const.c (mask_with_trailing_zeros): New function.
382         (fold_binary_loc): Make sure we don't recurse infinitely
383         when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
384         Use mask_with_trailing_zeros where appropriate.
386 2013-09-17  Yuri Rumyantsev  <ysrumyan@gmail.com>
388         * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
389         of 'prev' var to get better distance estimation.
391 2013-09-17  Eric Botcazou  <ebotcazou@adacore.com>
393         * tree-inline.h (struct copy_body_data): Add transform_parameter.
394         * tree-inline.c (is_parameter_of): New predicate.
395         (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
396         a parameter has been remapped.
397         (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
398         (optimize_inline_calls): Initialize transform_parameter.
399         (copy_gimple_seq_and_replace_locals): Likewise.
400         (tree_function_versioning): Likewise.
401         (maybe_inline_call_in_expr): Likewise.
403 2013-09-17  Nick Clifton  <nickc@redhat.com>
405         * config/msp430/msp430-protos.h: Add prototypes for new functions.
406         * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
407         interrupt handlers.
408         (is_attr_func): New function.
409         (msp430_is_interrupt_func): New function.
410         (is_naked_func): New function.
411         (is_reentrant_func): New function.
412         (is_critical_func): New function.
413         (msp430_start_function): Add annotations for function attributes.
414         (msp430_attr): New function.
415         (msp430_attribute_table): New.
416         (msp430_function_section): New function.
417         (TARGET_ASM_FUNCTION_SECTION): Define.
418         (msp430_builtin): New enum.
419         (msp430_init_builtins): New function.
420         (msp430_builtin_devl): New function.
421         (msp430_expand_builtin): New function.
422         (TARGET_INIT_BUILTINS): Define.
423         (TARGET_EXPAND_BUILTINS): Define.
424         (TARGET_BUILTIN_DECL): Define.
425         (msp430_expand_prologue): Add support for naked, interrupt,
426         critical and reentrant functions.
427         (msp430_expand_epilogue): Likewise.
428         (msp430_print_operand): Handle 'O' character.
429         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
430         NO_TRAMPOLINES.
431         * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
432         UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
433         (pushm): Use a 'n' rather than an 'i' constraint.
434         (msp_return): Add generation of the interrupt return instruction.
435         (disable_interrupts): New pattern.
436         (enable_interrupts): New pattern.
437         (push_intr_state): New pattern.
438         (pop_intr_state): New pattern.
439         (bic_SR): New pattern.
440         (bis_SR): New pattern.
441         * doc/extend.texi: Document MSP430 function attributes and builtin
442         functions.
444 2013-09-17  Richard Biener  <rguenther@suse.de>
446         PR tree-optimization/58432
447         * tree-loop-distribution.c (tree_loop_distribution): Also
448         scan PHIs for outside loop uses and seed a partition from them.
450 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
452         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
453         (restructure_reference): Call backtrace_base_for_ref.
455 2013-09-17  Alan Modra  <amodra@gmail.com>
457         PR target/57589
458         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11 patch.
460 2013-09-16  DJ Delorie  <dj@redhat.com>
462         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
463         vregs location for RL78/G10.
464         (rl78_expand_prologue): Avoid SEL on G10.
465         (rl78_expand_epilogue): Likewise.
466         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
467         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
468         __RL78_G10__ when appropriate.
469         (ASM_SPEC): Pass -mg10 along to the assembler.
470         * config/rl78/rl78.md (sel_rb): Disable for G10.
471         * config/rl78/rl78.opt: Add -mg10 option.
472         * config/rl78/t-rl78: Add -mg10 multilib.
474 2013-09-16  Xinliang David Li  <davidxl@google.com>
476         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
477         * omp-low.c (omp_max_vf): Ditto.
478         (expand_omp_simd): Ditto.
479         * tree-vectorizer.c (vectorize_loops): Ditto.
480         (gate_vect_slp): Ditto.
481         (gate_increase_alignment): Ditto.
482         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
483         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
484         (gate_tree_vectorize): Name change.
485         (tree_vectorize): Ditto.
486         (pass_vectorize::gate): Call new function.
487         (pass_vectorize::execute): Ditto.
488         opts.c: O3 default setting change.
489         (finish_options): Check new flag.
490         * doc/invoke.texi: Document new flags.
491         * common.opt: New flags.
493 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
495         * doc/tm.texi.in (Cond Exec Macros): Remove node.
496         (Condition Code): Don't reference it.
497         * doc/tm.texi: Regenerate.
499 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
501         PR middle-end/58418
502         * lra-constraints.c (undo_optional_reloads): Consider all optional
503         reload even if it did not get a hard reg.
505 2013-09-16  Teresa Johnson  <tejohnson@google.com>
507         * dumpfile.c (dump_loc): Remove newline emission.
508         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
509         emission to dump_printf_loc calls where missing.
510         (vect_mark_for_runtime_alias_test): Ditto.
511         (vect_analyze_data_ref_dependence): Ditto.
512         (vect_analyze_data_ref_dependences): Ditto.
513         (vect_slp_analyze_data_ref_dependence): Ditto.
514         (vect_slp_analyze_data_ref_dependences): Ditto.
515         (vect_compute_data_ref_alignment): Ditto.
516         (vect_update_misalignment_for_peel): Ditto.
517         (vect_verify_datarefs_alignment): Ditto.
518         (vector_alignment_reachable_p): Ditto.
519         (vect_get_data_access_cost): Ditto.
520         (vect_enhance_data_refs_alignment): Ditto.
521         (vect_find_same_alignment_drs): Ditto.
522         (vect_analyze_data_refs_alignment): Ditto.
523         (vect_analyze_group_access): Ditto.
524         (vect_analyze_data_ref_access): Ditto.
525         (vect_analyze_data_ref_accesses): Ditto.
526         (vect_prune_runtime_alias_test_list): Ditto.
527         (vect_analyze_data_refs): Ditto.
528         (vect_create_addr_base_for_vector_ref): Ditto.
529         (vect_create_data_ref_ptr): Ditto.
530         (vect_grouped_store_supported): Ditto.
531         (vect_grouped_load_supported): Ditto.
532         * value-prof.c (check_counter): Ditto.
533         (check_ic_target): Ditto.
534         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
535         (vect_recog_widen_mult_pattern): Ditto.
536         (vect_recog_widen_sum_pattern): Ditto.
537         (vect_recog_over_widening_pattern): Ditto.
538         (vect_recog_widen_shift_pattern): Ditto.
539         (vect_recog_rotate_pattern): Ditto.
540         (vect_recog_vector_vector_shift_pattern): Ditto.
541         (vect_recog_divmod_pattern): Ditto.
542         (vect_recog_mixed_size_cond_pattern): Ditto.
543         (vect_recog_bool_pattern): Ditto.
544         (vect_pattern_recog_1): Ditto.
545         (vect_pattern_recog): Ditto.
546         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
547         (vect_is_simple_iv_evolution): Ditto.
548         (vect_analyze_scalar_cycles_1): Ditto.
549         (vect_get_loop_niters): Ditto.
550         (vect_analyze_loop_1): Ditto.
551         (vect_analyze_loop_form): Ditto.
552         (vect_analyze_loop_operations): Ditto.
553         (vect_analyze_loop_2): Ditto.
554         (vect_analyze_loop): Ditto.
555         (report_vect_op): Ditto.
556         (vect_is_slp_reduction): Ditto.
557         (vect_is_simple_reduction_1): Ditto.
558         (vect_get_known_peeling_cost): Ditto.
559         (vect_estimate_min_profitable_iters): Ditto.
560         (vect_model_reduction_cost): Ditto.
561         (vect_model_induction_cost): Ditto.
562         (get_initial_def_for_induction): Ditto.
563         (vect_create_epilog_for_reduction): Ditto.
564         (vectorizable_reduction): Ditto.
565         (vectorizable_induction): Ditto.
566         (vectorizable_live_operation): Ditto.
567         (vect_loop_kill_debug_uses): Ditto.
568         (vect_transform_loop): Ditto.
569         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
570         (vect_stmt_relevant_p): Ditto.
571         (process_use): Ditto.
572         (vect_mark_stmts_to_be_vectorized): Ditto.
573         (vect_model_simple_cost): Ditto.
574         (vect_model_promotion_demotion_cost): Ditto.
575         (vect_model_store_cost): Ditto.
576         (vect_get_store_cost): Ditto.
577         (vect_model_load_cost): Ditto.
578         (vect_get_load_cost): Ditto.
579         (vect_init_vector_1): Ditto.
580         (vect_get_vec_def_for_operand): Ditto.
581         (vect_finish_stmt_generation): Ditto.
582         (vectorizable_call): Ditto.
583         (vectorizable_conversion): Ditto.
584         (vectorizable_assignment): Ditto.
585         (vectorizable_shift): Ditto.
586         (vectorizable_operation): Ditto.
587         (vectorizable_store): Ditto.
588         (vectorizable_load): Ditto.
589         (vectorizable_condition): Ditto.
590         (vect_analyze_stmt): Ditto.
591         (vect_transform_stmt): Ditto.
592         (vect_is_simple_use): Ditto.
593         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
594         (vect_can_advance_ivs_p): Ditto.
595         (vect_update_ivs_after_vectorizer): Ditto.
596         (vect_do_peeling_for_loop_bound): Ditto.
597         (vect_gen_niters_for_prolog_loop): Ditto.
598         (vect_update_inits_of_drs): Ditto.
599         (vect_create_cond_for_alias_checks): Ditto.
600         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
601         (vect_build_slp_tree_1): Ditto.
602         (vect_supported_load_permutation_p): Ditto.
603         (vect_analyze_slp_instance): Ditto.
604         (vect_analyze_slp): Ditto.
605         (vect_make_slp_decision): Ditto.
606         (vect_detect_hybrid_slp): Ditto.
607         (vect_bb_vectorization_profitable_p): Ditto.
608         (vect_slp_analyze_bb_1): Ditto.
609         (vect_update_slp_costs_according_to_vf): Ditto.
610         (vect_get_mask_element): Ditto.
611         (vect_transform_slp_perm_load): Ditto.
612         (vect_schedule_slp_instance): Ditto.
613         (vect_schedule_slp): Ditto.
614         (vect_slp_transform_bb): Ditto.
615         * profile.c (read_profile_edge_counts): Ditto.
616         (compute_branch_probabilities): Ditto.
617         * coverage.c (get_coverage_counts): Ditto.
619 2013-09-16  Diego Novillo  <dnovillo@google.com>
621         * tree-core.h: Add missing comment lines from refactoring
622         of tree.h.
624 2013-09-16  Jan Hubicka  <jh@suse.cz>
626         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
627         abstract functions; for static functions check the presence
628         of body.
630 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
632         * config/aarch64/aarch64-simd-builtins.def (fma): New.
633         * config/aarch64/aarch64-simd.md
634         (aarch64_mla_elt<mode>): New.
635         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
636         (aarch64_mls_elt<mode>): Likewise.
637         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
638         (aarch64_fma4_elt<mode>): Likewise.
639         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
640         (aarch64_fma4_elt_to_128v2df): Likewise.
641         (aarch64_fma4_elt_to_64df): Likewise.
642         (fnma<mode>4): Likewise.
643         (aarch64_fnma4_elt<mode>): Likewise.
644         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
645         (aarch64_fnma4_elt_to_128v2df): Likewise.
646         (aarch64_fnma4_elt_to_64df): Likewise.
647         * config/aarch64/iterators.md (VDQSF): New.
648         * config/aarch64/arm_neon.h
649         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
650         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
652 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
654         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
655         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
656         (aarch64_mul3_elt_to_128df): Likewise.
657         (aarch64_mul3_elt_to_64v2df): Likewise.
658         * config/aarch64/iterators.md (VEL): Also handle DFmode.
659         (VMUL): New.
660         (VMUL_CHANGE_NLANES) Likewise.
661         (h_con): Likewise.
662         (f): Likewise.
663         * config/aarch64/arm_neon.h
664         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
666 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
668         * config/aarch64/arm_neon.h
669         (vcvtx_high_f32_f64): Fix parameters.
671 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
672             Uros Bizjak  <ubizjak@gmail.com>
674         * config/alpha.c: Include tree-ssanames.h.
676 2013-09-16  Richard Biener  <rguenther@suse.de>
678         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
679         (dot_rdg_1): Handle control_dd.
680         (create_edge_for_control_dependence): New function.
681         (create_rdg_edges): Add control dependences if asked for.
682         (build_rdg): Likewise.
683         (generate_loops_for_partition): If there are not necessary
684         control stmts remove all their dependencies.
685         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
686         (distribute_loop): Pass on control dependences.
687         (tree_loop_distribution): Compute control dependences and remove
688         restriction on number of loop nodes.
690 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
692         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
693         for internal calls.
695 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
697         * cse.c (try_const_anchors): Punt on CC modes.
699 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
701         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
703 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
705         PR target/58382
706         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
707         calls to word_mode.
709 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
711         PR target/48094
712         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
713         seen.
714         (darwin_objc1_section): Likewise.
715         (darwin_file_end): Emit Image Info section when required.
717 2013-09-14  Jan Hubicka  <jh@suse.cz>
719         * tree-into-ssa.c (gate_into_ssa): New.
720         (pass_data_build_ssa): Use it.
721         * cgraph.h (expand_thunk): Update prototype.
722         * cgraphunit.c (analyze_function): Expand thunks early.
723         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
724         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
725         set lowered flag; do not add new function.
726         (assemble_thunks_and_aliases): Update.
727         * tree-ssa.c (gate_init_datastructures): New gate.
728         (pass_data_init_datastructures): Use it.
730 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
732         PR target/58269
733         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
734         xmm register set described in the psABI.
736 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
738         * dwarf2out.c (should_emit_struct_debug): Add check
739         for type_decl variable is not NULL.
741 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
743         * config.gcc: Use new winnt-c.c target hooks
744         * config/t-winnt: New file
745         * config/winnt-c.c: New file
746         * doc/tm.texi.in: Document new hook
747         * doc/tm.texi: Regenerated
749 2013-09-13  Jan Hubicka  <jh@suse.cz>
751         PR middle-end/58094
752         * ipa-inline.c (check_callers): New function.
753         (check_caller_edge): Remove.
754         (want_inline_function_to_all_callers_p): Also permit alises that are
755         called dirrectly.
756         (inline_to_all_callers): Terminate the walk when devirtualization
757         introduce new calls.
759 2013-09-13  Jan Hubicka  <jh@suse.cz>
761         * ipa-inline-analysis.c (struct growth_data): Add node.
762         (do_estimate_growth_1): Fix detection of recursion.
764 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
766         PR tree-optimization/58392
767         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
768         to avoid shadowing of outer loop variable.  If
769         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
770         replace_by_duplicate_decl simduid of loops that have it set and
771         set dest_cfun->has_simduid_loops and/or
772         dest_cfun->has_force_vect_loops.
773         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
774         instead of maybe_lookup_decl.
775         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
776         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
777         call.  Copy over force_vect and copy and remap simduid.  Set
778         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
779         (copy_cfg_body): Remove blocks_to_copy argument.  Use
780         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
781         caller.  Don't set cfun->has_simduid_loops and/or
782         cfun->has_force_vect_loops here.
783         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
784         caller.
785         (expand_call_inline, tree_function_versioning): Adjust copy_body
786         callers.
788 2013-09-13  Martin Jambor  <mjambor@suse.cz>
790         PR bootstrap/58388
791         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
792         the assert if the edge was a speculative one.
794 2013-09-13  Richard Biener  <rguenther@suse.de>
796         * tree-data-ref.h (known_dependences_p): Move here ...
797         * tree-loop-distribution.c (known_dependences_p): ... from here.
798         (dump_rdg_component, debug_rdg_component): Remove.
799         (dump_rdg): Adjust.
800         (generate_loops_for_partition): Use gimple_uid instead of
801         relying on matching stmt visit order.
802         (rdg_build_partitions): Take starting stmt vector.
803         (ldist_gen): Merge into ...
804         (distribute_loop): ... this function.  Do not compute starting
805         vertices vector.
806         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
808 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
810         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
811         Set type attribute correctly. Set predicable_short_it attribute.
812         (cmpsi_shiftsi): Remove %? from output template.
814 2013-09-13  Richard Biener  <rguenther@suse.de>
816         * tree-loop-distribution.c (struct rdg_component,
817         rdg_defs_used_in_other_loops_p, free_rdg_components,
818         rdg_build_components): Remove.
819         (stmts_from_loop): Do not record virtual PHIs.
820         (generate_loops_for_partition): Skip virtual PHIs.
821         (build_rdg_partition_for_component): Rename to ...
822         (build_rdg_partition_for_vertex): ... this and adjust.
823         (rdg_build_partitions): Take a vector of starting vertices
824         instead of components.  Remove unnecessary leftover handling.
825         (ldist_gen): Do not build components or record other stores.
826         (distribute_loop): Do not distribute loops containing stmts
827         with side-effects.
829 2013-09-13  Christian Bruel  <christian.bruel@st.com>
831         PR target/58314
832         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
834 2013-09-13  Kai Tietz  <ktietz@redhat.com>
836         * config.gcc: Separate cases for mingw and cygwin targets,
837         and add 64-bit cygwin target case.
839         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
840         dll-export inline-functions.
841         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
844 2013-09-13  Jeff Law  <law@redhat.com>
846         PR middle-end/58387
847         Revert:
848         2013-09-06  Jeff Law  <law@redhat.com>
850         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
851         edge implied equivalences into successor phis.
853 2013-09-12  DJ Delorie  <dj@redhat.com>
855         * config/rl78/rl78-virt.md: Change from | to \; for asm line
856         separators.
858 2013-09-12  Brooks Moses  <bmoses@google.com>
860         PR driver/42955
861         * Makefile.in: Do not install driver binaries in $(target)/bin.
863 2013-09-12  DJ Delorie  <dj@redhat.com>
865         * config/rl78/rl78.opt (mrelax): New.
866         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
867         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
869         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
870         between SP and FP.
871         (rl78_expand_epilogue): Likewise.
873 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
875         PR middle-end/58335
876         * lra-eliminations.c (remove_reg_equal_offset_note): New.
877         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
878         pointer elimination with using remove_reg_equal_offset_note.
880 2013-09-12  DJ Delorie  <dj@redhat.com>
882         * config/msp430/: New port.
883         * config.gcc (msp430): Added.
884         * doc/invoke.texi: Document MSP430 options.
885         * doc/install.texi: Document msp430-elf
886         * doc/md.texi: Document msp430-elf
887         * doc/contrib.texi: Document msp430-elf
889         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
890         PSImode.
892 2013-09-12  Martin Jambor  <mjambor@suse.cz>
894         PR ipa/58389
895         * ipa-prop.c (remove_described_reference): Give up if the edge in the
896         reference descriptor is NULL.
897         (ipa_edge_removal_hook): If owning a reference descriptor, set its
898         edge to NULL.
900 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
902         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
903         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
904         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
905         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
906         make_temp_ssa_name): move to tree-ssanames.h
907         * tree-ssa-alias.h: Move prototype.
908         * tree-ssa.h: Include tree-ssanames.h.
909         * tree-ssanames.c (FREE_SSANAMES): Move to here.
910         * tree-ssanames.h: New.  Move items from tree-flow*.h
911         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
913 2013-09-12  Richard Biener  <rguenther@suse.de>
915         PR tree-optimization/58404
916         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
917         propagate non-invariant addresses into dereferences wrapped
918         in component references.
920 2013-09-12  Richard Biener  <rguenther@suse.de>
922         PR tree-optimization/58402
923         * passes.def: Move pass_late_warn_uninitialized later.
925 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
927         * tree-ssa.h: New.  Move content from tree-flow.h and
928         tree-flow-inline.h.
929         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
930         Move prototypes belonging to tree-ssa.c.
931         * tree-flow-inline.h (redirect_edge_var_map_def,
932         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
933         tree-ssa.h.
934         * gimple.h: Adjust prototypes.
935         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
936         to...
937         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
938         * tree.h: Move prototype to tree-ssa.h.
939         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
940         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
941         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
942         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
943         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
944         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
945         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
946         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
947         graphite-blocking.c, graphite-clast-to-gimple.c,
948         graphite-dependences.c, graphite-interchange.c,
949         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
950         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
951         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
952         ipa-reference.c, ipa-split.c, ipa-utils.c,
953         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
954         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
955         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
956         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
957         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
958         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
959         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
960         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
961         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
962         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
963         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
964         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
965         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
966         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
967         value-prof.c, var-tracking.c,
968         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
970 2013-09-12  Richard Biener  <rguenther@suse.de>
972         PR tree-optimization/58396
973         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
974         (build_rdg): Take a loop-nest parameter, fix memleaks.
975         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
977 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
979         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
980         for SLM.
982 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
984         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
985         The FMA instruction names should have a 'v' prefix.
987 2013-09-12  Richard Biener  <rguenther@suse.de>
989         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
990         (dot_rdg): Use popen instead of system in optional code.
991         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
992         (already_processed_vertex_p): Adjust.
993         (has_anti_or_output_dependence, predecessor_has_mem_write,
994         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
995         rdg_flag_uses): Remove.
996         (rdg_flag_vertex): Simplify.
997         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
998         remove recursion.
999         (build_rdg_partition_for_component): Process the first vertex
1000         of a component only.
1001         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
1003 2013-09-12  Alan Modra  <amodra@gmail.com>
1005         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
1007 2013-09-11  DJ Delorie  <dj@redhat.com>
1008             Nick Clifton  <nickc@redhat.com>
1010         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
1011         (rl78_stack_based_mem): New.
1012         * config/rl78/constraints.md (Iv08): New.
1013         (Iv16): New.
1014         (Iv24): New.
1015         (Is09): New.
1016         (Is17): New.
1017         (Is25): New.
1018         (ISsi): New.
1019         (IShi): New.
1020         (ISqi): New.
1021         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
1022         (movhi): Likewise.
1023         (movsi): Change from expand to insn-and-split.
1024         (ashrsi3): Clobber AX.
1025         (lshrsi3): New.
1026         (ashlsi3): New.
1027         (cbranchsi4): New.
1028         * config/rl78/rl78.md (CC_REG): Fix.
1029         (addsi3): Allow memory and immediate operands.
1030         (addsi3_internal): Split into...
1031         (addsi3_internal_virt): ...new, and ...
1032         (addsi3_internal_real): ...new.
1033         (subsi): New.
1034         (subsi3_internal_virt): New.
1035         (subsi3_internal_real): New.
1036         (mulsi3): Add memory operand.
1037         (mulsi3_rl78): Likewise.
1038         (mulsi3_g13): Likewise.
1039         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
1040         (cbranchqi4_real): Add more constraint options.
1041         (cbranchhi4_real): Expand pattern.
1042         (cbranchhi4_real_signed): New.
1043         (cbranchhi4_real_inverted): New.
1044         (cbranchsi4_real_lt): New.
1045         (cbranchsi4_real_ge): New.
1046         (cbranchsi4_real_signed): New.
1047         (cbranchsi4_real): New.
1048         (peephole2): New.
1049         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
1050         constant shifts.
1051         (lshrsi3_virt): Likewise.
1052         (ashlsi3_virt): Likewise.
1053         (cbranchqi4_virt_signed): New.
1054         (cbranchhi4_virt_signed): New.
1055         (cbranchsi4_virt): New.
1056         * config/rl78/rl78.c: Whitespace fixes throughout.
1057         (move_elim_pass): New.
1058         (pass_data_rl78_move_elim): New.
1059         (pass_rl78_move_elim): New.
1060         (make_pass_rl78_move_elim): New.
1061         (rl78_devirt_info): Run devirt earlier.
1062         (rl78_move_elim_info): New.
1063         (rl78_asm_file_start): Register it.
1064         (rl78_split_movsi): New.
1065         (rl78_as_legitimate_address): Allow virtual base registers when
1066         appropriate.
1067         (rl78_addr_space_convert): Remove spurious debug stuff.
1068         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
1069         (rl78_print_operand): More cases for not printing '#'.
1070         (rl78_expand_compare): Remove most of the logic.
1071         (content_memory): New.
1072         (clear_content_memory): New.
1073         (get_content_index): New.
1074         (get_content_name): New.
1075         (display_content_memory): New.
1076         (update_content): New.
1077         (record_content): New.
1078         (already_contains): New.
1079         (insn_ok_now): Re-recog insns with virtual registers.
1080         (add_postponed_content_update): New.
1081         (process_postponed_content_update): New.
1082         (gen_and_emit_move): New.
1083         (transcode_memory_rtx): Record new location content.
1084         Use gen_and_emit_move.
1085         (force_into_acc): New.
1086         (move_to_acc): Use gen_and_emit_move.
1087         (move_from_acc): Likewise.
1088         (move_acc_to_reg): Likewise.
1089         (move_to_x): Likewise.
1090         (move_to_hl): Likewise.
1091         (move_to_de): Likewise.
1092         (rl78_alloc_physical_registers_op1): Record location content.
1093         (has_constraint): New.
1094         (rl78_alloc_physical_registers_op2): Record location content.
1095         Optimize use of HL.
1096         (rl78_alloc_physical_registers_ro1): Likewise.
1097         (rl78_alloc_physical_registers_cmp): Likewise.
1098         (rl78_alloc_physical_registers_umul): Likewise.
1099         (rl78_alloc_address_registers_macax): New.
1100         (rl78_alloc_physical_registers): Initialize and set location
1101         content memory as needed.
1102         (rl78_reorg): Make sure split2 is called.
1103         (rl78_rtx_costs): New.
1105 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
1107         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
1108         for (not (neg ...)) and (neg (not ...)) cases.
1110 2013-09-11  Richard Biener  <rguenther@suse.de>
1112         PR middle-end/58377
1113         * passes.def: Split critical edges before late uninit warning passes.
1114         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
1116 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
1118         PR tree-optimization/58385
1119         * fold-const.c (build_range_check): If both low and high are NULL,
1120         use omit_one_operand_loc to preserve exp side-effects.
1122 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1124         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
1126 2013-09-11  Richard Biener  <rguenther@suse.de>
1128         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
1129         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
1130         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
1131         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
1132         stmts_from_loop, known_dependences_p, build_empty_rdg,
1133         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
1134         * tree-loop-distribution.c: ... here.
1135         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
1136         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
1137         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
1138         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
1139         * tree-loop-distribution.c: ... here.
1140         * tree-loop-distribution.c: Include gimple-pretty-print.h.
1141         (struct partition_s): Add loops member.
1142         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
1143         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
1144         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
1146 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1147             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1148             Sergey Lega  <sergey.s.lega@intel.com>
1149             Anna Tikhonova  <anna.tikhonova@intel.com>
1150             Ilya Tocar  <ilya.tocar@intel.com>
1151             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1152             Ilya Verbin  <ilya.verbin@intel.com>
1153             Kirill Yukhin  <kirill.yukhin@intel.com>
1154             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1156         * config/i386/constraints.md (k): New.
1157         (Yk): Ditto.
1158         * config/i386/i386.c (const regclass_map): Add new mask registers.
1159         (dbx_register_map): Ditto.
1160         (dbx64_register_map): Ditto.
1161         (svr4_dbx_register_map): Ditto.
1162         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
1163         disabled.
1164         (ix86_preferred_reload_class): Disable constants for mask registers.
1165         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
1166         (ix86_hard_regno_mode_ok): Support new mask registers.
1167         (x86_order_regs_for_local_alloc): Ditto.
1168         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
1169         (FIXED_REGISTERS): Add new mask registers.
1170         (CALL_USED_REGISTERS): Ditto.
1171         (REG_ALLOC_ORDER): Ditto.
1172         (VALID_MASK_REG_MODE): New.
1173         (FIRST_MASK_REG): Ditto.
1174         (LAST_MASK_REG): Ditto.
1175         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
1176         (MAYBE_MASK_CLASS_P): New.
1177         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
1178         (REG_CLASS_CONTENTS): Ditto.
1179         (MASK_REGNO_P): New.
1180         (ANY_MASK_REG_P): Ditto.
1181         (HI_REGISTER_NAMES): Add new mask registers.
1182         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
1183         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
1184         mask registers.
1185         (attribute "type"): Add mskmov, msklog.
1186         (attribute "length_immediate"): Support them.
1187         (attribute "memory"): Ditto.
1188         (attribute "prefix_0f"): Ditto.
1189         (*movhi_internal): Support new mask registers.
1190         (*movqi_internal): Ditto.
1191         (define_split): Split out clobber pattern is a logic
1192         insn on mask registers.
1193         (*k<logic><mode>): New.
1194         (*andhi_1): Extend to support mask regs.
1195         (*andqi_1): Extend to support mask regs.
1196         (kandn<mode>): New.
1197         (define_split): Split and-not to and and not if operands
1198         are not mask regs.
1199         (*<code><mode>_1): Separate HI mode to new pattern...
1200         (*<code>hi_1): This.
1201         (*<code>qi_1): Extend to support mask regs.
1202         (kxnor<mode>): New.
1203         (kortestzhi): Ditto.
1204         (kortestchi): Ditto.
1205         (kunpckhi): Ditto.
1206         (*one_cmpl<mode>2_1): Remove HImode and handle it...
1207         (*one_cmplhi2_1): ...Here, now with mask registers support.
1208         (*one_cmplqi2_1): Support new mask registers.
1209         (HI/QImode arithmetics splitter): Don't split if mask registers
1210         are used.
1211         (HI/QImode not splitter): Ditto.
1212         * config/i386/predicated.md (mask_reg_operand): New.
1213         (general_reg_operand): Ditto.
1215 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1217         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
1218         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
1220 2013-09-10  Jeff Law  <law@redhat.com>
1222         PR tree-optimization/58380
1223         * tree-ssa-threadupdate.c (thread_block): Recognize another case
1224         of threading through a buried loop header.
1226         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
1227         return value for single successor case.
1229 2013-09-10  Jan Hubicka  <jh@suse.cz>
1231         * ipa-devirt.c (ipa_devirt): Enable with LTO.
1233 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
1235         PR target/58361
1236         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
1237         support conditional execution.
1238         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
1240 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
1242         * lra.c (lra): Clear lra_optional_reload_pseudos before every
1243         constraint pass.
1244         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
1245         Check destination too to check move insn.
1246         (undo_optional_reloads): Add check that the original peudo did not
1247         changed its allocation and the optional reload was inherited on last
1248         inheritance pass.  Break loop after deciding to keep optional reload.
1249         (lra_undo_inheritance): Add check that inherited pseudo still in
1250         memory.
1252 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
1254         * config/aarch64/aarch64.md (generic_sched): New.
1255         * config/aarch64/aarch64-generic.md (load): Make conditional
1256         on generic_sched attribute.
1257         (nonload): Likewise.
1259 2013-09-10  Jan Hubicka  <jh@suse.cz>
1261         * lto-cgraph.c: Include ipa-utils.h.
1262         (compute_ltrans_boundary): Also add possible targets into the boundary.
1264 2013-09-10  Jan Hubicka  <jh@suse.cz>
1266         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
1267         VAR_DECL of vtable rather than full expression.
1269 2013-09-10  Jan Hubicka  <jh@suse.cz>
1270             Paolo Carlini  <paolo.carlini@oracle.com>
1272         * cgraphunit.c (analyze_functions): Save input_location, set it
1273         to UNKNOWN_LOCATION and restore it at the end.
1275 2013-09-10  Martin Jambor  <mjambor@suse.cz>
1277         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
1278         represented by a thunk.
1280 2013-09-10  Jeff Law  <law@redhat.com>
1282         PR tree-optimization/58343
1283         * tree-ssa-threadupdate.c (thread_block): Identify and disable
1284         jump threading requests through loop headers buried in the middle
1285         of a jump threading path.
1287         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
1288         in return value/type.
1290 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
1292         PR rtl-optimization/58365
1293         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
1294         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
1295         it differs.
1297 2013-09-10  Richard Biener  <rguenther@suse.de>
1299         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
1300         * tree-data-ref.c (create_rdg_vertices): Collect all data
1301         references, signal failure to the caller, use data-ref API.
1302         (build_rdg): Compute data references only once.  Maintain lifetime
1303         of data references and data dependences from within RDG.
1304         (free_rdg): Free dependence relations.
1305         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
1306         inventing extra dependences.
1307         (distribute_loop): Update for RDG API changes.
1309 2013-09-10  Kai Tietz  <ktietz@redhat.com>
1311         * doc/invoke.texi (fms-extensions): Document changed
1312         behavior for ms-abi targets.
1313         * config/i386/i386.c (ix86_option_override_internal):
1314         Set default value of option -fms-extension for ms-abi targets.
1316 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
1318         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
1320 2013-09-10  Alan Modra  <amodra@gmail.com>
1322         PR target/58330
1323         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
1325 2013-09-10  Alan Modra  <amodra@gmail.com>
1327         * config/rs6000/predicates.md (add_cint_operand): New.
1328         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
1329         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
1330         using add_cint_operand.
1331         (largetoc_high_plus_aix): Likewise.
1333 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
1335         PR tree-optimization/58364
1336         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
1337         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
1338         the current range can't be an unconditional true or false.
1340 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
1342         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
1344 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
1346         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
1348 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
1350         PR c++/43452
1351         * doc/invoke.texi (-Wdelete-incomplete): Document it.
1353 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
1355         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
1356         NO_REGS for immediate that can't be moved directly into FP_REGS.
1358 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1360         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
1361         comparison with negated operand.
1362         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
1363         RTL form.
1365 2013-09-09  Richard Biener  <rguenther@suse.de>
1367         PR middle-end/58326
1368         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
1369         of a subloop record all its block as affecting loop-closed SSA form.
1371 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
1373         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
1374         of an rtx/bitpos pair.
1375         (store_fixed_bit_field): Update accordingly.
1377 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
1379         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
1380         GEN_INT.
1381         * builtins.c (expand_errno_check): Likewise.
1382         * dwarf2cfi.c (init_return_column_size): Likewise.
1383         * except.c (sjlj_mark_call_sites): Likewise.
1384         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
1385         * lra-constraints.c (emit_inc): Likewise.
1386         * ree.c (combine_set_extension): Likewise.
1387         * regmove.c (fixup_match_2): Likewise.
1388         * reload1.c (inc_for_reload): Likewise.
1390 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
1392         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
1393         (force_to_mode, simplify_shift_const_1, simplify_comparison):
1394         Use gen_int_mode with the mode of the associated simplify_* call.
1395         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
1396         * expmed.c (expand_shift_1): Likewise.
1397         * function.c (instantiate_virtual_regs_in_insn): Likewise.
1398         * loop-iv.c (iv_number_of_iterations): Likewise.
1399         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
1400         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
1402 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
1404         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
1405         of the associated expand_* call.
1406         (asan_emit_stack_protection): Likewise.
1407         * builtins.c (round_trampoline_addr): Likewise.
1408         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
1409         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
1410         (emit_store_flag): Likewise.
1411         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
1412         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
1413         Likewise.
1414         * function.c (instantiate_virtual_regs_in_insn): Likewise.
1415         * ifcvt.c (noce_try_store_flag_constants): Likewise.
1416         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
1417         * modulo-sched.c (generate_prolog_epilog): Likewise.
1418         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
1419         (expand_ctz, expand_ffs, expand_unop): Likewise.
1421 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
1423         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
1424         of the associated gen_rtx_* call.
1425         * caller-save.c (init_caller_save): Likewise.
1426         * combine.c (find_split_point, make_extraction): Likewise.
1427         (make_compound_operation): Likewise.
1428         * dwarf2out.c (mem_loc_descriptor): Likewise.
1429         * explow.c (plus_constant, probe_stack_range): Likewise.
1430         * expmed.c (expand_mult_const): Likewise.
1431         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
1432         * reload1.c (init_reload): Likewise.
1433         * valtrack.c (cleanup_auto_inc_dec): Likewise.
1434         * var-tracking.c (adjust_mems): Likewise.
1435         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
1436         rather than gen_rtx_fmt_ee.
1438 2013-09-09  Jan Hubicka  <jh@suse.cz>
1440         PR middle-end/58294
1441         * value-prof.c (gimple_ic): Copy also abnormal edges.
1443 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
1445         * asan.c (asan_shadow_cst): Use gen_int_mode.
1447 2013-09-08  Jan Hubicka  <jh@suse.cz>
1449         * ipa-profile.c: Add toplevel comment.
1450         (ipa_propagate_frequency_1): Be more conservative when profile is read.
1451         (contains_hot_call_p): New function.
1452         (ipa_propagate_frequency): Set frequencies based on counts when
1453         profile is read.
1454         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
1455         profile; do not tamper with profile after inlining if it is read.
1457 2013-09-08  Jan Hubicka  <jh@suse.cz>
1459         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
1460         speculative edges.
1462 2013-09-08  Jan Hubicka  <jh@suse.cz>
1464         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
1465         summary generation.
1467 2013-09-08  Jeff Law  <law@redhat.com>
1469         PR bootstrap/58340
1470         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
1471         of 'found'.
1473 2013-09-08  Andi Kleen  <ak@linux.intel.com>
1475         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
1477 2013-09-08  Jan Hubicka  <jh@suse.cz>
1479         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
1481 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
1483         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
1484         for non-debug insns.
1485         * lra.c (new_insn_reg): Take the containing insn as a parameter.
1486         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
1487         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
1488         accordingly.
1490 2013-09-08  Jan Hubicka  <jh@suse.cz>
1492         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
1493         targets and devirtualize to BUILT_IN_UNREACHABLE.
1494         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
1495         * ipa.c (walk_polymorphic_call_targets): New function.
1496         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
1497         functions; use the new timevar.
1498         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
1499         was removed from the program.
1500         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
1501         not consider it in the walk when its vtable is dead.
1502         (possible_polymorphic_call_targets_1): Pass anonymous flag to
1503         record_binfo.
1504         (devirt_variable_node_removal_hook): New function.
1505         (possible_polymorphic_call_targets): Also register
1506         devirt_variable_node_removal_hook.
1507         (ipa_devirt): Do not do non-speculative devirtualization.
1508         (gate_ipa_devirt): One execute if devirtualizing speculatively.
1510 2013-09-08  Jan Hubicka  <jh@suse.cz>
1512         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
1513         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
1514         varpool_remove_variable_insertion_hook): Declare.
1515         * varpool.c (varpool_node_hook_list): New structure.
1516         (first_varpool_node_removal_hook,
1517         first_varpool_variable_insertion_hook): New variables.
1518         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
1519         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
1520         varpool_remove_variable_insertion_hook,
1521         varpool_call_variable_insertion_hooks): New functions.
1522         (varpool_remove_node): Use it.
1524 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
1526         PR c++/54941
1527         * diagnostic.c (diagnostic_build_prefix): When s.file is
1528         "<built-in>" don't output line and column numbers.
1530 2013-09-06  Jan Hubicka  <jh@suse.cz>
1532         * cgraphunit.c (expand_thunk): Get body before touching arguments.
1533         * lto-streamer-out.c: Stream thunks, too.
1534         * lto-streamer-in.c (input_function): Pop cfun here
1535         (lto_read_body): Instead of here.
1537 2013-09-06  Caroline Tice  <cmtice@google.com>
1539         * doc/install.texi: Add documentation for the --enable-vtable-verify
1540         and the --disable-libvtv configure options.
1542 2013-09-06  Jeff Law  <law@redhat.com>
1544         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
1545         edge implied equivalences into successor phis.
1547 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
1549         * resource.c (mark_referenced_resources): Handle COND_EXEC.
1551 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
1553         * resource.c (mark_target_live_regs): Compute resources taking
1554         into account if a call is predicated or not.
1556 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
1558         * toplev.c (output_stack_usage): Be prepared for suffixes created by
1559         the compiler in the function names.
1561 2013-09-06  Jan Hubicka  <jh@suse.cz>
1563         PR middle-end/58094
1564         * ipa-inline.c (has_caller_p): New function.
1565         (want_inline_function_to_all_callers_p): Use it.
1566         (sum_callers, inline_to_all_callers): Break out from ...
1567         (ipa_inline): ... here.
1569 2013-09-06  Jan Hubicka  <jh@suse.cz>
1571         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
1572         only when AVX is enabled.
1574 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1576         * config/aarch64/aarch64.md
1577         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
1578         is fpsimd_<load/store>2.
1579         (load_pair<mode>): Likewise.
1580         (store_pair<mode>): Likewise.
1582 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1584         * config/arm/types.md (type): Add "mrs" type.
1585         * config/aarch64/aarch64.md
1586         (aarch64_load_tp_hard): Make type "mrs".
1587         * config/arm/arm.md
1588         (load_tp_hard): Make type "mrs".
1589         * config/arm/cortex-a15.md: Update with new attributes.
1590         * config/arm/cortex-a5.md: Update with new attributes.
1591         * config/arm/cortex-a53.md: Update with new attributes.
1592         * config/arm/cortex-a7.md: Update with new attributes.
1593         * config/arm/cortex-a8.md: Update with new attributes.
1594         * config/arm/cortex-a9.md: Update with new attributes.
1595         * config/arm/cortex-m4.md: Update with new attributes.
1596         * config/arm/cortex-r4.md: Update with new attributes.
1597         * config/arm/fa526.md: Update with new attributes.
1598         * config/arm/fa606te.md: Update with new attributes.
1599         * config/arm/fa626te.md: Update with new attributes.
1600         * config/arm/fa726te.md: Update with new attributes.
1602 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1604         * config/aarch64/aarch64.md
1605         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
1606         (*movtf_aarch64): Likewise.
1607         * config/arm/arm.md
1608         (thumb1_movdi_insn): Use "multiple" for type where more than one
1609         instruction is used for a move.
1610         (*arm32_movhf): Likewise.
1611         (*thumb_movdf_insn): Likewise.
1613 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1615         * config/arm/types.md (type): Rename fcpys to fmov.
1616         * config/arm/vfp.md
1617         (*arm_movsi_vfp): Rename type fcpys as fmov.
1618         (*thumb2_movsi_vfp): Likewise
1619         (*movhf_vfp_neon): Likewise
1620         (*movhf_vfp): Likewise
1621         (*movsf_vfp): Likewise
1622         (*thumb2_movsf_vfp): Likewise
1623         (*movsfcc_vfp): Likewise
1624         (*thumb2_movsfcc_vfp): Likewise
1625         * config/aarch64/aarch64-simd.md
1626         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
1627         * config/aarch64/aarch64.md
1628         (*movsi_aarch64): Replace type mov_reg with fmovs.
1629         (*movdi_aarch64): Likewise
1630         (*movsf_aarch64): Likewise
1631         (*movdf_aarch64): Likewise
1632         * config/arm/arm.c
1633         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
1634         * config/arm/iwmmxt.md
1635         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
1636         * config/arm/arm1020e.md: Update with new attributes.
1637         * config/arm/cortex-a15-neon.md: Update with new attributes.
1638         * config/arm/cortex-a5.md: Update with new attributes.
1639         * config/arm/cortex-a53.md: Update with new attributes.
1640         * config/arm/cortex-a7.md: Update with new attributes.
1641         * config/arm/cortex-a8-neon.md: Update with new attributes.
1642         * config/arm/cortex-a9.md: Update with new attributes.
1643         * config/arm/cortex-m4-fpu.md: Update with new attributes.
1644         * config/arm/cortex-r4f.md: Update with new attributes.
1645         * config/arm/marvell-pj4.md: Update with new attributes.
1646         * config/arm/vfp11.md: Update with new attributes.
1648 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1650         * config/aarch64/aarch64.md
1651         (*madd<mode>): Fix type attribute.
1652         (*maddsi_uxtw): Likewise.
1653         (*msub<mode>): Likewise.
1654         (*msubsi_uxtw): Likewise.
1655         (<su_optab>maddsidi4): Likewise.
1656         (<su_optab>msubsidi4): Likewise.
1658 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1660         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
1661         * config/arm/arm.md (core_cycles): Remove fdiv.
1662         * config/arm/vfp.md:
1663         (*sqrtsf2_vfp): Update for attribute changes.
1664         (*sqrtdf2_vfp): Likewise.
1665         * config/aarch64/aarch64.md:
1666         (sqrt<mode>2): Update for attribute changes.
1667         * config/arm/arm1020e.md: Update with new attributes.
1668         * config/arm/cortex-a15-neon.md: Update with new attributes.
1669         * config/arm/cortex-a5.md: Update with new attributes.
1670         * config/arm/cortex-a53.md: Update with new attributes.
1671         * config/arm/cortex-a7.md: Update with new attributes.
1672         * config/arm/cortex-a8-neon.md: Update with new attributes.
1673         * config/arm/cortex-a9.md: Update with new attributes.
1674         * config/arm/cortex-m4-fpu.md: Update with new attributes.
1675         * config/arm/cortex-r4f.md: Update with new attributes.
1676         * config/arm/marvell-pj4.md: Update with new attributes.
1677         * config/arm/vfp11.md: Update with new attributes.
1679 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1681         * config/arm/types.md
1682         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
1683         * config/aarch64/aarch64.md
1684         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
1685         new attributes.
1686         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
1687         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
1688         (float<GPI:mode><GPF:mode>2): Likewise.
1689         * config/arm/vfp.md
1690         (*truncsisf2_vfp): Update with new attributes.
1691         (*truncsidf2_vfp): Likewise.
1692         (fixuns_truncsfsi2): Likewise.
1693         (fixuns_truncdfsi2): Likewise.
1694         (*floatsisf2_vfp): Likewise.
1695         (*floatsidf2_vfp): Likewise.
1696         (floatunssisf2): Likewise.
1697         (floatunssidf2): Likewise.
1698         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
1699         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
1700         * config/arm/arm1020e.md: Update with new attributes.
1701         * config/arm/cortex-a15-neon.md: Update with new attributes.
1702         * config/arm/cortex-a5.md: Update with new attributes.
1703         * config/arm/cortex-a53.md: Update with new attributes.
1704         * config/arm/cortex-a7.md: Update with new attributes.
1705         * config/arm/cortex-a8-neon.md: Update with new attributes.
1706         * config/arm/cortex-a9.md: Update with new attributes.
1707         * config/arm/cortex-m4-fpu.md: Update with new attributes.
1708         * config/arm/cortex-r4f.md: Update with new attributes.
1709         * config/arm/marvell-pj4.md: Update with new attributes.
1710         * config/arm/vfp11.md: Update with new attributes.
1712 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1714         * config/aarch64/arm_neon.h
1715         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
1716         (vqtbx<1,2,3,4><q>_s8): Likewise.
1718 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1720         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
1721         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
1722         (add<mode>3): Add type attribute.
1723         (add<mode>3): Likewise.
1724         (usadd<mode>3): Likewise.
1725         (ssadd<mode>3): Likewise.
1726         (sub<mode>3): Likewise.
1727         (sub<mode>3): Likewise.
1728         (ussub<mode>3): Likewise.
1729         (sssub<mode>3): Likewise.
1730         (ssmulsa3): Likewise.
1731         (usmulusa3): Likewise.
1732         (arm_usatsihi): Likewise.
1733         * config/arm/vfp.md
1734         (*movdi_vfp): Add types for all instructions.
1735         (*movdi_vfp_cortexa8): Likewise.
1736         (*movhf_vfp_neon): Likewise.
1737         (*movhf_vfp): Likewise.
1738         (*movdf_vfp): Likewise.
1739         (*thumb2_movdf_vfp): Likewise.
1740         (*thumb2_movdfcc_vfp): Likewise.
1741         * config/arm/arm.md: Add type attribute to all insn patterns.
1742         (*thumb1_adddi3): Add type attribute.
1743         (*arm_adddi3): Likewise.
1744         (*adddi_sesidi_di): Likewise.
1745         (*adddi_zesidi_di): Likewise.
1746         (*thumb1_addsi3): Likewise.
1747         (addsi3_compare0): Likewise.
1748         (*addsi3_compare0_scratch): Likewise.
1749         (*compare_negsi_si): Likewise.
1750         (cmpsi2_addneg): Likewise.
1751         (*addsi3_carryin_<optab>): Likewise.
1752         (*addsi3_carryin_alt2_<optab>): Likewise.
1753         (*addsi3_carryin_clobercc_<optab>): Likewise.
1754         (*subsi3_carryin): Likewise.
1755         (*subsi3_carryin_const): Likewise.
1756         (*subsi3_carryin_compare): Likewise.
1757         (*subsi3_carryin_compare_const): Likewise.
1758         (*arm_subdi3): Likewise.
1759         (*thumb_subdi3): Likewise.
1760         (*subdi_di_zesidi): Likewise.
1761         (*subdi_di_sesidi): Likewise.
1762         (*subdi_zesidi_di): Likewise.
1763         (*subdi_sesidi_di): Likewise.
1764         (*subdi_zesidi_ze): Likewise.
1765         (thumb1_subsi3_insn): Likewise.
1766         (*arm_subsi3_insn): Likewise.
1767         (*anddi3_insn): Likewise.
1768         (*anddi_zesidi_di): Likewise.
1769         (*anddi_sesdi_di): Likewise.
1770         (*ne_zeroextracts): Likewise.
1771         (*ne_zeroextracts): Likewise.
1772         (*ite_ne_zeroextr): Likewise.
1773         (*ite_ne_zeroextr): Likewise.
1774         (*anddi_notdi_di): Likewise.
1775         (*anddi_notzesidi): Likewise.
1776         (*anddi_notsesidi): Likewise.
1777         (andsi_notsi_si): Likewise.
1778         (thumb1_bicsi3): Likewise.
1779         (*iordi3_insn): Likewise.
1780         (*iordi_zesidi_di): Likewise.
1781         (*iordi_sesidi_di): Likewise.
1782         (*thumb1_iorsi3_insn): Likewise.
1783         (*xordi3_insn): Likewise.
1784         (*xordi_zesidi_di): Likewise.
1785         (*xordi_sesidi_di): Likewise.
1786         (*arm_xorsi3): Likewise.
1787         (*andsi_iorsi3_no): Likewise.
1788         (*smax_0): Likewise.
1789         (*smax_m1): Likewise.
1790         (*arm_smax_insn): Likewise.
1791         (*smin_0): Likewise.
1792         (*arm_smin_insn): Likewise.
1793         (*arm_umaxsi3): Likewise.
1794         (*arm_uminsi3): Likewise.
1795         (*minmax_arithsi): Likewise.
1796         (*minmax_arithsi_): Likewise.
1797         (*satsi_<SAT:code>): Likewise.
1798         (arm_ashldi3_1bit): Likewise.
1799         (arm_ashrdi3_1bit): Likewise.
1800         (arm_lshrdi3_1bit): Likewise.
1801         (*arm_negdi2): Likewise.
1802         (*thumb1_negdi2): Likewise.
1803         (*arm_negsi2): Likewise.
1804         (*thumb1_negsi2): Likewise.
1805         (*negdi_extendsid): Likewise.
1806         (*negdi_zero_extend): Likewise.
1807         (*arm_abssi2): Likewise.
1808         (*thumb1_abssi2): Likewise.
1809         (*arm_neg_abssi2): Likewise.
1810         (*thumb1_neg_abss): Likewise.
1811         (one_cmpldi2): Likewise.
1812         (extend<mode>di2): Likewise.
1813         (*compareqi_eq0): Likewise.
1814         (*arm_extendhisi2addsi): Likewise.
1815         (*arm_movdi): Likewise.
1816         (*thumb1_movdi_insn): Likewise.
1817         (*arm_movt): Likewise.
1818         (*thumb1_movsi_insn): Likewise.
1819         (pic_add_dot_plus_four): Likewise.
1820         (pic_add_dot_plus_eight): Likewise.
1821         (tls_load_dot_plus_eight): Likewise.
1822         (*thumb1_movhi_insn): Likewise.
1823         (*thumb1_movsf_insn): Likewise.
1824         (*movdf_soft_insn): Likewise.
1825         (*thumb_movdf_insn): Likewise.
1826         (cbranchsi4_insn): Likewise.
1827         (cbranchsi4_scratch): Likewise.
1828         (*negated_cbranchsi4): Likewise.
1829         (*tbit_cbranch): Likewise.
1830         (*tlobits_cbranch): Likewise.
1831         (*tstsi3_cbranch): Likewise.
1832         (*cbranchne_decr1): Likewise.
1833         (*addsi3_cbranch): Likewise.
1834         (*addsi3_cbranch_scratch): Likewise.
1835         (*arm_cmpdi_insn): Likewise.
1836         (*arm_cmpdi_unsig): Likewise.
1837         (*arm_cmpdi_zero): Likewise.
1838         (*thumb_cmpdi_zero): Likewise.
1839         (*deleted_compare): Likewise.
1840         (*mov_scc): Likewise.
1841         (*mov_negscc): Likewise.
1842         (*mov_notscc): Likewise.
1843         (*cstoresi_eq0_thumb1_insn): Likewise.
1844         (cstoresi_nltu_thumb1): Likewise.
1845         (cstoresi_ltu_thu): Likewise.
1846         (thumb1_addsi3_addgeu): Likewise.
1847         (*arm_jump): Likewise.
1848         (*thumb_jump): Likewise.
1849         (*check_arch2): Likewise.
1850         (arm_casesi_internal): Likewise.
1851         (thumb1_casesi_dispatch): Likewise.
1852         (*arm_indirect_jump): Likewise.
1853         (*thumb1_indirect_jump): Likewise.
1854         (nop): Likewise.
1855         (*and_scc): Likewise.
1856         (*ior_scc): Likewise.
1857         (*compare_scc): Likewise.
1858         (*cond_move): Likewise.
1859         (*cond_arith): Likewise.
1860         (*cond_sub): Likewise.
1861         (*cmp_ite0): Likewise.
1862         (*cmp_ite1): Likewise.
1863         (*cmp_and): Likewise.
1864         (*cmp_ior): Likewise.
1865         (*ior_scc_scc): Likewise.
1866         (*ior_scc_scc_cmp): Likewise.
1867         (*and_scc_scc): Likewise.
1868         (*and_scc_scc_cmp): Likewise.
1869         (*and_scc_scc_nod): Likewise.
1870         (*negscc): Likewise.
1871         (movcond_addsi): Likewise.
1872         (movcond): Likewise.
1873         (*ifcompare_plus_move): Likewise.
1874         (*if_plus_move): Likewise.
1875         (*ifcompare_move_plus): Likewise.
1876         (*if_move_plus): Likewise.
1877         (*ifcompare_arith_arith): Likewise.
1878         (*if_arith_arith): Likewise.
1879         (*ifcompare_arith_move): Likewise.
1880         (*if_arith_move): Likewise.
1881         (*ifcompare_move_arith): Likewise.
1882         (*if_move_arith): Likewise.
1883         (*ifcompare_move_not): Likewise.
1884         (*if_move_not): Likewise.
1885         (*ifcompare_not_move): Likewise.
1886         (*if_not_move): Likewise.
1887         (*ifcompare_shift_move): Likewise.
1888         (*if_shift_move): Likewise.
1889         (*ifcompare_move_shift): Likewise.
1890         (*if_move_shift): Likewise.
1891         (*ifcompare_shift_shift): Likewise.
1892         (*ifcompare_not_arith): Likewise.
1893         (*ifcompare_arith_not): Likewise.
1894         (*if_arith_not): Likewise.
1895         (*ifcompare_neg_move): Likewise.
1896         (*if_neg_move): Likewise.
1897         (*ifcompare_move_neg): Likewise.
1898         (*if_move_neg): Likewise.
1899         (prologue_thumb1_interwork): Likewise.
1900         (*cond_move_not): Likewise.
1901         (*sign_extract_onebit): Likewise.
1902         (*not_signextract_onebit): Likewise.
1903         (stack_tie): Likewise.
1904         (align_4): Likewise.
1905         (align_8): Likewise.
1906         (consttable_end): Likewise.
1907         (consttable_1): Likewise.
1908         (consttable_2): Likewise.
1909         (consttable_4): Likewise.
1910         (consttable_8): Likewise.
1911         (consttable_16): Likewise.
1912         (*thumb1_tablejump): Likewise.
1913         (prefetch): Likewise.
1914         (force_register_use): Likewise.
1915         (thumb_eh_return): Likewise.
1916         (load_tp_hard): Likewise.
1917         (load_tp_soft): Likewise.
1918         (tlscall): Likewise.
1919         (*arm_movtas_ze): Likewise.
1920         (*arm_rev): Likewise.
1921         (*arm_revsh): Likewise.
1922         (*arm_rev16): Likewise.
1923         * config/arm/thumb2.md
1924         (*thumb2_smaxsi3): Likewise.
1925         (*thumb2_sminsi3): Likewise.
1926         (*thumb32_umaxsi3): Likewise.
1927         (*thumb2_uminsi3): Likewise.
1928         (*thumb2_negdi2): Likewise.
1929         (*thumb2_abssi2): Likewise.
1930         (*thumb2_neg_abss): Likewise.
1931         (*thumb2_movsi_insn): Likewise.
1932         (tls_load_dot_plus_four): Likewise.
1933         (*thumb2_movhi_insn): Likewise.
1934         (*thumb2_mov_scc): Likewise.
1935         (*thumb2_mov_negs): Likewise.
1936         (*thumb2_mov_negs): Likewise.
1937         (*thumb2_mov_nots): Likewise.
1938         (*thumb2_mov_nots): Likewise.
1939         (*thumb2_movsicc_): Likewise.
1940         (*thumb2_movsfcc_soft_insn): Likewise.
1941         (*thumb2_indirect_jump): Likewise.
1942         (*thumb2_and_scc): Likewise.
1943         (*thumb2_ior_scc): Likewise.
1944         (*thumb2_ior_scc_strict_it): Likewise.
1945         (*thumb2_cond_move): Likewise.
1946         (*thumb2_cond_arith): Likewise.
1947         (*thumb2_cond_ari): Likewise.
1948         (*thumb2_cond_sub): Likewise.
1949         (*thumb2_negscc): Likewise.
1950         (*thumb2_movcond): Likewise.
1951         (thumb2_casesi_internal): Likewise.
1952         (thumb2_casesi_internal_pic): Likewise.
1953         (*thumb2_alusi3_short): Likewise.
1954         (*thumb2_mov<mode>_shortim): Likewise.
1955         (*thumb2_addsi_short): Likewise.
1956         (*thumb2_subsi_short): Likewise.
1957         (thumb2_addsi3_compare0): Likewise.
1958         (*thumb2_cbz): Likewise.
1959         (*thumb2_cbnz): Likewise.
1960         (*thumb2_one_cmplsi2_short): Likewise.
1961         (*thumb2_negsi2_short): Likewise.
1962         (*orsi_notsi_si): Likewise.
1963         * config/arm/arm1020e.md: Update with new attributes.
1964         * config/arm/arm1026ejs.md: Update with new attributes.
1965         * config/arm/arm1136jfs.md: Update with new attributes.
1966         * config/arm/arm926ejs.md: Update with new attributes.
1967         * config/arm/cortex-a15.md: Update with new attributes.
1968         * config/arm/cortex-a5.md: Update with new attributes.
1969         * config/arm/cortex-a53.md: Update with new attributes.
1970         * config/arm/cortex-a7.md: Update with new attributes.
1971         * config/arm/cortex-a8.md: Update with new attributes.
1972         * config/arm/cortex-a9.md: Update with new attributes.
1973         * config/arm/cortex-m4.md: Update with new attributes.
1974         * config/arm/cortex-r4.md: Update with new attributes.
1975         * config/arm/fa526.md: Update with new attributes.
1976         * config/arm/fa606te.md: Update with new attributes.
1977         * config/arm/fa626te.md: Update with new attributes.
1978         * config/arm/fa726te.md: Update with new attributes.
1980 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
1982         * config/aarch64/aarch64-simd.md
1983         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
1984         <vwx> iterator to ensure correct register choice.
1985         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
1986         (aarch64_sqdmull_n<mode>): Likewise.
1987         (aarch64_sqdmull2_n<mode>_internal): Likewise.
1988         * config/aarch64/arm_neon.h
1989         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
1990         (vml<as><q>_n_<su>16): Likewise.
1991         (vml<as>l_high_lane<q>_<su>16): Likewise.
1992         (vml<as>l_high_n_<su>16): Likewise.
1993         (vml<as>l_lane<q>_<su>16): Likewise.
1994         (vml<as>l_n_<su>16): Likewise.
1995         (vmul<q>_lane<q>_<su>16): Likewise.
1996         (vmul<q>_n_<su>16): Likewise.
1997         (vmull_lane<q>_<su>16): Likewise.
1998         (vmull_n_<su>16): Likewise.
1999         (vmull_high_lane<q>_<su>16): Likewise.
2000         (vmull_high_n_<su>16): Likewise.
2001         (vqrdmulh<q>_n_s16): Likewise.
2003 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
2005         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
2006         have the correct lane parameter.
2008 2013-09-06  Richard Biener <rguenther@suse.de>
2010         * cfganal.c (control_dependences::~control_dependences):
2011         Properly free all of the vector.
2013 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
2015         PR target/58269
2016         * config/i386/i386.c (ix86_conditional_register_usage):
2017         Proper initialize extended SSE registers.
2019 2013-09-06  Jan Hubicka  <jh@suse.cz>
2021         PR tree-optimization/58311
2022         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
2024 2013-09-06  Jan Hubicka  <jh@suse.cz>
2026         * Makefile.in (tree-sra.o): Update dependencies.
2027         * tree-sra.c: Include ipa-utils.h
2028         (scan_function): Use recursive_call_p.
2029         (has_caller_p): New function.
2030         (cgraph_for_node_and_aliases): Count also callers of aliases.
2032 2013-09-06  Jan Hubicka  <jh@suse.cz>
2034         PR middle-end/58094
2035         * cgraph.h (symtab_semantically_equivalent_p): Declare.
2036         * tree-tailcall.c: Include ipa-utils.h.
2037         (find_tail_calls): Use it.
2038         * ipa-pure-const.c (check_call): Likewise.
2039         * ipa-utils.c (recursive_call_p): New function.
2040         * ipa-utils.h (recursive_call_p): Dclare.
2041         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
2042         (symtab_semantically_equivalent_p): New function.
2043         * Makefile.in (tree-tailcall.o): Update dependencies.
2045 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
2047         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
2048         non-inlinable part.
2050 2013-09-06  Richard Biener  <rguenther@suse.de>
2052         * lto-streamer.h (lto_global_var_decls): Remove.
2053         * Makefile.in (OBJS): Remove lto-symtab.o.
2054         (lto-symtab.o): Remove.
2055         (GTFILES): Remove lto-symtab.c
2056         * lto-symtab.c: Move to lto/
2058 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2060         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
2061         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
2062         (UNSPEC_FPINT_RINT): New constant definitions.
2063         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
2064         definition with 2 attributes.
2065         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
2066         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
2067         definitions.
2069 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2071         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
2072         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
2073         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
2074         attribute to "z196_alone".
2075         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
2076         "zEC12_simple".
2078 2013-09-06  Richard Biener  <rguenther@suse.de>
2080         * basic-block.h (class control_dependences): New.
2081         * tree-ssa-dce.c (control_dependence_map): Remove.
2082         (cd): New global.
2083         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
2084         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
2085         find_pdom, find_control_dependence, find_all_control_dependences):
2086         Move to cfganal.c.
2087         (mark_control_dependent_edges_necessary,
2088         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
2089         tree_dce_done, perform_tree_ssa_dce): Adjust.
2090         * cfganal.c (set_control_dependence_map_bit,
2091         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
2092         find_all_control_dependences): Move from tree-ssa-dce.c and
2093         implement as methods of control_dependences class.
2094         (control_dependences::control_dependences): New.
2095         (control_dependences::~control_dependences): Likewise.
2096         (control_dependences::get_edges_dependent_on): Likewise.
2097         (control_dependences::get_edge): Likewise.
2099 2013-09-04  Jan Hubicka  <jh@suse.cz>
2101         * tree.c (types_same_for_odr): Drop overactive check.
2102         * ipa-devirt.c (hash_type_name): Likewise.
2104 2013-09-04  Jan Hubicka  <jh@suse.cz>
2106         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
2107         (analyze_functions): ... here.
2109 2013-09-04  Jan Hubicka  <jh@suse.cz>
2111         PR middle-end/58201
2112         * cgraphunit.c (analyze_functions): Clear AUX fields
2113         after processing; initialize assembler name has.
2115 2013-09-05  Jeff Law  <law@redhat.com>
2117         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
2118         from thread_around_empty_block.  Record threading path into PATH.
2119         Recurse if threading through the initial block is successful.
2120         (thread_across_edge): Corresponding changes to slightly simplify.
2122 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
2124         * config/aarch64/aarch64.md
2125         (type): Remove frecpe, frecps, frecpx.
2126         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
2127         fix to be a TARGET_SIMD instruction.
2128         (aarch64_frecps): Remove.
2129         * config/aarch64/aarch64-simd.md
2130         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
2131         (aarch64_frecps<mode>): Handle all float/vector of float modes.
2133 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
2134             Sofiane Naci  <sofiane.naci@arm.com>
2136         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
2137         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
2138         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
2139         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
2140         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
2141         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
2142         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
2143         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
2144         into "clz, "rbit".  Rename "shift" to "shift_imm".
2145         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
2146         changes.  Update for attribute changes all occurrences of arlo_* and
2147         shift* types.
2148         * config/arm/arm-fixed.md: Update for attribute changes
2149         all occurrences of arlo_* types.
2150         * config/arm/thumb2.md: Update for attribute changes all occurrences
2151         of arlo_* types.
2152         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
2153         (cortexa7_older_only): Likewise.
2154         (cortexa7_younger):  Likewise.
2155         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
2156         (1020alu_shift_op): Likewise.
2157         (1020alu_shift_reg_op): Likewise.
2158         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
2159         (alu_shift_op): Likewise.
2160         (alu_shift_reg_op): Likewise.
2161         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
2162         (11_alu_shift_op): Likewise.
2163         (11_alu_shift_reg_op): Likewise.
2164         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
2165         (9_alu_shift_reg_op): Likewise.
2166         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
2167         attribute changes.
2168         (cortex_a15_alu_shift): Likewise.
2169         (cortex_a15_alu_shift_reg): Likewise.
2170         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
2171         attribute changes.
2172         (cortex_a5_alu_shift): Likewise.
2173         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
2174         attribute changes.
2175         (cortex_a53_alu_shift): Likewise.
2176         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
2177         attribute changes.
2178         (cortex_a7_alu_reg): Likewise.
2179         (cortex_a7_alu_shift): Likewise.
2180         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
2181         attribute changes.
2182         (cortex_a8_alu_shift): Likewise.
2183         (cortex_a8_alu_shift_reg): Likewise.
2184         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
2185         (cortex_a9_dp_shift): Likewise.
2186         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
2187         attribute changes.
2188         * config/arm/cortex-r4.md
2189         (cortex_r4_alu): Update for attribute changes.
2190         (cortex_r4_mov): Likewise.
2191         (cortex_r4_alu_shift_reg): Likewise.
2192         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
2193         (526_alu_shift_op): Likewise.
2194         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
2195         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
2196         (626te_alu_shift_op): Likewise.
2197         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
2198         (726te_alu_shift_op): Likewise.
2199         (726te_alu_shift_reg_op): Likewise.
2200         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
2201         (mp626_alu_shift_op): Likewise.
2202         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
2203         (pj4_alu_conds): Likewise.
2204         (pj4_shift): Likewise.
2205         (pj4_shift_conds): Likewise.
2206         (pj4_alu_shift): Likewise.
2207         (pj4_alu_shift_conds): Likewise.
2208         * config/aarch64/aarch64.md: Update for attribute change
2209         all occurrences of arlo_* and shift* types.
2211 2013-09-05  Mike Stump  <mikestump@comcast.net>
2213         * tree.h: Move documentation for tree_function_decl to tree-core.h
2214         with the declaration.
2216 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
2218         PR target/58139
2219         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
2220         looking for widest mode.
2222 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
2224         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
2226 2013-09-05  Richard Biener  <rguenther@suse.de>
2228         PR tree-optimization/58137
2229         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
2230         Do not create vectors of pointers.
2231         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
2232         types for the components of the vector initializer.
2233         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
2234         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
2236 2013-09-05  Martin Jambor  <mjambor@suse.cz>
2238         * ipa-prop.c (remove_described_reference): Accept missing references,
2239         return false if that hppens, otherwise return true.
2240         (cgraph_node_for_jfunc): New function.
2241         (try_decrement_rdesc_refcount): Likewise.
2242         (try_make_edge_direct_simple_call): Use them.
2243         (ipa_edge_removal_hook): Remove references from rdescs.
2244         (ipa_edge_duplication_hook): Clone rdescs and their references
2245         when the new edge has the same caller as the old one.
2246         * cgraph.c (cgraph_resolve_speculation): Remove speculative
2247         reference before removing any edges.
2249 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
2251         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
2252         initial store.
2253         * thumb2.md (thumb2_storewb_parisi): New pattern.
2255 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
2257         * config/aarch64/aarch64-option-extensions.def: Add
2258         AARCH64_OPT_EXTENSION of 'crc'.
2259         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
2260         (AARCH64_ISA_CRC): Ditto.
2261         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
2262         description of the CRC extension.
2264 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2266         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
2267         * config/rs6000/linux.h: Ditto.
2268         * alpha/linux.h: Ditto.
2269         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
2270         no_c99_libc_has_function.
2271         * config/c6x/uclinux-elf.h: Ditto.
2272         * config/lm32/uclinux-elf.h: Ditto.
2273         * config/m68k/uclinux.h: Ditto.
2274         * config/moxie/uclinux.h: Ditto.
2275         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
2276         (crisv32-*-linux*, cris-*-linux*): Ditto.
2277         * config/bfin/bfin.c: Include "tm_p.h".
2279 2013-09-05  Richard Biener  <rguenther@suse.de>
2281         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
2282         check for a definition without a basic-block.
2284 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
2285             Sofiane Naci <sofiane.naci@arm.com>
2287         * config/aarch64/aarch64.md
2288         (*movti_aarch64): Rename r_2_f and f_2_r.
2289         (*movsf_aarch64): Likewise.
2290         (*movdf_aarch64): Likewise.
2291         (*movtf_aarch64): Likewise.
2292         (aarch64_movdi_<mode>low): Likewise.
2293         (aarch64_movdi_<mode>high): Likewise.
2294         (aarch64_mov<mode>high_di): Likewise.
2295         (aarch64_mov<mode>low_di): Likewise.
2296         (aarch64_movtilow_tilow): Likewise.
2297         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
2298         values to config/arm/types.md
2299         (attribute "conds"): Update for attribute change.
2300         (anddi3_insn): Likewise.
2301         (iordi3_insn): Likewise.
2302         (xordi3_insn): Likewise.
2303         (one_cmpldi2): Likewise.
2304         * config/arm/types.md (type): Add Neon types.
2305         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
2306         use "type" attribute.
2307         (movmisalign<mode>_neon_store): Likewise.
2308         (movmisalign<mode>_neon_load): Likewise.
2309         (vec_set<mode>_internal): Likewise.
2310         (vec_setv2di_internal): Likewise.
2311         (vec_extract<mode>): Likewise.
2312         (vec_extractv2di): Likewise.
2313         (add<mode>3_neon): Likewise.
2314         (adddi3_neon): Likewise.
2315         (sub<mode>3_neon): Likewise.
2316         (subdi3_neon): Likewise.
2317         (mul<mode>3_neon): Likewise.
2318         (mul<mode>3add<mode>_neon): Likewise.
2319         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
2320         (fma<VCVTF:mode>4)): Likewise.
2321         (fma<VCVTF:mode>4_intrinsic): Likewise.
2322         (fmsub<VCVTF:mode>4)): Likewise.
2323         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
2324         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
2325         (ior<mode>3): Likewise.
2326         (and<mode>3): Likewise.
2327         (anddi3_neon): Likewise.
2328         (orn<mode>3_neon): Likewise.
2329         (orndi3_neon): Likewise.
2330         (bic<mode>3_neon): Likewise.
2331         (bicdi3_neon): Likewise.
2332         (xor<mode>3): Likewise.
2333         (one_cmpl<mode>2): Likewise.
2334         (abs<mode>2): Likewise.
2335         (neg<mode>2): Likewise.
2336         (umin<mode>3_neon): Likewise.
2337         (umax<mode>3_neon): Likewise.
2338         (smin<mode>3_neon): Likewise.
2339         (smax<mode>3_neon): Likewise.
2340         (vashl<mode>3): Likewise.
2341         (vashr<mode>3_imm): Likewise.
2342         (vlshr<mode>3_imm): Likewise.
2343         (ashl<mode>3_signed): Likewise.
2344         (ashl<mode>3_unsigned): Likewise.
2345         (neon_load_count): Likewise.
2346         (ashldi3_neon_noclobber): Likewise.
2347         (signed_shift_di3_neon): Likewise.
2348         (unsigned_shift_di3_neon): Likewise.
2349         (ashrdi3_neon_imm_noclobber): Likewise.
2350         (lshrdi3_neon_imm_noclobber): Likewise.
2351         (widen_ssum<mode>3): Likewise.
2352         (widen_usum<mode>3): Likewise.
2353         (quad_halves_<code>v4si): Likewise.
2354         (quad_halves_<code>v4sf): Likewise.
2355         (quad_halves_<code>v8hi): Likewise.
2356         (quad_halves_<code>v16qi): Likewise.
2357         (reduc_splus_v2di): Likewise.
2358         (neon_vpadd_internal<mode>): Likewise.
2359         (neon_vpsmin<mode>): Likewise.
2360         (neon_vpsmax<mode>): Likewise.
2361         (neon_vpumin<mode>): Likewise.
2362         (neon_vpumax<mode>): Likewise.
2363         (ss_add<mode>_neon): Likewise.
2364         (us_add<mode>_neon): Likewise.
2365         (ss_sub<mode>_neon): Likewise.
2366         (us_sub<mode>_neon): Likewise.
2367         (neon_vadd<mode>_unspec): Likewise.
2368         (neon_vaddl<mode>): Likewise.
2369         (neon_vaddw<mode>): Likewise.
2370         (neon_vhadd<mode>): Likewise.
2371         (neon_vqadd<mode>): Likewise.
2372         (neon_vaddhn<mode>): Likewise.
2373         (neon_vmul<mode>): Likewise.
2374         (neon_vmla<mode>): Likewise.
2375         (neon_vmlal<mode>): Likewise.
2376         (neon_vmls<mode>): Likewise.
2377         (neon_vmlsl<mode>): Likewise.
2378         (neon_vqdmulh<mode>): Likewise.
2379         (neon_vqdmlal<mode>): Likewise.
2380         (neon_vqdmlsl<mode>): Likewise.
2381         (neon_vmull<mode>): Likewise.
2382         (neon_vqdmull<mode>): Likewise.
2383         (neon_vsub<mode>_unspec): Likewise.
2384         (neon_vsubl<mode>): Likewise.
2385         (neon_vsubw<mode>): Likewise.
2386         (neon_vqsub<mode>): Likewise.
2387         (neon_vhsub<mode>): Likewise.
2388         (neon_vsubhn<mode>): Likewise.
2389         (neon_vceq<mode>): Likewise.
2390         (neon_vcge<mode>): Likewise.
2391         (neon_vcgeu<mode>): Likewise.
2392         (neon_vcgt<mode>): Likewise.
2393         (neon_vcgtu<mode>): Likewise.
2394         (neon_vcle<mode>): Likewise.
2395         (neon_vclt<mode>): Likewise.
2396         (neon_vcage<mode>): Likewise.
2397         (neon_vcagt<mode>): Likewise.
2398         (neon_vtst<mode>): Likewise.
2399         (neon_vabd<mode>): Likewise.
2400         (neon_vabdl<mode>): Likewise.
2401         (neon_vaba<mode>): Likewise.
2402         (neon_vabal<mode>): Likewise.
2403         (neon_vmax<mode>): Likewise.
2404         (neon_vmin<mode>): Likewise.
2405         (neon_vpaddl<mode>): Likewise.
2406         (neon_vpadal<mode>): Likewise.
2407         (neon_vpmax<mode>): Likewise.
2408         (neon_vpmin<mode>): Likewise.
2409         (neon_vrecps<mode>): Likewise.
2410         (neon_vrsqrts<mode>): Likewise.
2411         (neon_vqabs<mode>): Likewise.
2412         (neon_vqneg<mode>): Likewise.
2413         (neon_vcls<mode>): Likewise.
2414         (clz<mode>2): Likewise.
2415         (popcount<mode>2): Likewise.
2416         (neon_vrecpe): Likewise.
2417         (neon_vrsqrte): Likewise.
2418         (neon_vget_lane<mode>_sext_internal): Likewise.
2419         (neon_vget_lane<mode>_zext_internal): Likewise.
2420         (neon_vdup_n<mode>): Likewise.
2421         (neon_vdup_nv2di): Likewise.
2422         (neon_vdpu_lane<mode>_internal): Likewise.
2423         (neon_vswp<mode>): Likewise.
2424         (float<mode><V_cvtto>2): Likewise.
2425         (floatuns<mode><V_cvtto>2): Likewise.
2426         (fix_trunc<mode><V_cvtto>)2): Likewise
2427         (fixuns_trunc<mode><V_cvtto)2): Likewise.
2428         (neon_vcvt<mode>): Likewise.
2429         (neon_vcvtv4sfv4hf): Likewise.
2430         (neon_vcvtv4hfv4sf): Likewise.
2431         (neon_vcvt_n<mode>): Likewise.
2432         (neon_vmovn<mode>): Likewise.
2433         (neon_vqmovn<mode>): Likewise.
2434         (neon_vqmovun<mode>): Likewise.
2435         (neon_vmovl<mode>): Likewise.
2436         (neon_vmul_lane<mode>): Likewise.
2437         (neon_vmull_lane<mode>): Likewise.
2438         (neon_vqdmull_lane<mode>): Likewise.
2439         (neon_vqdmulh_lane<mode>): Likewise.
2440         (neon_vmla_lane<mode>): Likewise.
2441         (neon_vmlal_lane<mode>): Likewise.
2442         (neon_vqdmlal_lane<mode>): Likewise.
2443         (neon_vmls_lane<mode>): Likewise.
2444         (neon_vmlsl_lane<mode>): Likewise.
2445         (neon_vqdmlsl_lane<mode>): Likewise.
2446         (neon_vext<mode>): Likewise.
2447         (neon_vrev64<mode>): Likewise.
2448         (neon_vrev32<mode>): Likewise.
2449         (neon_vrev16<mode>): Likewise.
2450         (neon_vbsl<mode>_internal): Likewise.
2451         (neon_vshl<mode>): Likewise.
2452         (neon_vqshl<mode>): Likewise.
2453         (neon_vshr_n<mode>): Likewise.
2454         (neon_vshrn_n<mode>): Likewise.
2455         (neon_vqshrn_n<mode>): Likewise.
2456         (neon_vqshrun_n<mode>): Likewise.
2457         (neon_vshl_n<mode>): Likewise.
2458         (neon_vqshl_n<mode>): Likewise.
2459         (neon_vqshlu_n<mode>): Likewise.
2460         (neon_vshll_n<mode>): Likewise.
2461         (neon_vsra_n<mode>): Likewise.
2462         (neon_vsri_n<mode>): Likewise.
2463         (neon_vsli_n<mode>): Likewise.
2464         (neon_vtbl1v8qi): Likewise.
2465         (neon_vtbl2v8qi): Likewise.
2466         (neon_vtbl3v8qi): Likewise.
2467         (neon_vtbl4v8qi): Likewise.
2468         (neon_vtbx1v8qi): Likewise.
2469         (neon_vtbx2v8qi): Likewise.
2470         (neon_vtbx3v8qi): Likewise.
2471         (neon_vtbx4v8qi): Likewise.
2472         (neon_vtrn<mode>_internal): Likewise.
2473         (neon_vzip<mode>_internal): Likewise.
2474         (neon_vuzp<mode>_internal): Likewise.
2475         (neon_vld1<mode>): Likewise.
2476         (neon_vld1_lane<mode>): Likewise.
2477         (neon_vld1_dup<mode>): Likewise.
2478         (neon_vld1_dupv2di): Likewise.
2479         (neon_vst1<mode>): Likewise.
2480         (neon_vst1_lane<mode>): Likewise.
2481         (neon_vld2<mode>): Likewise.
2482         (neon_vld2_lane<mode>): Likewise.
2483         (neon_vld2_dup<mode>): Likewise.
2484         (neon_vst2<mode>): Likewise.
2485         (neon_vst2_lane<mode>): Likewise.
2486         (neon_vld3<mode>): Likewise.
2487         (neon_vld3qa<mode>): Likewise.
2488         (neon_vld3qb<mode>): Likewise.
2489         (neon_vld3_lane<mode>): Likewise.
2490         (neon_vld3_dup<mode>): Likewise.
2491         (neon_vst3<mode>): Likewise.
2492         (neon_vst3qa<mode>): Likewise.
2493         (neon_vst3qb<mode>): Likewise.
2494         (neon_vst3_lane<mode>): Likewise.
2495         (neon_vld4<mode>): Likewise.
2496         (neon_vld4qa<mode>): Likewise.
2497         (neon_vld4qb<mode>): Likewise.
2498         (neon_vld4_lane<mode>): Likewise.
2499         (neon_vld4_dup<mode>): Likewise.
2500         (neon_vst4<mode>): Likewise.
2501         (neon_vst4qa<mode>): Likewise.
2502         (neon_vst4qb<mode>): Likewise.
2503         (neon_vst4_lane<mode>): Likewise.
2504         (neon_vec_unpack<US>_lo_<mode>): Likewise.
2505         (neon_vec_unpack<US>_hi_<mode>): Likewise.
2506         (neon_vec_<US>mult_lo_<mode>): Likewise.
2507         (neon_vec_<US>mult_hi_<mode>): Likewise.
2508         (neon_vec_<US>shiftl_<mode>): Likewise.
2509         (neon_unpack<US>_<mode>): Likewise.
2510         (neon_vec_<US>mult_<mode>): Likewise.
2511         (vec_pack_trunc_<mode>): Likewise.
2512         (neon_vec_pack_trunk_<mode>): Likewise.
2513         (neon_vabd<mode>_2): Likewise.
2514         (neon_vabd<mode>_3): Likewise.
2515         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
2516         (thumb2_movsi_vfp): Likewise.
2517         (movdi_vfp): Likewise.
2518         (movdi_vfp_cortexa8): Likewise.
2519         (movhf_vfp_neon): Likewise.
2520         (movhf_vfp): Likewiwse.
2521         (movsf_vfp): Likewiwse.
2522         (thumb2_movsf_vfp): Likewiwse.
2523         (movdf_vfp): Likewise.
2524         (thumb2_movdf_vfp): Likewise.
2525         (movsfcc_vfp): Likewise.
2526         (thumb2_movsfcc_vfp): Likewise.
2527         (movdfcc_vfp): Likewise.
2528         (thumb2_movdfcc_vfp): Likewise.
2529         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
2530         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
2531         (v10_v2c): Likewise.
2532         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
2533         attribute change.
2534         (cortex_a15_neon_int_2): Likewise.
2535         (cortex_a15_neon_int_3): Likewise.
2536         (cortex_a15_neon_int_4): Likewise.
2537         (cortex_a15_neon_int_5): Likewise.
2538         (cortex_a15_neon_vqneg_vqabs): Likewise.
2539         (cortex_a15_neon_vmov): Likewise.
2540         (cortex_a15_neon_vaba): Likewise.
2541         (cortex_a15_neon_vaba_qqq): Likewise.
2542         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2543         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
2544         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
2545         scalar_64_32_long_scalar): Likewise.
2546         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2547         (cortex_a15_neon_mla_qqq_8_16): Likewise.
2548         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
2549         lotype_qdd_64_32_long): Likewise.
2550         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
2551         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
2552         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
2553         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
2554         (cortex_a15_neon_shift_1): Likewise.
2555         (cortex_a15_neon_shift_2): Likewise.
2556         (cortex_a15_neon_shift_3): Likewise.
2557         (cortex_a15_neon_vshl_ddd): Likewise.
2558         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
2559         (cortex_a15_neon_vsra_vrsra): Likewise.
2560         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
2561         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
2562         (cortex_a15_neon_fp_vmul_ddd): Likewise.
2563         (cortex_a15_neon_fp_vmul_qqd): Likewise.
2564         (cortex_a15_neon_fp_vmla_ddd): Likewise.
2565         (cortex_a15_neon_fp_vmla_qqq): Likewise.
2566         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
2567         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
2568         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
2569         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
2570         (cortex_a15_neon_bp_simple): Likewise.
2571         (cortex_a15_neon_bp_2cycle): Likewise.
2572         (cortex_a15_neon_bp_3cycle): Likewise.
2573         (cortex_a15_neon_vld1_1_2_regs): Likewise.
2574         (cortex_a15_neon_vld1_3_4_regs): Likewise.
2575         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
2576         (cortex_a15_neon_vld2_4_regs): Likewise.
2577         (cortex_a15_neon_vld3_vld4): Likewise.
2578         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
2579         (cortex_a15_neon_vst1_3_4_regs): Likewise.
2580         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
2581         (cortex_a15_neon_vst3_vst4): Likewise.
2582         (cortex_a15_neon_vld1_vld2_lane): Likewise.
2583         (cortex_a15_neon_vld3_vld4_lane" 10
2584         (cortex_a15_neon_vst1_vst2_lane): Likewise.
2585         (cortex_a15_neon_vst3_vst4_lane): Likewise.
2586         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
2587         (cortex_a15_neon_ldm_2): Likewise.0
2588         (cortex_a15_neon_stm_2): Likewise.
2589         (cortex_a15_neon_mcr): Likewise.
2590         (cortex_a15_neon_mcr_2_mcrr): Likewise.
2591         (cortex_a15_neon_mrc): Likewise.
2592         (cortex_a15_neon_mrrc): Likewise.
2593         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
2594         change.
2595         (cortex_a15_alu_shift): Likewise.
2596         (cortex_a15_alu_shift_reg): Likewise.
2597         (cortex_a15_mult32): Likewise.
2598         (cortex_a15_mult64): Likewise.
2599         (cortex_a15_block): Likewise.
2600         (cortex_a15_branch): Likewise.
2601         (cortex_a15_load1): Likewise.
2602         (cortex_a15_load3): Likewise.
2603         (cortex_a15_store1): Likewise.
2604         (cortex_a15_store3): Likewise.
2605         (cortex_a15_call): Likewise.
2606         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
2607         (cortex_a5_f2r): Likewise.
2608         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
2609         change.
2610         (cortex_a53_f2r): Likewise.
2611         * config/arm/cortex-a7.md
2612         (cortex_a7_branch): Update for attribute change.
2613         (cortex_a7_call): Likewise.
2614         (cortex_a7_alu_imm): Likewise.
2615         (cortex_a7_alu_reg): Likewise.
2616         (cortex_a7_alu_shift): Likewise.
2617         (cortex_a7_mul): Likewise.
2618         (cortex_a7_load1): Likewise.
2619         (cortex_a7_store1): Likewise.
2620         (cortex_a7_load2): Likewise.
2621         (cortex_a7_store2): Likewise.
2622         (cortex_a7_load3): Likewise.
2623         (cortex_a7_store3): Likewise.
2624         (cortex_a7_load4): Likewise.
2625         (cortex_a7_store4): Likewise.
2626         (cortex_a7_fpalu): Likewise.
2627         (cortex_a7_fconst): Likewise.
2628         (cortex_a7_fpmuls): Likewise.
2629         (cortex_a7_neon_mul): Likewise.
2630         (cortex_a7_fpmacs): Likewise.
2631         (cortex_a7_neon_mla: Likewise.
2632         (cortex_a7_fpmuld: Likewise.
2633         (cortex_a7_fpmacd: Likewise.
2634         (cortex_a7_fpfmad: Likewise.
2635         (cortex_a7_fdivs: Likewise.
2636         (cortex_a7_fdivd: Likewise.
2637         (cortex_a7_r2f: Likewise.
2638         (cortex_a7_f2r: Likewise.
2639         (cortex_a7_f_flags: Likewise.
2640         (cortex_a7_f_loads: Likewise.
2641         (cortex_a7_f_loadd: Likewise.
2642         (cortex_a7_f_stores: Likewise.
2643         (cortex_a7_f_stored: Likewise.
2644         (cortex_a7_neon): Likewise.
2645         * config/arm/cortex-a8-neon.md
2646         (cortex_a8_neon_mrc): Update for attribute change.
2647         (cortex_a8_neon_mrrc): Likewise.
2648         (cortex_a8_neon_int_1): Likewise.
2649         (cortex_a8_neon_int_2): Likewise.
2650         (cortex_a8_neon_int_3): Likewise.
2651         (cortex_a8_neon_int_4): Likewise.
2652         (cortex_a8_neon_int_5): Likewise.
2653         (cortex_a8_neon_vqneg_vqabs): Likewise.
2654         (cortex_a8_neon_vmov): Likewise.
2655         (cortex_a8_neon_vaba): Likewise.
2656         (cortex_a8_neon_vaba_qqq): Likewise.
2657         (cortex_a8_neon_vsma): Likewise.
2658         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2659         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
2660         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
2661         long_scalar): Likewise.
2662         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2663         (cortex_a8_neon_mla_qqq_8_16): Likewise.
2664         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
2665         long_scalar_qdd_64_32_long): Likewise.
2666         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
2667         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
2668         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
2669         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
2670         (cortex_a8_neon_shift_1): Likewise.
2671         (cortex_a8_neon_shift_2): Likewise.
2672         (cortex_a8_neon_shift_3): Likewise.
2673         (cortex_a8_neon_vshl_ddd): Likewise.
2674         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
2675         (cortex_a8_neon_vsra_vrsra): Likewise.
2676         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
2677         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
2678         (cortex_a8_neon_fp_vsum): Likewise.
2679         (cortex_a8_neon_fp_vmul_ddd): Likewise.
2680         (cortex_a8_neon_fp_vmul_qqd): Likewise.
2681         (cortex_a8_neon_fp_vmla_ddd): Likewise.
2682         (cortex_a8_neon_fp_vmla_qqq): Likewise.
2683         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
2684         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
2685         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
2686         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
2687         (cortex_a8_neon_bp_simple): Likewise.
2688         (cortex_a8_neon_bp_2cycle): Likewise.
2689         (cortex_a8_neon_bp_3cycle): Likewise.
2690         (cortex_a8_neon_ldr): Likewise.
2691         (cortex_a8_neon_str): Likewise.
2692         (cortex_a8_neon_vld1_1_2_regs): Likewise.
2693         (cortex_a8_neon_vld1_3_4_regs): Likewise.
2694         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
2695         (cortex_a8_neon_vld2_4_regs): Likewise.
2696         (cortex_a8_neon_vld3_vld4): Likewise.
2697         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
2698         (cortex_a8_neon_vst1_3_4_regs): Likewise.
2699         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
2700         (cortex_a8_neon_vst3_vst4): Likewise.
2701         (cortex_a8_neon_vld1_vld2_lane): Likewise.
2702         (cortex_a8_neon_vld3_vld4_lane): Likewise.
2703         (cortex_a8_neon_vst1_vst2_lane): Likewise.
2704         (cortex_a8_neon_vst3_vst4_lane): Likewise.
2705         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
2706         (cortex_a8_neon_mcr): Likewise.
2707         (cortex_a8_neon_mcr_2_mcrr): Likewise.
2708         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
2709         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
2710         change.
2711         (ca9_neon_mrrc): Likewise.
2712         (cortex_a9_neon_int_1): Likewise.
2713         (cortex_a9_neon_int_2): Likewise.
2714         (cortex_a9_neon_int_3): Likewise.
2715         (cortex_a9_neon_int_4): Likewise.
2716         (cortex_a9_neon_int_5): Likewise.
2717         (cortex_a9_neon_vqneg_vqabs): Likewise.
2718         (cortex_a9_neon_vmov): Likewise.
2719         (cortex_a9_neon_vaba): Likewise.
2720         (cortex_a9_neon_vaba_qqq): Likewise.
2721         (cortex_a9_neon_vsma): Likewise.
2722         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2723         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
2724         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
2725         long_scalar): Likewise.
2726         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2727         (cortex_a9_neon_mla_qqq_8_16): Likewise.
2728         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
2729         long_scalar_qdd_64_32_long): Likewise.
2730         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
2731         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
2732         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
2733         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
2734         (cortex_a9_neon_shift_1): Likewise.
2735         (cortex_a9_neon_shift_2): Likewise.
2736         (cortex_a9_neon_shift_3): Likewise.
2737         (cortex_a9_neon_vshl_ddd): Likewise.
2738         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
2739         (cortex_a9_neon_vsra_vrsra): Likewise.
2740         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
2741         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
2742         (cortex_a9_neon_fp_vsum): Likewise.
2743         (cortex_a9_neon_fp_vmul_ddd): Likewise.
2744         (cortex_a9_neon_fp_vmul_qqd): Likewise.
2745         (cortex_a9_neon_fp_vmla_ddd): Likewise.
2746         (cortex_a9_neon_fp_vmla_qqq): Likewise.
2747         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
2748         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
2749         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
2750         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
2751         (cortex_a9_neon_bp_simple): Likewise.
2752         (cortex_a9_neon_bp_2cycle): Likewise.
2753         (cortex_a9_neon_bp_3cycle): Likewise.
2754         (cortex_a9_neon_ldr): Likewise.
2755         (cortex_a9_neon_str): Likewise.
2756         (cortex_a9_neon_vld1_1_2_regs): Likewise.
2757         (cortex_a9_neon_vld1_3_4_regs): Likewise.
2758         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
2759         (cortex_a9_neon_vld2_4_regs): Likewise.
2760         (cortex_a9_neon_vld3_vld4): Likewise.
2761         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
2762         (cortex_a9_neon_vst1_3_4_regs): Likewise.
2763         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
2764         (cortex_a9_neon_vst3_vst4): Likewise.
2765         (cortex_a9_neon_vld1_vld2_lane): Likewise.
2766         (cortex_a9_neon_vld3_vld4_lane): Likewise.
2767         (cortex_a9_neon_vst1_vst2_lane): Likewise.
2768         (cortex_a9_neon_vst3_vst4_lane): Likewise.
2769         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
2770         (cortex_a9_neon_mcr): Likewise.
2771         (cortex_a9_neon_mcr_2_mcrr): Likewise.
2772         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
2773         (cortex_a9_fps): Likewise.
2774         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
2775         change.
2776         (cortex_m4_fmuls): Likewise.
2777         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
2778         change.
2779         (cortex_r4_mrc): Likewise.
2780         * config/arm/iterators.md: Update comment referring to neon_type.
2781         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
2782         (iwmmxt_movsi_insn): Likewise.
2783         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
2784         attribute change.
2785         (pj4_core_to_vfp): Likewise.
2786         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
2787         attribute change.
2788         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
2789         (vfp_fstore): Likewise.
2790         * doc/md.texi: Change references to neon_type to refer to type.
2792 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
2794         * tree.h (DECL_BUILT_IN): Fix typo in comment.
2796 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
2798         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
2799         lglobl if not weak.
2801 2013-09-04  Easwaran Raman  <eraman@google.com>
2803         PR middle-end/57370
2804         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
2805         (build_and_add_sum): Use it.
2806         (appears_later_in_bb): Simplify code.
2808 2013-09-04  Teresa Johnson  <tejohnson@google.com>
2810         * dumpfile.c (dump_finish): Don't close stderr/stdout.
2812 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
2814         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
2816 2013-09-04  Jan Hubicka  <jh@suse.cz>
2818         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
2819         * ipa-devirt.c: Include diganostic.h
2820         (odr_type_d): Add types and types_set.
2821         (hash_type_name): Work for types with vtables during LTO.
2822         (odr_hasher::remove): Fix comment; destroy types_set.
2823         (add_type_duplicate): New function,
2824         (get_odr_type): Use it.
2825         (dump_type_inheritance_graph): Dump type duplicates.
2826         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
2827         graph.
2828         * tree.c (types_same_for_odr): Give exact answers on types with
2829         virtual tables.
2831 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
2833         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
2834         explaining their differences.
2836 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
2838         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
2840 2013-09-03  Jeff Law  <law@redhat.com>
2842         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
2843         when not threading through a joiner block.  Pass joiner/no joiner
2844         state to register_jump_thread.
2845         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
2846         state from argument rather than implying on path length.
2847         Dump the entire jump thread path into debugging dump.
2848         * tree-flow.h (register_jump_thread): Update prototype.
2850 2013-08-29  Xinliang David Li  <davidxl@google.com>
2852         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
2853         Remove a trivial gcc_assert.
2855 2013-08-29  Xinliang David Li  <davidxl@google.com>
2857         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
2858         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
2859         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
2860         Delay base decl alignment adjustment.
2861         * tree-vectorizer.c (vect_destroy_datarefs): New function.
2862         * tree-vectorizer.h: New data structure.
2863         (set_dr_misalignment): New function.
2864         (dr_misalignment): Ditto.
2865         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
2866         (vectorizable_load): Ditto.
2867         (ensure_base_align): New function.
2868         (vectorize_loops): Add dbg_cnt support.
2869         (execute_vect_slp): Ditto.
2870         * dbgcnt.def: New debug counter.
2871         * Makefile: New dependency.
2873 2013-09-03  Meador Inge  <meadori@codesourcery.com>
2875         Revert:
2877         2013-08-30  Meador Inge  <meadori@codesourcery.com>
2879         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
2881 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
2883         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
2884         function descriptor.
2886 2013-09-03  Richard Biener  <rguenther@suse.de>
2888         * tree-affine.c (add_elt_to_tree): Fix association issue,
2889         avoid useless converts and make sure to always return a
2890         properly typed result.
2892 2013-09-03  Richard Biener  <rguenther@suse.de>
2894         PR middle-end/57656
2895         * fold-const.c (negate_expr_p): Fix division case.
2896         (negate_expr): Likewise.
2898 2013-09-03  Richard Biener  <rguenther@suse.de>
2900         PR lto/58285
2901         * tree-streamer-out.c: Include tm.h.
2902         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
2904 2013-09-03  Jan Hubicka  <jh@suse.cz>
2906         * tree-profile.c (tree_profiling): Cleanup CFG when done.
2908 2013-09-03  Alan Modra  <amodra@gmail.com>
2910         * config.gcc (powerpc*-*-linux*): Add support for little-endian
2911         multilibs to big-endian target and vice versa.
2912         * config/rs6000/t-linux64: Use := assignment on all vars.
2913         (MULTILIB_EXTRA_OPTS): Remove fPIC.
2914         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
2915         * config/rs6000/t-linux64le: New file.
2916         * config/rs6000/t-linux64bele: New file.
2917         * config/rs6000/t-linux64lebe: New file.
2919 2013-09-02  Jan Hubicka  <jh@suse.cz>
2921         * ipa-inline-transform.c (inline_transform): Do not
2922         optimize_inline_calls when not optimizing.
2924 2013-09-02  Jan Hubicka  <jh@suse.cz>
2926         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
2927         duplicated nodes for assembler names.
2928         * symtab.c (symtab_unregister_node): Do not attempt to unlink
2929         hard registers from assembler name hash.
2931 2013-09-02  Jan Hubicka  <jh@suse.cz>
2933         * ipa-split.c (execute_split_functions): Split externally visible
2934         functions called once.
2936 2013-09-02  Martin Jambor  <mjambor@suse.cz>
2938         PR ipa/58106
2939         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
2940         linked list.  When finding the correct duplicate, also consider also
2941         the caller in additon to its inlined_to node.
2943 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
2945         * config/aarch64/aarch64-simd-builtins.def
2946         (dup_lane_scalar): Remove.
2947         * config/aarch64/aarch64-simd.md
2948         (aarch64_simd_dup): Add 'w->w' alternative.
2949         (aarch64_dup_lane<mode>): Allow for VALL.
2950         (aarch64_dup_lane_scalar<mode>): Remove.
2951         (aarch64_dup_lane_<vswap_width_name><mode>): New.
2952         (aarch64_get_lane_signed<mode>): Add w->w altenative.
2953         (aarch64_get_lane_unsigned<mode>): Likewise.
2954         (aarch64_get_lane<mode>): Likewise.
2955         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
2956         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
2957         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
2958         (VCON): Change container of V2SF.
2959         (vswap_width_name): Likewise.
2960         * config/aarch64/arm_neon.h
2961         (__aarch64_vdup_lane_any): New.
2962         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
2963         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
2964         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
2966 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
2968         PR middle-end/56382
2969         * expr.c (emit_move_complex): Do not move complex FP values as parts if
2970         the source or the destination is a single hard register.
2972 2013-09-02  Richard Biener  <rguenther@suse.de>
2974         PR middle-end/57511
2975         * tree-scalar-evolution.c (instantiate_scev_name): Allow
2976         non-linear SCEVs.
2978 2013-09-02  Richard Biener  <rguenther@suse.de>
2980         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
2981         arithmetic to sizetype.
2983 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
2985         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
2986         Find auto-increment use both before and after candidate.
2988 2013-09-02  Marek Polacek  <polacek@redhat.com>
2990         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
2992 2013-09-01  Jan Hubicka  <jh@suse.cz>
2994         * Makefile.in: Add ipa-profile.o
2995         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
2996         * cgraph.c (struct cgraph_propagate_frequency_data,
2997         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
2998         ipa-profile.c; replace cgraph_ by ipa_ prefix.
2999         * cgraph.h (cgraph_propagate_frequency): Remove.
3000         * ipa-inline-analysis.c: Include ipa-utils.h;
3001         drop duplicated cfgloop.h.
3002         (inline_update_callee_summaries): Update.
3003         * ipa-profile.c: New file.
3004         * ipa-utils.h (ipa_propagate_frequency): Declare.
3005         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
3006         data-streamer.h, value-prof.h.
3007         (symtab_remove_unreachable_nodes): Update profile.
3008         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
3009         account_time_size, cmp_counts, dump_histogram,
3010         ipa_profile_generate_summary, ipa_profile_write_summary,
3011         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
3012         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
3013         Move to ipa-profile.c.
3015 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
3017         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
3019 2013-09-01  Jan Hubicka  <jh@suse.cz>
3021         * common.opt (fdevirtualize-speculatively): New function.
3022         * invoke.texi (fdevirtualize-speculatively): Document.
3023         * ipa-devirt.c: Include ipa-inline.h
3024         (likely_target_p): New function.
3025         (ipa_devirt): New function.
3026         (gate_ipa_devirt): New function.
3027         (pass_data_ipa_devirt): New static var.
3028         (pass_ipa_devirt): Likewise.
3029         (make_pass_ipa_devirt): New function.
3030         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
3031         (common_handle_option): Disable devirtualization when
3032         value range profiling is available.
3033         * passes.def (pass_ipa_devirt): Add.
3034         * timever.def (TV_IPA_DEVIRT): New timevar.
3035         * tree-pass.h (make_pass_ipa_devirt):
3037 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
3039         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
3040         include sanitize(undefined).
3042 2013-08-31  Diego Novillo  <dnovillo@google.com>
3044         * Makefile.in (TREE_CORE_H): Define.
3045         (TREE_H): Use.
3046         (GTFILES): Add tree-core.h.
3047         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
3048         size the array.
3049         * tree-core.h: New file.
3050         Move all data structures, enum, typedefs, global
3051         declarations and constants from ...
3052         * tree.h: ... here.
3054 2013-08-31  Jan Hubicka  <jh@suse.cz>
3056         * bulitins.c (expand_builtin): Do not early exit for gcov
3057         instrumented functions.
3059 2013-08-31  Marek Polacek  <polacek@redhat.com>
3061         * ubsan.c: Include tm_p.h.
3063 2013-08-31  Jan Hubicka  <jh@suse.cz>
3065         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
3066         warning.
3068         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
3069         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
3070         * tree-cfg.c (verify_gimple_label): ... here.
3071         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
3072         (ipa_merge_profiles): New function.
3073         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
3074         (lto_input_function_body): Likewise.
3075         * ipa-utils.h (ipa_merge_profiles): Declare.
3076         * lto-streamer.h (lto_input_function_body): Update prototype.
3077         (emit_label_in_global_context_p): Remove.
3078         * lto-symtab.c: Include ipa-utils.h
3079         (lto_cgraph_replace_node): Use ipa_merge_profiles.
3081 2013-08-31  Jan Hubicka  <jh@suse.cz>
3083         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
3085 2013-08-31  Jan Hubicka  <jh@suse.cz>
3087         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
3089 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
3091         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
3092         "cmp" RTX before signed_comparison_operator check to account
3093         for "code" changes.
3095 2013-08-30  Jan Hubicka  <jh@suse.cz>
3097         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
3098         (detect_type_change_1): Rename to ...
3099         (detect_type_change): ... this one; early return on non-polymorphic
3100         types.
3101         (detect_type_change_ssa): Add comp_type parameter; update
3102         use of detect_type_change.
3103         (compute_complex_assign_jump_func): Add param_type parameter;
3104         update use of detect_type_change_ssa.
3105         (compute_complex_ancestor_jump_func): Likewise.
3106         (ipa_get_callee_param_type): New function.
3107         (ipa_compute_jump_functions_for_edge): Compute parameter type;
3108         update calls to the jump function computation functions.
3110 2013-08-30  Teresa Johnson  <tejohnson@google.com>
3111             Steven Bosscher  <steven@gcc.gnu.org>
3113         * cfgrtl.c (fixup_new_cold_bb): New routine.
3114         (commit_edge_insertions): Invoke fixup_partitions.
3115         (find_partition_fixes): New routine.
3116         (fixup_partitions): Ditto.
3117         (verify_hot_cold_block_grouping): Update comments.
3118         (rtl_verify_edges): Invoke find_partition_fixes.
3119         (rtl_verify_bb_pointers): Update comments.
3120         (rtl_verify_bb_layout): Ditto.
3121         * basic-block.h (probably_never_executed_edge_p): Declare.
3122         (fixup_partitions): Ditto.
3123         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
3124         * bb-reorder.c (sanitize_hot_paths): New function.
3125         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
3126         sanitize_hot_paths.
3127         * predict.c (probably_never_executed_edge_p): New routine.
3128         * cfg.c (check_bb_profile): Add partition insanity warnings.
3130 2013-08-30  Meador Inge  <meadori@codesourcery.com>
3132         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
3134 2013-08-30  Marek Polacek  <polacek@redhat.com>
3136         * Makefile.in (ubsan.o): Add.
3137         (c-family/c-ubsan.o): Add.
3138         (builtins.o): Add ubsan.h dependency.
3139         * ubsan.h: New file.
3140         * ubsan.c: New file.
3141         * common.opt: Add -fsanitize=undefined option.
3142         (flag_sanitize): Add variable.
3143         (fsanitize=): Add option.  Add Driver.
3144         (fsanitize=thread): Remove option.
3145         (fsanitize=address): Likewise.
3146         (static-libubsan): New option.
3147         * doc/invoke.texi: Document the new flag and -static-libubsan.
3148         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
3149         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
3150         * builtin-attrs.def (ATTR_COLD): Define.
3151         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
3152         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
3153         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
3154         * flag-types.h (sanitize_code): New enum.
3155         * opts.c (common_handle_option): Parse command line arguments
3156         of -fsanitize=.  Add -fsanitize=unreachable option.
3157         * varasm.c (get_variable_section): Adjust.
3158         (assemble_noswitch_variable): Likewise.
3159         (assemble_variable): Likewise.
3160         (output_constant_def_contents): Likewise.
3161         (categorize_decl_for_section): Likewise.
3162         (place_block_symbol): Likewise.
3163         (output_object_block): Likewise.
3164         * builtins.def: Likewise.
3165         * toplev.c (compile_file): Likewise.
3166         (process_options): Likewise.
3167         * cppbuiltin.c: Likewise.
3168         * tsan.c (tsan_pass): Likewise.
3169         (tsan_gate): Likewise.
3170         (tsan_gate_O0): Likewise.
3171         * cfgexpand.c (partition_stack_vars): Likewise.
3172         (expand_stack_vars): Likewise.
3173         (defer_stack_allocation): Likewise.
3174         (expand_used_vars): Likewise.
3175         * cfgcleanup.c (old_insns_match_p): Likewise.
3176         * asan.c (asan_finish_file): Likewise.
3177         (asan_instrument): Likewise.
3178         (gate_asan): Likewise.
3179         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
3180         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
3181         (asan_global_struct): Use pointer_sized_int_node instead
3182         calling build_nonstandard_integer_type.
3183         (initialize_sanitizer_builtins): Likewise.
3184         (asan_finish_file): Likewise.
3185         * gcc.c: Document %{%:function(args):X}.
3186         (static_spec_functions): Add sanitize.
3187         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
3188         store funcval != NULL there.
3189         (do_spec_1): Adjust handle_spec_function caller.
3190         (handle_braces): Allow %:function(args) as condition.
3191         (sanitize_spec_function): New function.
3192         (ADD_STATIC_LIBUBSAN_LIBS): Define.
3193         (LIBUBSAN_SPEC): Likewise.
3194         (LIBUBSAN_EARLY_SPEC): Likewise.
3195         (SANITIZER_SPEC): Handle libubsan.
3196         (SANITIZER_EARLY_SPEC): Likewise.
3197         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
3198         instead of fsanitize=address.
3199         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
3200         instead of fsanitize=address*.
3201         * builtins.c: Include ubsan.h.
3202         (fold_builtin_0): Instrument __builtin_unreachable.
3203         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
3204         instead of flag_asan.
3205         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
3206         (pointer_sized_int_node): Define.
3207         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
3209 2013-08-30  Mike Stump  <mikestump@comcast.net>
3211         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
3212         with RE patterns.
3214 2013-08-29  Jan Hubicka  <jh@suse.cz>
3216         * cgraph.c (cgraph_function_body_availability): Handle weakref
3217         correctly.
3218         * passes.def: Remove pass_fixup_cfg.
3219         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
3220         track when we need to remove functions.
3221         (gate_ipa_inline): Execute inlining always; add comment why.
3222         (pass_data_ipa_inline): Remove TODO_remove_functions.
3223         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
3224         do not produce summaries.
3225         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
3226         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
3227         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
3228         constant pool and vtable.
3230 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
3232         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
3233         New arm_neon.h's internal macros to specify 64-bit constants.
3234         Avoid using stdint.h's macros.
3236 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
3238         * recog.c (verify_changes): Verify that changes[i].old is non-zero
3239         before applying REG_P.
3241 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
3243         PR tree-optimization/58277
3244         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
3245         after seeing too many stmts with vdef in between dombb and current
3246         bb, invalidate everything.
3248 2013-08-30  Richard Biener  <rguenther@suse.de>
3250         * fold-const.c (fold_single_bit_test): Fix overflow test.
3252 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
3254         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
3255         and which can live in a register, always retrieve the value on entry.
3256         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
3257         passed by invisible reference specially.
3258         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
3259         (vt_add_function_parameter): Correctly deal with a parameter passed by
3260         invisible reference.
3262 2013-08-30  Jan Hubicka  <jh@suse.cz>
3264         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
3266 2013-08-30  Richard Biener  <rguenther@suse.de>
3268         PR tree-optimization/58228
3269         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
3270         allow invariant loads in nested loop vectorization.
3272 2013-08-30  Richard Biener  <rguenther@suse.de>
3274         PR tree-optimization/58223
3275         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
3276         (has_anti_or_output_dependence): ... this and adjust to also
3277         look for output dependences.
3278         (mark_nodes_having_upstream_mem_writes): Adjust.
3279         (rdg_flag_uses): Likewise.
3281 2013-08-30  Richard Biener  <rguenther@suse.de>
3283         PR tree-optimization/58010
3284         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
3285         assert that we have a loop-closed PHI.
3287 2013-08-29  Jan Hubicka  <jh@suse.cz>
3289         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
3290         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
3291         * lto-section-in.c (lto_free_function_in_decl_state): New function.
3292         (lto_free_function_in_decl_state_for_node): New function.
3294 2013-08-29  Xinliang David Li  <davidxl@google.com>
3296         * loop-unroll.c (report_unroll_peel): Minor message change.
3297         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
3298         Emit alignment peeling message with default -fopt-info.
3299         (vect_loop_versioning): Emit loop version info message.
3300         * tree-vectorizer.c (vectorize_loops): Minor message change.
3301         (execute_vect_slp): Ditto.
3303 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
3305         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
3306         of the clone from the DECL_NAME of the original function.
3308 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
3310         * passes.c (register_pass): Add overload.
3311         * tree-pass.h (register_pass): Forward declare it.  Add comment.
3313 2013-08-29  Jan Hubicka  <jh@suse.cz>
3315         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
3316         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
3317         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
3318         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
3319         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
3320         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
3321         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
3322         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
3324 2013-08-29  Teresa Johnson  <tejohnson@google.com>
3326         * dumpfile.c (dump_loc): Output column number.
3327         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
3328         * doc/invoke.texi: Document optall -fopt-info flag.
3329         * profile.c (read_profile_edge_counts): Use new dump framework.
3330         (compute_branch_probabilities): Ditto.
3331         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
3332         when pass not in any opt group.
3333         * pass_manager.h (pass_manager::get_pass_profile): New method.
3334         * value-prof.c (check_counter): Use new dump framework.
3335         (check_ic_target): Ditto.
3336         * coverage.c (get_coverage_counts): Ditto.
3337         (coverage_init): Setup new dump framework.
3339 2013-08-29  Richard Biener  <rguenther@suse.de>
3341         PR tree-optimization/58246
3342         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
3343         handle the dominance check inside a basic-block.
3345 2013-08-29  Richard Biener  <rguenther@suse.de>
3347         PR middle-end/57287
3348         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
3349         of default defs that appear in abnormal PHI nodes.
3351 2013-08-29  Richard Biener  <rguenther@suse.de>
3353         PR tree-optimization/57685
3354         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
3355         single-use operands to avoid exponential complexity.
3357 2013-08-28  Dehao Chen  <dehao@google.com>
3359         * ipa-inline.c (edge_badness): Fix integer underflow.
3361 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
3363         * gtm-builtins.def (_ITM_free): Declare leaf.
3365 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
3367         PR target/58067
3368         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
3369         (*tls_local_dynamic_base_64_largepic): Likewise.
3370         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
3371         Remove predicate from call operand.
3372         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
3373         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
3375 2013-08-28  Jeff Law  <law@redhat.com>
3377         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
3378         checks for the number of predecessors and successors allowed.
3379         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
3380         which require copying a joiner block if there is a request which
3381         is a subpath that requires no joiner block copying.
3383 2013-08-28  Jan Hubicka  <jh@suse.cz>
3385         * lto-streamer-out.c (DFS_write_tree_body): Drop
3386         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
3387         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
3388         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
3389         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
3390         Do not read DECL_ERROR_ISSUED.
3391         (unpack_ts_decl_with_vis_value_fields): Do not read
3392         DECL_DEFER_OUTPUT.
3393         (lto_input_ts_binfo_tree_pointers): Do not read
3394         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
3395         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
3396         write DECL_ERROR_ISSUED..
3397         (pack_ts_decl_with_vis_value_fields): Do not write
3398         DECL_DEFER_OUTPUT.
3399         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
3400         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
3401         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
3402         * tree.h (tree_decl_common): Update comment.
3403         (DECL_ERROR_ISSUED): Remove.
3405 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
3407         PR middle-end/58257
3408         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
3410 2013-08-28  Jan Hubicka  <jh@suse.cz>
3412         * builtins.def (free): Declare leaf.
3414 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
3416         * gdbhooks.py: New.
3417         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
3418         * configure: Regenerate.
3420 2013-08-27  Martin Jambor  <mjambor@suse.cz>
3422         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
3423         (ipa_ancestor_jf_data): Likewise.
3424         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
3425         (ipa_get_jf_pass_through_type_preserved): New function.
3426         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
3427         (ipa_get_jf_ancestor_type_preserved): New function.
3428         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
3429         (ipa_get_jf_ancestor_result): Likewise.
3430         (propagate_vals_accross_pass_through): Use
3431         ipa_get_jf_pass_through_result to do all the value mappings.
3432         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
3433         type_preserved flag.
3434         (ipa_set_jf_cst_copy): New function.
3435         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
3436         (ipa_set_jf_arith_pass_through): Likewise.
3437         (ipa_set_ancestor_jf): Likewise.
3438         (compute_complex_assign_jump_func): Set type_preserved instead of
3439         punting.
3440         (ipa_compute_jump_functions_for_edge): Likewise.
3441         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
3442         (update_jump_functions_after_inlining): Update type_preserved.
3443         Explicitely create jump functions when combining one with pass_through.
3444         (ipa_write_jump_function): Stream the type_preserved flags.
3445         (ipa_read_jump_function): Likewise.
3447 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
3448             Aldy Hernandez  <aldyh@redhat.com>
3450         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
3451         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
3452         * function.h (struct function): Add has_force_vect_loops and
3453         has_simduid_loops.
3454         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
3455         * gimple.c (gimple_build_omp_critical): Add KIND argument and
3456         handle it.
3457         * gimple.def: Update CLAUSES comments.
3458         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
3459         (gimple_build_omp_for): Add argument to prototype.
3460         (gimple_omp_for_kind): New.
3461         (gimple_omp_for_set_kind): New.
3462         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
3463         GOVD_DATA_SHARE_CLASS.
3464         (enum omp_region_type): Add ORT_SIMD.
3465         (gimple_add_tmp_var): Handle ORT_SIMD.
3466         (gimplify_var_or_parm_decl): Same.
3467         (is_gimple_stmt): Same.
3468         (omp_firstprivatize_variable): Same.
3469         (omp_add_variable): Only use splay_tree_insert if lookup failed.
3470         (omp_notice_variable): Handle ORT_SIMD.
3471         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
3472         (omp_check_private): Handle ORT_SIMD.
3473         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
3474         OMP_CLAUSE_SAFELEN.
3475         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
3476         Handle OMP_CLAUSE_LASTPRIVATE.
3477         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
3478         OMP_CLAUSE_SAFELEN.
3479         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
3480         (gimplify_expr): Handle OMP_SIMD.
3481         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
3482         (expand_GOMP_SIMD_VF): New.
3483         (expand_GOMP_SIMD_LAST_LANE): New.
3484         * internal-fn.def (GOMP_SIMD_LANE): New.
3485         (GOMP_SIMD_VF): New.
3486         (GOMP_SIMD_LAST_LANE): New.
3487         * omp-low.c: Include target.h.
3488         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
3489         OMP_CLAUSE_SAFELEN.
3490         (check_omp_nesting_restrictions): Same.
3491         (omp_max_vf): New.
3492         (lower_rec_simd_input_clauses): New.
3493         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
3494         OMP_CLAUSE_LINEAR.
3495         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
3496         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
3497         (expand_omp_build_assign): New.
3498         (expand_omp_for_init_counts): New.
3499         (expand_omp_for_init_vars): New.
3500         (extract_omp_for_update_vars): New.
3501         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
3502         and rewrite accordingly.
3503         (expand_omp_simd): New.
3504         (expand_omp_for): Use expand_omp_simd.
3505         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
3506         (lower_omp_for): Do not lower the body.
3507         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
3508         in their own loops.
3509         * tree-flow.h (find_omp_clause): Remove prototype.
3510         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
3511         forcing vectorization of the loop, or if flag_tree_vectorize.
3512         (gate_tree_if_conversion): Similarly.
3513         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
3514         gimple_build_omp_for.
3515         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
3516         * tree-parloops (create_parallel_loop): Pass kind argument to
3517         gimple_build_omp_for.
3518         * tree-pretty-print.c (dump_omp_clause): Add cases for
3519         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
3520         OMP_CLAUSE__SIMDUID_.
3521         (dump_generic_node): Handle OMP_SIMD.
3522         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
3523         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
3524         unroll OMP_SIMD loops here.
3525         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
3526         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
3527         loop->safelen.
3528         (vect_analyze_data_refs): Handle simd loops.
3529         * tree-vect-loop.c (vectorizable_live_operation): Handle
3530         IFN_GOMP_SIMD*.
3531         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
3532         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
3533         (vectorizable_load): Same.
3534         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
3535         (struct simduid_to_vf): New.
3536         (simduid_to_vf::hash): New.
3537         (simduid_to-vf::equal): New.
3538         (struct simd_array_to_simduid): New.
3539         (simd_array_to_simduid::hash): New.
3540         (simd_array_to_simduid::equal): New.
3541         (adjust_simduid_builtins): New.
3542         (struct note_simd_array_uses_struct): New.
3543         (note_simd_array_uses_cb): New.
3544         (note_simd_array_uses): New.
3545         (vectorize_loops): Handle simd hints and adjust simd builtins
3546         accordingly.
3547         * tree-vectorizer.h (struct _stmt_vec_info): Add
3548         simd_lane_access_p field.
3549         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
3550         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
3551         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
3552         (omp_clause_code_name): Same.
3553         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
3554         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
3555         * tree.def (OMP_SIMD): New entry.
3556         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
3557         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
3558         (OMP_CLAUSE_DECL): Adjust range for new clauses.
3559         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
3560         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
3561         (OMP_CLAUSE_LINEAR_STEP): New.
3562         (OMP_CLAUSE_SAFELEN_EXPR): New.
3563         (OMP_CLAUSE__SIMDUID__DECL): New.
3564         (find_omp_clause): New prototype.
3566 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
3568         * config/i386/driver-i386.c (host_detect_local_cpu): Update
3569         Haswell processor detection.
3571 2013-08-27  Christian Widmer  <shadow@umbrox.de>
3573         PR target/57927
3574         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
3575         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
3576         AVX2 capable processors.
3578 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
3580         * config/aarch64/arm_neon.h: Replace all inline asm implementations
3581         of vget_low_* with implementations in terms of other intrinsics.
3583 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
3585         PR middle-end/57219
3586         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
3587         values to -1, 0 and 1.
3589 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
3591         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
3592         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
3593         (<optab><mode>3_insn): Remove.
3594         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
3595         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
3596         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
3597         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
3598         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
3599         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
3600         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
3601         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
3602         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
3603         (ror<mode>3_insn): Likewise.
3604         * config/aarch64/predicates.md (aarch64_simd_register): New.
3606 2013-08-27  Richard Biener  <rguenther@suse.de>
3608         PR tree-optimization/57521
3609         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
3610         one edge is non-critical.
3611         (find_phi_replacement_condition): Make sure to use a non-critical
3612         edge.  Cleanup and remove old bug workarounds.
3613         (bb_postdominates_preds): Remove.
3614         (if_convertible_loop_p_1): Do not compute post-dominators.
3615         (combine_blocks): Do not free post-dominators.
3616         (main_tree_if_conversion): Likewise.
3617         (pass_data_if_conversion): Add TODO_verify_ssa.
3619 2013-08-27  DJ Delorie  <dj@redhat.com>
3621         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
3623 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
3625         * function.c (assign_parm_find_data_types): Set passed_mode and
3626         nominal_mode to the TYPE_MODE of nominal_type for the built
3627         pointer type in case of the struct-pass-by-reference.
3629 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
3631         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
3632         (UINT16_TYPE): Change default to "unsigned int".
3634         * config/avr/avr.opt (mfract-convert-truncate): New option.
3635         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
3636         is set, round negative fractional integers according to n1169
3637         when converting to integer types.
3639 2013-08-26  Jan Hubicka  <jh@suse.cz>
3641         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
3642         methods can not be called indirectly when their address is not taken.
3644 2013-08-26  Jan Hubicka  <jh@suse.cz>
3646         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
3647         ctor_for_folding.
3649 2013-08-26  Jan Hubicka  <jh@suse.cz>
3651         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
3652         can be unshared.
3654 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
3656         * reload.c (find_valid_class): Allow classes that do not include
3657         FIRST_PSEUDO_REGISTER - 1.
3659 2013-08-26  Jan Hubicka  <jh@suse.cz>
3661         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
3662         fix edge count/frequency when speculation failed; fix type check
3663         for the direct call.
3665 2013-08-26  Jan Hubicka  <jh@suse.cz>
3667         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
3669 2013-08-26  Jan Hubicka  <jh@suse.cz>
3671         * ipa-inline-transform.c (inline_transform): Be ready for basic block
3672         to be changed by edge redirection.
3674 2013-08-26  Jan Hubicka  <jh@suse.cz>
3676         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
3677         formating; add sanity check.
3678         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
3679         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
3680         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
3681         (resolve_noninline_speculation): Update callee keys, too.
3683 2013-08-26  Jan Hubicka  <jh@suse.cz>
3685         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
3686         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
3688 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
3690         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
3691         into proper place.
3693 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
3695         * config/i386/i386.c (ix86_debug_options): Remove prototype.
3696         (x86_64_elf_select_section): Ditto.
3697         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
3698         arguments.
3699         (ix86_pass_by_reference): Ditto.
3700         (output_set_got): Ditto.
3701         (ix86_unary_operator_ok): Ditto.
3702         (ix86_expand_builtin): Ditto.
3704 2013-08-23  Jan Hubicka  <jh@suse.cz>
3706         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
3708 2013-08-23  Jan Hubicka  <jh@suse.cz>
3710         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
3711         (tree_decl_with_vis): Add FINAL field.
3713 2013-08-23  Jeff Law  <law@redhat.com>
3715         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
3716         the debugging dump when the expression is fully redundant.
3718 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3720         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
3721         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
3722         * pretty-print.c (pp_formatted_text_data): Likewise.
3723         (pp_write_text_to_stream): Likewise.
3724         (pp_write_text_as_dot_label_to_stream): Likewise.
3725         (pp_append_r): Likewise.
3726         (pp_format): Likewise.
3727         (pp_flush): Likewise.
3728         (pp_clear_output_area): Likewise.
3729         (pp_append_text): Likewise.
3730         (pp_formatted_text): Likewise.
3731         (pp_remaining_character_count_for_line): Likewise.
3732         (pp_newline): Likewise.
3733         (pp_character): Likewise.
3734         (output_buffer::~output_buffer): Define.
3735         (pretty_printer::~pretty_printer): Destruct output buffer.
3736         * pretty-print.h (output_buffer::~output_buffer): Declare.
3737         (pretty_printer::~pretty_printer): Declare virtual.
3739 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
3741         PR other/57324
3742         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
3743         HOST_WIDE_INT_M1U): New macros.
3744         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
3745         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
3746         unsigned -1 for lshift.
3747         * cse.c (cse_insn): Likewise.
3748         * double-int.c (rshift_double, lshift_double): Likewise.
3749         * builtins.c (fold_builtin_bitop): Likewise.
3750         * combine.c (force_to_mode): Likewise.
3751         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
3752         * simplify-rtx.c (simplify_const_unary_operation,
3753         simplify_const_binary_operation): Likewise.
3754         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
3755         check_va_list_escapes): Likewise.
3756         * rtlanal.c (nonzero_bits1): Likewise.
3757         * expmed.c (expand_smod_pow2): Likewise.
3758         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
3760 2013-08-23  Jan Hubicka  <jh@suse.cz>
3762         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
3763         as having address taken.
3765 2013-08-23  Jan Hubicka  <jh@suse.cz>
3767         * ipa-utils.h (method_class_type): Declare.
3768         * ipa-devirt.c (method_class_type): Export.
3770         * cgraphunit.c (analyze_functions): Do basic devirtualization;
3771         do not walk base classes of anonymous types.
3773 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
3775         PR rtl-optimization/58220
3776         PR regression/58221
3777         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
3778         handle SEQUENCE insns properly.
3780 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3782         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
3783         definition.
3784         (pp_newline_and_indent): Likewise.
3785         (pp_separate_with): Likewise.
3786         * pretty-print.c (pp_newline_and_flush): Define.
3787         (pp_newline_and_indent): Likewise.
3788         (pp_separate_with): Likewise.
3790 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
3792         PR target/58218
3793         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
3794         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
3796 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
3798         * config/i386/predicates.md (ext_sse_reg_operand): New.
3799         * config/i386/i386.md (*movti_internal): Use
3800         predicate to determine if EVEX is needed.
3801         (*movsi_internal): Ditto.
3802         (*movdf_internal): Ditto.
3803         (*movsf_internal): Ditto.
3804         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
3806 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
3808         PR tree-optimization/58209
3809         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
3810         (find_tail_calls): Give up for pointer result types if m is non-NULL.
3811         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
3812         emit POINTER_PLUS_EXPR.
3813         (create_tailcall_accumulator): For pointer result type accumulate in
3814         sizetype type.
3816 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
3818         * configure.ac: Add backslashes missing from the last change.
3819         * configure: Regenerate.
3821 2013-08-22  Jan Hubicka  <jh@susue.cz>
3823         * ipa.c (function_and_variable_visibility): First remember function
3824         was global and then make it local.
3826 2013-08-22  Julian Brown  <julian@codesourcery.com>
3828         * configure.ac: Add aarch64 to list of arches which use "nop" in
3829         debug_line test.
3830         * configure: Regenerate.
3832 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3834         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
3835         gnu_libc_has_function.
3836         * config/s390/tpf.h: Likewise.
3838 2013-08-22  Jan Hubicka  <jh@susue.cz>
3840         * timevar.c (validate_phases): Add cast.
3842 2013-08-22  Jan Hubicka  <jh@susue.cz>
3844         * timevar.c (validate_phases): Use size_t for memory.
3845         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
3847 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3849         * pretty-print.h (output_buffer::output_buffer): Declare.
3850         (pretty_printer::pretty_printer): Likewise.
3851         (pp_construct): Remove.
3852         * pretty-print.c (output_buffer::output_buffer): Define.
3853         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
3854         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
3855         (print_gimple_expr): Likewise.
3856         (print_gimple_seq): Likewise.
3857         (gimple_dump_bb): Likewise.
3858         * sched-vis.c (dump_value_slim): Likewise.
3859         (dump_insn_slim): Likewise.
3860         (dump_rtl_slim): Likewise.
3861         (str_pattern_slim): Likewise.
3862         * tree-mudflap.c (mf_varname_tree): Likewise.
3863         * graph.c (print_graph_cfg): Likewise.
3864         (start_graph_dump): Likewise.
3865         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
3866         placement-new.
3867         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
3868         pretty printer initialization.
3869         * coretypes.h (diagnostic_context): Remove superflous type alias
3870         declaration.
3871         (pretty_printer): Likewise.  Declare directly as a class.
3872         (pretty_print_info): Remove declaration as class.
3873         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
3874         and pp_clear_output_area.
3875         (asan_add_global): Likewise.
3877 2013-08-22  Jan Hubicka  <jh@suse.cz>
3879         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
3880         * ipa-utils.h (update_type_inheritance_graph): Declare.
3881         (possible_polymorphic_call_target_p): Declare.
3882         (possible_polymorphic_call_target_p): New.
3883         * ipa-devirt.c: Update toplevel comments.
3884         (cached_polymorphic_call_targets): Move up.
3885         (odr_type_d): Move ID down.
3886         (polymorphic_type_binfo_p): Update comment.
3887         (odr_hasher::remove): Likewise;
3888         (get_odr_type): Set anonymous_namespace.
3889         (dump_odr_type): Dump it.
3890         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
3891         (maybe_record_node): Record node in cached_polymorphic_call_targets.
3892         (record_binfo): Add comment.
3893         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
3894         built.
3895         (devirt_node_removal_hook): Do not iCE when cache is freed.
3896         (possible_polymorphic_call_target_p): New predicate.
3897         (update_type_inheritance_graph): New function.
3899 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3900             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3901             Sergey Lega  <sergey.s.lega@intel.com>
3902             Anna Tikhonova  <anna.tikhonova@intel.com>
3903             Ilya Tocar  <ilya.tocar@intel.com>
3904             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3905             Ilya Verbin  <ilya.verbin@intel.com>
3906             Kirill Yukhin  <kirill.yukhin@intel.com>
3907             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3909         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
3910         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
3911         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
3912         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
3913         (OPTION_MASK_ISA_AVX2_UNSET): Update.
3914         (OPTION_MASK_ISA_AVX512F_UNSET): New.
3915         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
3916         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
3917         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
3918         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
3919         OPT_mavx512pf, OPT_mavx512er cases.
3920         * config/i386/constraints.md (v): New constraint.
3921         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
3922         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
3923         (bit_AVX512CD): New.
3924         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
3925         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
3926         * config/i386/i386-c.c (ix86_target_macros_internal):
3927         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
3928         __AVX512PF__.
3929         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
3930         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
3931         * config/i386/i386.c (regclass_map, dbx_register_map)
3932         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
3933         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
3934         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
3935         -mavx512pf options.
3936         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
3937         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
3938         -mavx512pf options.  Fix formatting.
3939         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
3940         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
3941         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
3942         -mavx512cd, -mavx512pf options.
3943         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
3944         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
3945         (ix86_preferred_output_reload_class): Replace SSE_REGS with
3946         ALL_SSE_REGS.
3947         (ix86_hard_regno_mode_ok): Support 512-bit registers.
3948         (ix86_set_reg_reg_cost): Ditto.
3949         (x86_order_regs_for_local_alloc): Ditto.
3950         (MAX_VECT_LEN): Extend to 64-byte.
3951         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
3952         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
3953         (TARGET_AVX512ER, TARGET_AVX512CD): New.
3954         (BIGGEST_ALIGNMENT): Extend to 512-bits.
3955         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
3956         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
3957         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
3958         (SSE_REG_MODE_P): Support new modes.
3959         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
3960         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
3961         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
3962         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
3963         (REG_CLASS_CONTENTS): Add new registers.
3964         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
3965         (EXT_REX_SSE_REGNO_P): New.
3966         (HI_REGISTER_NAMES): Add new registers.
3967         * config/i386/i386.md: Define constants for new registers.
3968         (mode): Add new 512-bit modes.
3969         (prefix): Support evex prefix.
3970         (isa): Support avx512f, noavx512f, fma_avx512f.
3971         (ssemodesuffix): Add new 512-bit modes.
3972         (movxi): New.
3973         (*movxi_internal_avx512f): Ditto.
3974         (*movdi_internal): Replace constraint "x" with the new constraint "v".
3975         Support MODE_XI.
3976         (*movsi_internal): Likewise.
3977         (*movdf_internal): Likewise.
3978         (*movsf_internal): Likewise.
3979         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
3980         (<code><mode>3): Likewise.
3981         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
3982         New.
3983         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
3984         with the new constraint "v".
3985         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
3986         modes.
3987         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
3988         with the new constraint "v".
3989         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
3990         (<sse2>_storedqu<avxsizesuffix>): Likewise.
3991         (*<plusminus_insn><mode>3): Likewise.
3992         (<sse>_vm<plusminus_insn><mode>3): Likewise.
3993         (*mul<mode>3): Likewise.
3994         (<sse>_vmmul<mode>3): Likewise.
3995         (<sse>_div<mode>3): Likewise.
3996         (<sse>_vmdiv<mode>3): Likewise.
3997         (<sse>_sqrt<mode>2): Likewise.
3998         (<sse>_vmsqrt<mode>2): Likewise.
3999         (*<code><mode>3_finite): Likewise.
4000         (*<code><mode>3) <smaxmin>: Likewise.
4001         (<sse>_vm<code><mode>3): Likewise.
4002         (*<code><mode>3) <any_logic>: Likewise.
4003         (*fma_fmadd_<mode>): Likewise.
4004         (*fma_fmsub_<mode>): Likewise.
4005         (*fma_fnmadd_<mode>): Likewise.
4006         (*fma_fnmsub_<mode>): Likewise.
4007         (*fma_fmaddsub_<mode>): Likewise.
4008         (*fma_fmsubadd_<mode>): Likewise.
4009         (*fmai_fmadd_<mode>): Likewise.
4010         (*fmai_fmsub_<mode>): Likewise.
4011         (*fmai_fnmadd_<mode>): Likewise.
4012         (*fmai_fnmsub_<mode>): Likewise.
4013         (sse_cvtsi2ss): Likewise.
4014         (sse_cvtsi2ssq): Likewise.
4015         (sse_cvtss2si): Likewise.
4016         (sse_cvtss2si_2): Likewise.
4017         (sse_cvtss2siq): Likewise.
4018         (sse_cvtss2siq_2): Likewise.
4019         (sse_cvttss2si): Likewise.
4020         (sse_cvtss2siq_2): Likewise.
4021         (float<sseintvecmodelower><mode>2): Likewise.
4022         (sse2_cvtsd2si_2): Likewise.
4023         (sse2_cvtsd2siq_2): Likewise.
4024         (*<plusminus_insn><mode>3): Likewise.
4025         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
4026         (*<sse4_1_avx2>_mul<mode>3): Likewise.
4027         (ashr<mode>3): Likewise.
4028         (<shift_insn><mode>3): Likewise.
4029         (avx2_<code><mode>3): Likewise.
4030         (*avx2_<code><mode>3): Likewise.
4031         (*andnot<mode>3): Likewise.
4032         (*<code><mode>3) <any_logic>: Likewise.
4033         (abs<mode>2): Likewise.
4034         (avx2_permvar<mode>): Likewise.
4035         (avx2_perm<mode>_1): Likewise.
4036         (*avx_vpermilp<mode>): Likewise.
4037         (avx_vpermilvar<mode>3): Likewise.
4038         (avx2_ashrv<mode>): Likewise.
4039         (avx2_<shift_insn>v<mode>): Likewise.
4040         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
4041         -mavx512cd.
4042         * doc/rtl.texi: Document XImode.
4044 2013-08-21  Jeff Law  <law@redhat.com>
4046         * tree-flow.h (register_jump_thread): Pass vector of edges
4047         instead of each important edge.
4048         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
4049         thread path into a vector and pass that to register_jump_thread.
4050         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
4051         passed in edge vector to the current 3-edge form.
4053         Revert:
4054         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
4056         * dce.c (fini_dce): Call df_analyze again just in case
4057         delete_unmarked_insns removed anything.
4059 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
4061         * reload.h (struct reg_equivs): Rename to ..
4062         (struct reg_equivs_s): .. this.
4064 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
4066         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
4068 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
4070         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
4072 2013-08-21  Jeff Law  <law@redhat.com>
4074         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
4075         simplify assignments too.  If the RHS collapses to a singleton
4076         range, then return the value for the range.
4078 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
4080         * config/i386/sse.md (V16): Rename to...
4081         (VMOVE): this.
4082         (mov<mode>): Update iterator name.
4083         (*mov<mode>_internal): Ditto.
4084         (push<mode>1): Ditto.
4085         (movmisalign<mode>): Ditto.
4087 2013-08-20  Jan Hubicka  <jh@suse.cz>
4089         PR bootstrap/58186
4090         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
4091         entry for direct edges.
4092         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
4094 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
4096         Revert my last two changes, r201865 and r201864:
4098         Revert r201865:
4099         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
4101         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
4102         instances can own GC refs.
4104         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
4105         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
4106         (gcc::context::gt_pch_nx): Likewise.
4107         (gcc::context::gt_pch_nx):  Likewise.
4108         * ggc.h (gt_ggc_mx <T>): New.
4109         (gt_pch_nx_with_op <T>): New.
4110         (gt_pch_nx <T>): New.
4111         * passes.c (opt_pass::gt_ggc_mx): New.
4112         (opt_pass::gt_pch_nx): New.
4113         (opt_pass::gt_pch_nx_with_op): New.
4114         (pass_manager::gt_ggc_mx): New.
4115         (pass_manager::gt_pch_nx): New.
4116         (pass_manager::gt_pch_nx_with_op): New.
4117         (pass_manager::operator new): Use
4118         ggc_internal_cleared_alloc_stat rather than xcalloc.
4119         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
4120         (pass_manager::gt_ggc_mx): New.
4121         (pass_manager::gt_pch_nx): New.
4122         (pass_manager::gt_pch_nx_with_op): New.
4123         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
4124         (opt_pass::operator new): New.
4125         (opt_pass::gt_ggc_mx): New.
4126         (opt_pass::gt_pch_nx): New.
4127         (opt_pass::gt_pch_nx_with_op): New.
4129         Revert r201864:
4130         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
4132         * Makefile.in (GTFILES): Add context.h.
4133         * context.c (gcc::context::operator new): New.
4134         (gcc::context::gt_ggc_mx): New.
4135         (gcc::context::gt_pch_nx): New.
4136         (gcc::context::gt_pch_nx): New.
4137         * context.h (gcc::context): Add GTY((user)) marking.
4138         (gcc::context::operator new): New.
4139         (gcc::context::gt_ggc_mx): New.
4140         (gcc::context::gt_pch_nx): New.
4141         (gcc::context::gt_pch_nx): New.
4142         (g): Add GTY marking.
4143         (gt_ggc_mx (gcc::context *)): New.
4144         (gt_pch_nx (gcc::context *)): New.
4145         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
4146         void *cookie)): New.
4147         * gengtype.c (open_base_files) <ifiles>: Add context.h.
4149 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
4151         * dce.c (fini_dce): Call df_analyze again just in case
4152         delete_unmarked_insns removed anything.
4154 2013-08-20  Teresa Johnson  <tejohnson@google.com>
4156         PR rtl-optimizations/57451
4157         * final.c (reemit_insn_block_notes): Prevent lexical blocks
4158         from crossing split section boundaries.
4160 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
4162         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
4163         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
4164         with MULTLIB_DEFAULTS.
4166 2013-08-20  Nick Clifton  <nickc@redhat.com>
4168         * target.def (narrow_volatile_bitfield): Note that the default
4169         value is false, not !TARGET_STRICT_ALIGN.
4170         * doc/tm.texi: Regenerate.
4172 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
4174         Fix LIB_SPEC for systems without libpthread.
4176         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
4177         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
4178         for Android.
4179         * config/i386/linux-common.h: Likewise.
4180         * config/mips/linux-common.h: Likewise.
4182 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
4184         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
4185         checks.
4187 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
4189         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
4190         instances can own GC refs.
4192         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
4193         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
4194         (gcc::context::gt_pch_nx): Likewise.
4195         (gcc::context::gt_pch_nx):  Likewise.
4196         * ggc.h (gt_ggc_mx <T>): New.
4197         (gt_pch_nx_with_op <T>): New.
4198         (gt_pch_nx <T>): New.
4199         * passes.c (opt_pass::gt_ggc_mx): New.
4200         (opt_pass::gt_pch_nx): New.
4201         (opt_pass::gt_pch_nx_with_op): New.
4202         (pass_manager::gt_ggc_mx): New.
4203         (pass_manager::gt_pch_nx): New.
4204         (pass_manager::gt_pch_nx_with_op): New.
4205         (pass_manager::operator new): Use
4206         ggc_internal_cleared_alloc_stat rather than xcalloc.
4207         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
4208         (pass_manager::gt_ggc_mx): New.
4209         (pass_manager::gt_pch_nx): New.
4210         (pass_manager::gt_pch_nx_with_op): New.
4211         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
4212         (opt_pass::operator new): New.
4213         (opt_pass::gt_ggc_mx): New.
4214         (opt_pass::gt_pch_nx): New.
4215         (opt_pass::gt_pch_nx_with_op): New.
4217 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
4219         * Makefile.in (GTFILES): Add context.h.
4220         * context.c (gcc::context::operator new): New.
4221         (gcc::context::gt_ggc_mx): New.
4222         (gcc::context::gt_pch_nx): New.
4223         (gcc::context::gt_pch_nx): New.
4224         * context.h (gcc::context): Add GTY((user)) marking.
4225         (gcc::context::operator new): New.
4226         (gcc::context::gt_ggc_mx): New.
4227         (gcc::context::gt_pch_nx): New.
4228         (gcc::context::gt_pch_nx): New.
4229         (g): Add GTY marking.
4230         (gt_ggc_mx (gcc::context *)): New.
4231         (gt_pch_nx (gcc::context *)): New.
4232         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
4233         void *cookie)): New.
4234         * gengtype.c (open_base_files) <ifiles>: Add context.h.
4236 2013-08-20  Alan Modra  <amodra@gmail.com>
4238         PR target/57865
4239         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
4240         (rs6000_emit_epilogue): Likewise.
4242 2013-08-19  Dehao Chen  <dehao@google.com>
4244         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
4246 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
4247             Jakub Jelinek  <jakub@redhat.com>
4249         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
4250         (BUILT_IN_FABSD64): Likewise.
4251         (BUILT_IN_FABSD128): Likewise.
4252         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
4253         (fold_builtin_1): Likewise.
4254         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
4255         destination and source operands.
4256         (*abstd2_fpr): Likewise.
4257         (*nabstd2_fpr): Likewise.
4259 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
4261         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
4262         JUMP_P and INSN_P.
4264 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
4266         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
4267         incomplete.
4269 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4271         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
4272         * builtins.c (default_libc_has_function): New.
4273         (gnu_libc_has_function): Ditto.
4274         (no_c99_libc_has_function): Ditto.
4275         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
4276         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
4277         (fold_builtin_sincos): Likewise.
4278         (fold_builtin_cexp): Likewise.
4279         * builtins.def (DEF_C94_BUILTIN): Likewise.
4280         (DEF_C99_BUILTIN): Likewise.
4281         (DEF_C99_C90RES_BUILTIN): Likewise.
4282         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
4283         definitions to using this define.
4284         * config/darwin-protos.h (darwin_libc_has_function): New.
4285         * config/darwin.c (darwin_libc_has_function): Ditto.
4286         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
4287         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
4288         * config/darwin.h: Ditto.
4289         * config/elfos.h: Ditto.
4290         * config/freebsd.h: Ditto.
4291         * config/i386/cygming.h: Ditto.
4292         * config/i386/djgpp.h: Ditto.
4293         * config/i386/i386-interix.h: Ditto.
4294         * config/microblaze/microblaze.h: Ditto.
4295         * config/mmix/mmix.h: Ditto.
4296         * config/gnu-user.h: Ditto.
4297         * config/ia64/hpux.h: Ditto.
4298         * config/pa/pa-hpux.h: Ditto.
4299         * config/pdp11/pdp11.h: Ditto.
4300         * config/picochip/picochip.h: Ditto.
4301         * config/linux.h: Ditto.
4302         * config/netbsd.h: Ditto.
4303         * config/openbsd.h: Ditto.
4304         * config/rs6000/aix43.h: Ditto.
4305         * config/rs6000/aix51.h: Ditto.
4306         * config/rs6000/aix52.h: Ditto.
4307         * config/rs6000/aix53.h: Ditto.
4308         * config/rs6000/aix61.h: Ditto.
4309         * config/rs6000/darwin.h: Ditto.
4310         * config/rs6000/linux.h: Ditto.
4311         * config/rs6000/linux64.h: Ditto.
4312         * config/s390/tpf.h: Ditto.
4313         * config/sol2-10.h: Ditto.
4314         * config/sol2.h: Ditto.
4315         * config/vms/vms.h: Ditto.
4316         * config/vxworks.h: Ditto.
4317         * config/linux-android.c (linux_android_libc_has_function):
4318         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
4319         * config/linux-protos.h (linux_android_libc_has_function):
4320         New declaration.
4321         * config/i386/i386.c (ix86_libc_has_function): New.
4322         * config/i386/i386-protos.h
4323         (ix86_libc_has_function): New declaration.
4324         * config/i386/i386.md
4325         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
4326         ("isinf<mode>2): Likewise.
4327         * convert.c (convert_to_integer): Using new target hook
4328         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
4329         TARGET_C99_FUNCTIONS.
4330         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
4331         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
4332         * coretypes.h (function_class): New enum for different
4333         classes of functions.
4334         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
4335         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
4336         (TARGET_HAS_SINCOS): Likewise.
4337         (TARGET_LIBC_HAS_FUNCTION): New.
4338         * doc/tm.texi: Regenerated.
4339         * targhooks.h (default_libc_has_function): New declaration.
4340         (no_c99_libc_has_function): Ditto.
4341         (gnu_libc_has_function): Ditto.
4342         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
4343         and TARGET_HAS_SINCOS.
4345 2013-08-18  Jan Hubicka  <jh@suse.cz>
4347         * Makeifle-in (ipa-devirt.o): New.
4348         (GTFILES): Add ipa-utils.h and ipa-devirt.c
4349         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
4350         (analyze_functions): Look into possible targets of polymorphic call.
4351         * dumpfile.c (dump_files): Add type-inheritance dump.
4352         * dumpfile.h (TDI_inheritance): New.
4353         * ipa-devirt.c: New file.
4354         * ipa-utils.h (odr_type_d): Forward declare.
4355         (odr_type): New type.
4356         (build_type_inheritance_graph): Declare.
4357         (possible_polymorphic_call_targets): Declare and introduce inline
4358         variant when only edge is pased.
4359         (dump_possible_polymorphic_call_targets): Likewise.
4360         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
4361         * tree.c (type_in_anonymous_namespace_p): Break out from ...
4362         (types_same_for_odr): ... here.
4363         * tree.h (type_in_anonymous_namespace_p): Declare.
4365 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
4367         PR tree-optimization/58006
4368         * tree-parloops.c (take_address_of): Don't ICE if get_name
4369         returns NULL.
4370         (eliminate_local_variables_stmt): Remove clobber stmts.
4372 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
4374         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
4375         error message is issued for an alias to undefined symbol.
4377 2013-08-18  Jan Hubicka  <jh@suse.cz>
4379         * cgraph.c (cgraph_create_indirect_edge): Discover
4380         polymorphic calls and record basic info into indirect_info.
4381         * gimple-fold.c (gimple_fold_call): When doing BINFO based
4382         devirtualization, ignore objc function calls.
4383         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
4384         call with no parm index info.
4385         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
4386         * tree.c (virtual_method_call_p): New function.
4387         * tree.h (virtual_method_call_p): Declare.
4389 2013-08-16  Jan Hubicka  <jh@suse.cz>
4391         PR middle-end/58179
4392         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
4394 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
4396         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
4397         attribute.
4399 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
4401         * gengtype.c (type_for_name): Add special-case support for
4402         locating types within the "gcc::" namespace.
4403         (open_base_files): Emit a "using namespace gcc" directive.
4405 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
4407         PR target/58160
4408         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
4409         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
4411         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
4412         array instead of each individual operand as a separate argument.
4413         (emit_fusion_gpr_load): Likewise.
4414         (expand_fusion_gpr_load): Add new function declaration.
4416         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
4417         signature to have the operands passed as an array, instead of as
4418         separate arguments.  Allow ZERO_EXTEND to be in the memory
4419         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
4420         depend on the register live/dead flags when peepholes are run.
4421         (expand_fusion_gpr_load): New function to be called from the
4422         peephole2 pass, to change the register that addis sets to be the
4423         target register.
4424         (emit_fusion_gpr_load): Change the calling signature to have the
4425         operands passed as an array, instead of as separate arguments.
4426         Allow ZERO_EXTEND to be in the memory address, and also
4427         SIGN_EXTEND if -mpower8-fusion-sign.
4429         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
4430         unspec enumeration.
4431         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
4432         adjust the register addis loads up in the peephole2 pass.  Do not
4433         depend on the register live/dead state when the peephole pass is done.
4435 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
4437         * gengtype.c (create_user_defined_type): Ensure that the kind
4438         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
4439         declaration is seen before the GTY((user)) marking.
4441 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4443         PR target/58105
4444         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
4446 2013-08-16  Jan Hubicka  <jh@suse.cz>
4448         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
4449         arugment expected_type.
4450         (gimple_fold_call): Use it.
4451         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
4452         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
4453         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
4454         (try_make_edge_direct_virtual_call): Likewise.
4455         * tree.c (obj_type_ref_class): New.
4456         * tree.h (obj_type_ref_class): Use it.
4458 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
4460         * sched-vis.c (rtl_slim_pp_initialized): Remove.
4461         (rtl_slim_pp): Likewise.
4462         (init_rtl_slim_pretty_print): Likewise.
4463         (dump_value_slim):  Don't call it.  Use local pretty printer.
4464         (dump_insn_slim): Likewise.
4465         (dump_rtl_slim): Likewise.
4466         (str_pattern_slim): Likewise.
4467         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
4468         Simplify.
4470 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
4472         PR tree-optimization/58164
4473         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
4474         walk gimple_goto_dest of GIMPLE_GOTO.
4476         PR tree-optimization/58165
4477         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
4478         bi_call must be the last stmt in a bb, don't split_block, instead
4479         use fallthru edge from it and give up if there is none.
4480         Release conds vector when returning early.
4482 2013-08-14  Xinliang David Li  <davidxl@google.com>
4484         * config/i386/i386.c (ix86_option_override_internal):
4485         Remove unused variable and field.
4487 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4489         PR target/57949
4490         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
4491         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
4492         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
4493         and Linux, correct BLKmode alignment when 128-bit alignment is
4494         required and compatibility flag is not set.
4495         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
4496         for zero-size arguments when compatibility flag is not set.
4498 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
4500         PR tree-optimization/58145
4501         * tree-sra.c (build_ref_for_offset): If prev_base has
4502         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
4504 2013-08-14  Xinliang David Li  <davidxl@google.com>
4506         * config/i386/i386.c (ix86_option_override_internal):
4507         Fix uninitialized variable error.
4509 2013-08-14  Xinliang David Li  <davidxl@google.com>
4511         * config/i386/i386.opt: Define two new options.
4512         * config/i386/x86-tune.def: Add arch selector field in macros.
4513         * config/i386/i386.h: Adjust macro definition.
4514         * config/i386/i386.c (ix86_option_override_internal):
4515         Refactor the code.
4516         (parse_mtune_ctrl_str): New function.
4517         (set_ix86_tune_features): New function.
4518         (ix86_function_specific_restore): Call the new helper function.
4520 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
4522         PR rtl-optimization/57662
4523         * sel-sched.c (code_motion_process_successors): When the current insn
4524         is removed after the recursive traversal, break from the loop.
4525         Add comments and debug printouts.
4527 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
4528             Alexandre Oliva  <aoliva@redhat.com>
4530         PR target/58067
4531         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
4532         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
4533         there also UNSPEC_PLTOFF.
4535 2013-08-14  Marek Polacek  <polacek@redhat.com>
4537         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
4538         NUM_CONDITIONS bit positions.
4540 2013-08-13  Cary Coutant  <ccoutant@google.com>
4542         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
4543         (attr_checksum): Hash vector contents instead of pointer.
4544         (attr_checksum_ordered): Likewise.
4546 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
4548         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
4549         when Pmode != word_mode.  Add length_address attribute.
4550         (sse3_monitor_<mode>): Merge from sse3_monitor and
4551         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
4552         Pmode != word_mode.  Update insn length attribute.
4553         * config/i386/i386.c (ix86_option_override_internal): Update
4554         ix86_gen_monitor selection for merged sse3_monitor insn.
4556 2013-08-13  Julian Brown  <julian@codesourcery.com>
4558         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
4559         perform invalid legitimization on greater-than-word-size modes for
4560         TARGET_E500_DOUBLE.
4562 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
4564         * ira.c (setup_class_translate_array): Use aclass instead of cl
4565         for classes not fully covered by allocno classes.
4567 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
4569         PR tree-optimization/57661
4570         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
4571         * tree-inline.c (tree_function_versioning): Initialize it.
4572         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
4573         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
4574         that is not being copied.
4576         PR sanitizer/56417
4577         * asan.c (instrument_strlen_call): Fix typo in comment.
4578         Use char * type even for the lhs of POINTER_PLUS_EXPR.
4580 2013-08-13  Steve Ellcey  <sellcey@mips.com>
4582         * config/mips/mips.md (prefetch): Use lw instead of ld on
4583         loongson in 32bit mode.
4585 2013-08-13  Nick Clifton  <nickc@redhat.com>
4587         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
4589 2013-08-13  Jan Hubicka  <jh@suse.cz>
4591         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
4592         introduced edge; fix typo in sanity check.
4593         (cgraph_resolve_speculation): Export; improve diagnostic.
4594         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
4595         speculation at type mismatch.
4596         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
4597         (cgraph_resolve_speculation): Declare.
4598         (symtab_can_be_discarded): New function.
4599         * value-prof.c (gimple_ic_transform): Remove actual transform code.
4600         * ipa-inline-transform.c (speculation_removed): New global var.
4601         (clone_inlined_nodes): See if speculation can be removed.
4602         (inline_call): If speculations was removed, we growths may not match.
4603         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
4604         (speculation_useful_p): New function.
4605         (resolve_noninline_speculation): New function.
4606         (inline_small_functions): Resolve useless speculations.
4607         * ipa-inline.h (speculation_useful_p): Declare
4608         * ipa.c (can_replace_by_local_alias): Simplify.
4609         (ipa_profile): Produce speculative calls in non-lto, too;
4610         add simple cost model; produce local aliases.
4612 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
4614         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
4615         PASS_MANAGER_H.
4617 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
4619         * config/i386/i386.c (ix86_function_versions): Use error + inform.
4621 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
4623         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
4624         iterator instead of X87MODEF.
4626 2013-08-12  Perez Read  <netfirewall@gmail.com>
4628         PR target/58132
4629         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
4630         operand 0 for intel asm alternative.
4631         (*movabs<mode>_2): Ditto for operand 1.
4633 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
4635         * config/aarch64/arm_none.h
4636         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
4638 2013-08-12  Nick Clifton  <nickc@redhat.com>
4640         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
4642 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
4644         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
4645         expand for QI/HImode operand to produce more effictive code for
4646         unsigned char(short) --> float(double) conversion.
4648 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
4650         * doc/invoke.texi: Mention that -ftls-model does not force the final
4651         model.
4653 2013-08-12  Marek Polacek  <polacek@redhat.com>
4654             Marc Glisse  <marc.glisse@inria.fr>
4656         PR tree-optimization/57980
4657         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
4658         when creating -1 constant.
4660 2013-08-10  Jan Hubicka  <jh@suse.cz>
4662         Workaround binutils PR14342.
4663         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
4664         (gimple_init_edge_profiler): Likewise.
4665         (gimple_gen_ic_func_profiler): Likewise.
4667 2013-08-09  Jan Hubicka  <jh@suse.cz>
4669         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
4671 2013-08-09  Xinliang David Li  <davidxl@google.com>
4673         * config/i386/stringop.def: New file.
4674         * config/i386/stringop.opt: New file.
4675         * config/i386/i386-opts.h: Include stringopt.def.
4676         * config/i386/i386.opt: Include stringopt.opt.
4677         * config/i386/i386.c (ix86_option_override_internal):
4678         Override default size based stringop inline strategies with options.
4679         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
4680         New function.
4682 2013-08-09  Jan Hubicka  <jh@suse.cz>
4684         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
4686 2013-08-09  Jan Hubicka  <jh@suse.cz>
4688         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
4689         CGRAPH_FREQ_MAX.
4690         (dump_cgraph_node): Dump profile-id.
4691         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
4692         and common_target_probability.
4693         * lto-cgraph.c (lto_output_edge): Stream common targets.
4694         (lto_output_node): Stream profile ids.
4695         (input_node): Stream profile ids.
4696         (input_edge): Stream common targets.
4697         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
4698         * ipa.c: Include value-prof.h
4699         (ipa_profile_generate_summary): Turn indirect call statement histograms
4700         into common targets.
4701         (ipa_profile): Turn common targets into speculative edges.
4703 2013-08-09  Jan Hubicka  <jh@suse.cz>
4705         * cgraph.h (cgraph_node): Add profile_id.
4706         * value-prof.c (cgraph_node_map): Turn into pointer_map.
4707         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
4708         (del_node_map): Update.
4709         (find_func_by_funcdef_no): Replace by ...
4710         (find_func_by_profile_id): ... this one.
4711         (gimple_ic_transform): Do not remove useful histograms when
4712         speculation is not done; dump info when indirect call removal
4713         can happen at LTO.
4714         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
4715         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
4716         (__gcov_indirect_call_profiler_v2): .. this one.
4717         * profile.h (init_node_map): Update.
4718         * coverage.c (coverage_compute_profile_id): New function.
4719         * coverage.h (coverage_compute_profile_id): Declare.
4720         * tree-profile.c (init_ic_make_global_vars): Make
4721         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
4722         (gimple_init_edge_profiler): Update prototype of
4723         __gcov_indirect_call_profiler.
4724         (gimple_gen_ic_func_profiler): Simplify.
4725         (tree_profiling): Use init_node_map
4727 2013-08-09  Jan Hubicka  <jh@suse.cz>
4729         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
4730         non-speculative refs.
4731         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
4732         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
4733         (cgraph_set_call_stmt): Likewise.
4734         (cgraph_create_edge_1): Fix release checking compilatoin;
4735         clear lto_stmt_uid.
4736         (cgraph_free_edge): Free indirect info.
4737         (cgraph_turn_edge_to_speculative): New function.
4738         (cgraph_speculative_call_info): New function.
4739         (cgraph_make_edge_direct): Return direct edge; handle speculation.
4740         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
4741         (dump_cgraph_node): Dump speculation.
4742         (verify_edge_count_and_frequency): Accept speculative edges.
4743         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
4744         (verify_cgraph_node): Handle speculation.
4745         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
4746         (cgraph_set_call_stmt): Update prototype.
4747         (cgraph_make_edge_direct): Update prototype.
4748         (cgraph_speculative_call_info): Declare.
4749         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
4750         to change; update call of ipa_find_references.
4751         * ipa-ref.c (ipa_record_reference): Fix return value; clear
4752         lto_stmt_uid and speculative flags.
4753         (ipa_dump_references): Dump speculation.
4754         (ipa_clone_references): Clone speculative flag.
4755         (ipa_clone_referring): Likewise.
4756         (ipa_clone_ref): New function.
4757         (ipa_find_reference): Look into lto_stmt_uids
4758         (ipa_clear_stmts_in_references): Do not clear speculative calls.
4759         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
4760         (ipa_find_reference): Update declaration.
4761         (ipa_clone_ref): Declare.
4762         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
4763         stream speculative flag.
4764         (lto_output_ref): Stream statements uids and speculation.
4765         (input_ref): Likewise.
4766         (input_edge): Stream speuclation.
4767         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
4768         (cgraph_set_call_stmt_including_clones): Handle speculation.
4769         * ipa-inline.c (heap_edge_removal_hook): New function.
4770         (inline_small_functions): Register it.
4771         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
4772         also initialize refs.
4773         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
4774         edge to change.
4775         (try_make_edge_direct_simple_call): Likewise.
4776         (try_make_edge_direct_simple_call): Likewise.
4777         (update_indirect_edges_after_inlining): Likewise.
4778         (remove_described_reference): Look proper lto_stmt_uid.
4779         (propagate_controlled_uses): Likewise.
4780         (propagate_controlled_uses): Liekwise.
4781         * tree-inline.c (copy_bb): Copy speculative edges.
4782         (redirect_all_calls): New function.
4783         (copy_cfg_body): Do redirection after loop info is updated.
4784         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
4786 2013-08-09  Jan Hubicka  <jh@suse.cz>
4788         * lto-streamer-out.c (output_function): Renumber PHIs.
4789         * lto-streamer-in.c (input_function): Likewise.
4791 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
4793         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
4794         (get_lane_unsigned): Likewise.
4795         (dup_lane_scalar): Likewise.
4796         (get_lane): enable for VALL.
4797         * config/aarch64/aarch64-simd.md
4798         (aarch64_dup_lane_scalar<mode>): Remove.
4799         (aarch64_get_lane_signed<mode>): Likewise.
4800         (aarch64_get_lane_unsigned<mode>): Likewise.
4801         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
4802         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
4803         (aarch64_get_lane<mode>): Enable for all vector modes.
4804         (aarch64_get_lanedi): Remove misleading constraints.
4805         * config/aarch64/arm_neon.h
4806         (__aarch64_vget_lane_any): Define.
4807         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
4808         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
4809         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
4810         * config/aarch64/iterators.md (VDQQH): New.
4811         (VDQQHS): Likewise.
4812         (vwcore): Likewise.
4814 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
4816         * configure.ac: Add GAS check for LEON instructions on SPARC.
4817         * configure: Regenerate.
4818         * config.in: Likewise.
4819         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
4820         sparc*-*-* block.
4821         * config/sparc/sparc.opt (LEON, LEON3): New masks.
4822         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
4823         for LEON or LEON3.
4824         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
4825         (AS_LEON_FLAG): New macro.
4826         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
4827         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
4828         Deal with LEON and LEON3 for the memory model.
4829         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
4830         (atomic_compare_and_swap<mode>_1): Likewise.
4831         (*atomic_compare_and_swap<mode>_1): Likewise.
4833 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
4835         * config/arm/neon.md (vcond): Fix floating-point vector
4836         comparisons against 0.
4838 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
4840         * lra-constraints.c (emit_spill_move): Remove assert.
4841         (process_alt_operands): Add more debugging
4842         output.  Increase reject for spilling into memory.  Decrease
4843         reject for reloading scratch.
4844         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
4846 2013-08-08  Steve Ellcey  <sellcey@mips.com>
4848         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
4849         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
4850         micromips incompatible.  Add nan2008.
4851         (MULTILIB_DIRNAMES): Add nan2008.
4852         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
4853         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
4854         and micromips incompatible.  Add nan2008.
4855         (MULTILIB_DIRNAMES): Add nan2008.
4856         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
4858 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
4860         PR rtl-optimization/58079
4861         * combine.c (combine_simplify_rtx): Avoid using SUBST if
4862         simplify_comparison has widened a comparison with an integer.
4864 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4866         * config/arm/neon.md (movmisalign<mode>): Disable when we
4867         don't allow unaligned accesses.
4868         (*movmisalign<mode>_neon_store): Likewise.
4869         (*movmisalign<mode>_neon_load): Likewise.
4870         (*movmisalign<mode>_neon_store): Likewise.
4871         (*movmisalign<mode>_neon_load): Likewise.
4873 2013-08-08  Jan Hubicka  <jh@suse.cz>
4875         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
4876         (make_pass_rebuild_cgraph_edges): Also clear references.
4877         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
4878         * ipa-inline-transform.c (inline_transform): Remove all references
4879         after inlining.
4880         * cgraphunit.c (expand_function): Remove all references after
4881         expansion.
4882         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
4883         (ipa_find_reference): Rewrite to iterator.
4884         (remove_stmt_references): Likewise.
4885         (ipa_clear_stmts_in_references): New function.
4886         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
4887         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
4888         clear references.
4889         * ipa-split.c (split_function): Remove references in split function.
4891 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
4893         PR target/57431
4894         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
4895         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
4897 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
4899         PR target/56979
4900         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
4901         suggested mode for the assignment isn't compatible with the
4902         registers required.
4904 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4906         PR target/58065
4907         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
4909 2013-08-07  Xinliang David Li  <davidxl@google.com>
4911         * config/i386/i386.opt: New option -mtune-ctrl=.
4912         * config/i386/x86-tune.def: New file.
4913         * config/i386/i386.h: include x86-tune.def.
4914         * config/i386/i386.c (ix86_option_override_internal):
4915         Parsing -mtune-ctrl= option and set tune features.
4917 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
4919         PR other/12081
4920         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
4921         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
4922         to gen_2arg_fn_t.
4924 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
4926         * rtl.h (update_alignments): Declare.
4927         * final.c (grow_label_align): New function extracted from...
4928         (shorten_branches): ...here.  Call it.
4929         (update_alignments): New function.
4930         * reorg.c (sibling_labels): New variable.
4931         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
4932         the new label along with it onto the sibling_labels vector.
4933         (fill_simple_delay_slots): Adjust call to get_label_before.
4934         (fill_slots_from_thread): Likewise.
4935         (relax_delay_slots): Likewise.
4936         (make_return_insns): Likewise.
4937         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
4939 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
4941         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
4942         document its semantics.
4943         (diagnostic_report_diagnostic): Adjust accordingly.
4945 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
4947         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
4948         (sparc_option_override): ...and port to new C++ pass API.
4949         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
4951 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
4953         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
4955 2013-08-06  Caroline Tice  <cmtice@google.com>
4957         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
4958         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
4959         * tree-pass.h: Add pass_vtable_verify.
4960         * varasm.c (assemble_variable): Add code to properly set the comdat
4961         section and name for the .vtable_map_vars section.
4962         (assemble_vtyv_preinit_initializer): New function.
4963         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
4964         LINK_ONCE flag.
4965         * output.h: Add function decl for assemble_vtv_preinit_initializer.
4966         * vtable-verify.c: New file.
4967         * vtable-verify.h: New file.
4968         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
4969         initialiation levels.
4970         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
4971         * passes.def: Insert pass_vtable_verify.
4972         * aclocal.m4: Reorder includes.
4973         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
4974         -fvtv-counts options.
4975         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
4976         as appropriate, if -fvtable-verify=... is used.
4977         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
4978         -fvtable-verify=... is used.
4979         * Makefile.in (OBJS):  Add vtable-verify.o to list.
4980         (vtable-verify.o): Add new build rule.
4981         (GTFILES): Add vtable-verify.c to list.
4982         * common.opt (fvtable-verify=): New flag.
4983         (vtv_priority): Values for fvtable-verify= flag.
4984         (fvtv-counts): New flag.
4985         (fvtv-debug): New flag.
4986         * tree.h (save_vtable_map_decl): New extern function decl.
4988 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
4990         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
4991         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
4992         (pass_data_rl78_devirt): ...new pass_data instance and...
4993         (make_pass_rl78_devirt): ...new function.
4994         (rl78_asm_file_start): Port pass registration to new C++ API.
4996 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
4998         * coretypes.h (rtl_opt_pass): Add.
4999         (gcc::context): Add.
5000         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
5001         (epiphany_init): Port to new C++ pass API.
5002         (epiphany_optimize_mode_switching): Likewise.
5003         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
5004         (pass_manager::get_pass_mode_switching): New.
5005         (pass_manager::get_pass_peephole2): New.
5006         * mode-switching.c (pass_mode_switching): Add clone method.
5007         * recog.c (pass_peephole2): Add clone method.
5008         (pass_split_all_insns): Add clone method.
5010 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
5012         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
5013         (mips_option_override): ...here, porting to new C++ API for passes.
5015 2013-08-06  Jan Hubicka  <jh@suse.cz>
5017         * cgraph.c (cgraph_get_body): New function based on lto.c
5018         implementation.
5019         * cgraph.h (cgraph_get_body): Declare.
5020         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
5021         LTO paths.
5022         * cgraphunit.c (expand_function): Get body prior expanding.
5023         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
5024         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
5025         really need.
5026         * passes.c (do_per_function_toporder): Get body.
5027         * tree-inline.c (expand_call_inline): Get body prior inlining it.
5028         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
5030 2013-08-06  Martin Jambor  <mjambor@suse.cz>
5032         PR fortran/57987
5033         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
5034         is not re-finalized.  Rename second parameter to no_collect.
5036 2013-08-06  Martin Jambor  <mjambor@suse.cz>
5038         PR middle-end/58041
5039         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
5040         MEM_REF has proper alignment information.
5042 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
5044         PR other/12081
5045         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
5046         class insn_gen_fn.
5047         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
5048         rtx (*) (rtx, ...) with insn_gen_fn.
5049         * genoutput.c (output_insn_data): Cast gen_? function pointers to
5050         insn_gen_fn::stored_funcptr.  Add initializer braces.
5052 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
5054         Rewrite how instances of passes are cloned to remove assumptions
5055         about their sizes (thus allowing pass subclasses to have
5056         additional data fields, albeit non-GC-managed ones at this point).
5058         * passes.c (make_pass_instance): Now that passes have clone
5059         methods, rewrite this function to eliminate XNEW and memcpy
5060         calls that used hardcoded sizes.  Since this function no longer
5061         creates pass instances, rename it to...
5062         (add_pass_instance): ...this.  Document the old way that passes were
5063         numbered and flagged, and rework this function to continue using it.
5064         (next_pass_1): Add an initial_pass argument for use by
5065         add_pass_instance.
5066         (position_pass): When adding multiple instances of a pass, use
5067         the pass's clone method, rather than relying on the XNEW/memcpy
5068         within the former make_pass_instance (now add_pass_instance).
5069         (pass_manager::pass_manager): When invoking next_pass_1, also supply
5070         the initial instance of the current pass within the pass manager.
5072 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
5074         This is the automated part of the conversion of passes from C
5075         structs to C++ classes.
5077         Patch autogenerated by refactor_passes.py from
5078         https://github.com/davidmalcolm/gcc-refactoring-scripts
5079         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
5081         * asan.c (pass_asan): Convert from a global struct to a subclass of
5082         gimple_opt_pass along with...
5083         (pass_data_asan): ...new pass_data instance and...
5084         (make_pass_asan): ...new function.
5085         (pass_asan_O0): Convert from a global struct to a subclass of
5086         gimple_opt_pass along with...
5087         (pass_data_asan_O0): ...new pass_data instance and...
5088         (make_pass_asan_O0): ...new function.
5089         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
5090         subclass of rtl_opt_pass along with...
5091         (pass_data_inc_dec): ...new pass_data instance and...
5092         (make_pass_inc_dec): ...new function.
5093         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
5094         a subclass of rtl_opt_pass along with...
5095         (pass_data_reorder_blocks): ...new pass_data instance and...
5096         (make_pass_reorder_blocks): ...new function.
5097         (pass_duplicate_computed_gotos): Convert from a global struct to a
5098         subclass of rtl_opt_pass along with...
5099         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
5100         (make_pass_duplicate_computed_gotos): ...new function.
5101         (pass_partition_blocks): Convert from a global struct to a subclass of
5102         rtl_opt_pass along with...
5103         (pass_data_partition_blocks): ...new pass_data instance and...
5104         (make_pass_partition_blocks): ...new function.
5105         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
5106         struct to a subclass of rtl_opt_pass along with...
5107         (pass_data_branch_target_load_optimize1): ...new pass_data instance
5108         and...
5109         (make_pass_branch_target_load_optimize1): ...new function.
5110         (pass_branch_target_load_optimize2): Convert from a global struct to a
5111         subclass of rtl_opt_pass along with...
5112         (pass_data_branch_target_load_optimize2): ...new pass_data instance
5113         and...
5114         (make_pass_branch_target_load_optimize2): ...new function.
5115         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
5116         of rtl_opt_pass along with...
5117         (pass_data_jump): ...new pass_data instance and...
5118         (make_pass_jump): ...new function.
5119         (pass_jump2): Convert from a global struct to a subclass of
5120         rtl_opt_pass along with...
5121         (pass_data_jump2): ...new pass_data instance and...
5122         (make_pass_jump2): ...new function.
5123         * cfgexpand.c (pass_expand): Convert from a global struct to a
5124         subclass of rtl_opt_pass along with...
5125         (pass_data_expand): ...new pass_data instance and...
5126         (make_pass_expand): ...new function.
5127         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
5128         of rtl_opt_pass along with...
5129         (pass_data_free_cfg): ...new pass_data instance and...
5130         (make_pass_free_cfg): ...new function.
5131         (pass_into_cfg_layout_mode): Convert from a global struct to a
5132         subclass of rtl_opt_pass along with...
5133         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
5134         (make_pass_into_cfg_layout_mode): ...new function.
5135         (pass_outof_cfg_layout_mode): Convert from a global struct to a
5136         subclass of rtl_opt_pass along with...
5137         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
5138         (make_pass_outof_cfg_layout_mode): ...new function.
5139         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
5140         struct to a subclass of gimple_opt_pass along with...
5141         (pass_data_build_cgraph_edges): ...new pass_data instance and...
5142         (make_pass_build_cgraph_edges): ...new function.
5143         (pass_rebuild_cgraph_edges): Convert from a global struct to a
5144         subclass of gimple_opt_pass along with...
5145         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
5146         (make_pass_rebuild_cgraph_edges): ...new function.
5147         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
5148         subclass of gimple_opt_pass along with...
5149         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
5150         and...
5151         (make_pass_remove_cgraph_callee_edges): ...new function.
5152         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
5153         struct to a subclass of rtl_opt_pass along with...
5154         (pass_data_stack_adjustments): ...new pass_data instance and...
5155         (make_pass_stack_adjustments): ...new function.
5156         * combine.c (pass_combine): Convert from a global struct to a subclass
5157         of rtl_opt_pass along with...
5158         (pass_data_combine): ...new pass_data instance and...
5159         (make_pass_combine): ...new function.
5160         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
5161         global struct to a subclass of rtl_opt_pass along with...
5162         (pass_data_compare_elim_after_reload): ...new pass_data instance
5163         and...
5164         (make_pass_compare_elim_after_reload): ...new function.
5165         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
5166         of rtl_opt_pass along with...
5167         (pass_data_rtl_cprop): ...new pass_data instance and...
5168         (make_pass_rtl_cprop): ...new function.
5169         * cse.c (pass_cse): Convert from a global struct to a subclass of
5170         rtl_opt_pass along with...
5171         (pass_data_cse): ...new pass_data instance and...
5172         (make_pass_cse): ...new function.
5173         (pass_cse2): Convert from a global struct to a subclass of
5174         rtl_opt_pass along with...
5175         (pass_data_cse2): ...new pass_data instance and...
5176         (make_pass_cse2): ...new function.
5177         (pass_cse_after_global_opts): Convert from a global struct to a
5178         subclass of rtl_opt_pass along with...
5179         (pass_data_cse_after_global_opts): ...new pass_data instance and...
5180         (make_pass_cse_after_global_opts): ...new function.
5181         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
5182         of rtl_opt_pass along with...
5183         (pass_data_ud_rtl_dce): ...new pass_data instance and...
5184         (make_pass_ud_rtl_dce): ...new function.
5185         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
5186         rtl_opt_pass along with...
5187         (pass_data_fast_rtl_dce): ...new pass_data instance and...
5188         (make_pass_fast_rtl_dce): ...new function.
5189         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
5190         a subclass of rtl_opt_pass along with...
5191         (pass_data_df_initialize_opt): ...new pass_data instance and...
5192         (make_pass_df_initialize_opt): ...new function.
5193         (pass_df_initialize_no_opt): Convert from a global struct to a
5194         subclass of rtl_opt_pass along with...
5195         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
5196         (make_pass_df_initialize_no_opt): ...new function.
5197         (pass_df_finish): Convert from a global struct to a subclass of
5198         rtl_opt_pass along with...
5199         (pass_data_df_finish): ...new pass_data instance and...
5200         (make_pass_df_finish): ...new function.
5201         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
5202         rtl_opt_pass along with...
5203         (pass_data_rtl_dse1): ...new pass_data instance and...
5204         (make_pass_rtl_dse1): ...new function.
5205         (pass_rtl_dse2): Convert from a global struct to a subclass of
5206         rtl_opt_pass along with...
5207         (pass_data_rtl_dse2): ...new pass_data instance and...
5208         (make_pass_rtl_dse2): ...new function.
5209         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
5210         subclass of rtl_opt_pass along with...
5211         (pass_data_dwarf2_frame): ...new pass_data instance and...
5212         (make_pass_dwarf2_frame): ...new function.
5213         * except.c (pass_set_nothrow_function_flags): Convert from a global
5214         struct to a subclass of rtl_opt_pass along with...
5215         (pass_data_set_nothrow_function_flags): ...new pass_data instance
5216         and...
5217         (make_pass_set_nothrow_function_flags): ...new function.
5218         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
5219         subclass of rtl_opt_pass along with...
5220         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
5221         and...
5222         (make_pass_convert_to_eh_region_ranges): ...new function.
5223         * final.c (pass_compute_alignments): Convert from a global struct to a
5224         subclass of rtl_opt_pass along with...
5225         (pass_data_compute_alignments): ...new pass_data instance and...
5226         (make_pass_compute_alignments): ...new function.
5227         (pass_final): Convert from a global struct to a subclass of
5228         rtl_opt_pass along with...
5229         (pass_data_final): ...new pass_data instance and...
5230         (make_pass_final): ...new function.
5231         (pass_shorten_branches): Convert from a global struct to a subclass of
5232         rtl_opt_pass along with...
5233         (pass_data_shorten_branches): ...new pass_data instance and...
5234         (make_pass_shorten_branches): ...new function.
5235         (pass_clean_state): Convert from a global struct to a subclass of
5236         rtl_opt_pass along with...
5237         (pass_data_clean_state): ...new pass_data instance and...
5238         (make_pass_clean_state): ...new function.
5239         * function.c (pass_instantiate_virtual_regs): Convert from a global
5240         struct to a subclass of rtl_opt_pass along with...
5241         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
5242         (make_pass_instantiate_virtual_regs): ...new function.
5243         (pass_leaf_regs): Convert from a global struct to a subclass of
5244         rtl_opt_pass along with...
5245         (pass_data_leaf_regs): ...new pass_data instance and...
5246         (make_pass_leaf_regs): ...new function.
5247         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
5248         subclass of rtl_opt_pass along with...
5249         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
5250         and...
5251         (make_pass_thread_prologue_and_epilogue): ...new function.
5252         (pass_match_asm_constraints): Convert from a global struct to a
5253         subclass of rtl_opt_pass along with...
5254         (pass_data_match_asm_constraints): ...new pass_data instance and...
5255         (make_pass_match_asm_constraints): ...new function.
5256         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
5257         subclass of rtl_opt_pass along with...
5258         (pass_data_rtl_fwprop): ...new pass_data instance and...
5259         (make_pass_rtl_fwprop): ...new function.
5260         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
5261         rtl_opt_pass along with...
5262         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
5263         (make_pass_rtl_fwprop_addr): ...new function.
5264         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
5265         rtl_opt_pass along with...
5266         (pass_data_rtl_pre): ...new pass_data instance and...
5267         (make_pass_rtl_pre): ...new function.
5268         (pass_rtl_hoist): Convert from a global struct to a subclass of
5269         rtl_opt_pass along with...
5270         (pass_data_rtl_hoist): ...new pass_data instance and...
5271         (make_pass_rtl_hoist): ...new function.
5272         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
5273         subclass of gimple_opt_pass along with...
5274         (pass_data_lower_cf): ...new pass_data instance and...
5275         (make_pass_lower_cf): ...new function.
5276         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
5277         from a global struct to a subclass of gimple_opt_pass along with...
5278         (pass_data_strength_reduction): ...new pass_data instance and...
5279         (make_pass_strength_reduction): ...new function.
5280         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
5281         of rtl_opt_pass along with...
5282         (pass_data_rtl_ifcvt): ...new pass_data instance and...
5283         (make_pass_rtl_ifcvt): ...new function.
5284         (pass_if_after_combine): Convert from a global struct to a subclass of
5285         rtl_opt_pass along with...
5286         (pass_data_if_after_combine): ...new pass_data instance and...
5287         (make_pass_if_after_combine): ...new function.
5288         (pass_if_after_reload): Convert from a global struct to a subclass of
5289         rtl_opt_pass along with...
5290         (pass_data_if_after_reload): ...new pass_data instance and...
5291         (make_pass_if_after_reload): ...new function.
5292         * init-regs.c (pass_initialize_regs): Convert from a global struct to
5293         a subclass of rtl_opt_pass along with...
5294         (pass_data_initialize_regs): ...new pass_data instance and...
5295         (make_pass_initialize_regs): ...new function.
5296         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
5297         of ipa_opt_pass_d along with...
5298         (pass_data_ipa_cp): ...new pass_data instance and...
5299         (make_pass_ipa_cp): ...new function.
5300         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
5301         global struct to a subclass of gimple_opt_pass along with...
5302         (pass_data_inline_parameters): ...new pass_data instance and...
5303         (make_pass_inline_parameters): ...new function.
5304         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
5305         subclass of gimple_opt_pass along with...
5306         (pass_data_early_inline): ...new pass_data instance and...
5307         (make_pass_early_inline): ...new function.
5308         (pass_ipa_inline): Convert from a global struct to a subclass of
5309         ipa_opt_pass_d along with...
5310         (pass_data_ipa_inline): ...new pass_data instance and...
5311         (make_pass_ipa_inline): ...new function.
5312         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
5313         struct to a subclass of gimple_opt_pass along with...
5314         (pass_data_local_pure_const): ...new pass_data instance and...
5315         (make_pass_local_pure_const): ...new function.
5316         (pass_ipa_pure_const): Convert from a global struct to a subclass of
5317         ipa_opt_pass_d along with...
5318         (pass_data_ipa_pure_const): ...new pass_data instance and...
5319         (make_pass_ipa_pure_const): ...new function.
5320         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
5321         to a subclass of ipa_opt_pass_d along with...
5322         (pass_data_ipa_reference): ...new pass_data instance and...
5323         (make_pass_ipa_reference): ...new function.
5324         * ipa-split.c (pass_split_functions): Convert from a global struct to
5325         a subclass of gimple_opt_pass along with...
5326         (pass_data_split_functions): ...new pass_data instance and...
5327         (make_pass_split_functions): ...new function.
5328         (pass_feedback_split_functions): Convert from a global struct to a
5329         subclass of gimple_opt_pass along with...
5330         (pass_data_feedback_split_functions): ...new pass_data instance and...
5331         (make_pass_feedback_split_functions): ...new function.
5332         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
5333         global struct to a subclass of simple_ipa_opt_pass along with...
5334         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
5335         instance and...
5336         (make_pass_ipa_function_and_variable_visibility): ...new function.
5337         (pass_ipa_free_inline_summary): Convert from a global struct to a
5338         subclass of simple_ipa_opt_pass along with...
5339         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
5340         (make_pass_ipa_free_inline_summary): ...new function.
5341         (pass_ipa_whole_program_visibility): Convert from a global struct to a
5342         subclass of ipa_opt_pass_d along with...
5343         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
5344         and...
5345         (make_pass_ipa_whole_program_visibility): ...new function.
5346         (pass_ipa_profile): Convert from a global struct to a subclass of
5347         ipa_opt_pass_d along with...
5348         (pass_data_ipa_profile): ...new pass_data instance and...
5349         (make_pass_ipa_profile): ...new function.
5350         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
5351         ipa_opt_pass_d along with...
5352         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
5353         (make_pass_ipa_cdtor_merge): ...new function.
5354         * ira.c (pass_ira): Convert from a global struct to a subclass of
5355         rtl_opt_pass along with...
5356         (pass_data_ira): ...new pass_data instance and...
5357         (make_pass_ira): ...new function.
5358         (pass_reload): Convert from a global struct to a subclass of
5359         rtl_opt_pass along with...
5360         (pass_data_reload): ...new pass_data instance and...
5361         (make_pass_reload): ...new function.
5362         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
5363         subclass of rtl_opt_pass along with...
5364         (pass_data_cleanup_barriers): ...new pass_data instance and...
5365         (make_pass_cleanup_barriers): ...new function.
5366         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
5367         of rtl_opt_pass along with...
5368         (pass_data_loop2): ...new pass_data instance and...
5369         (make_pass_loop2): ...new function.
5370         (pass_rtl_loop_init): Convert from a global struct to a subclass of
5371         rtl_opt_pass along with...
5372         (pass_data_rtl_loop_init): ...new pass_data instance and...
5373         (make_pass_rtl_loop_init): ...new function.
5374         (pass_rtl_loop_done): Convert from a global struct to a subclass of
5375         rtl_opt_pass along with...
5376         (pass_data_rtl_loop_done): ...new pass_data instance and...
5377         (make_pass_rtl_loop_done): ...new function.
5378         (pass_rtl_move_loop_invariants): Convert from a global struct to a
5379         subclass of rtl_opt_pass along with...
5380         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
5381         (make_pass_rtl_move_loop_invariants): ...new function.
5382         (pass_rtl_unswitch): Convert from a global struct to a subclass of
5383         rtl_opt_pass along with...
5384         (pass_data_rtl_unswitch): ...new pass_data instance and...
5385         (make_pass_rtl_unswitch): ...new function.
5386         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
5387         subclass of rtl_opt_pass along with...
5388         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
5389         and...
5390         (make_pass_rtl_unroll_and_peel_loops): ...new function.
5391         (pass_rtl_doloop): Convert from a global struct to a subclass of
5392         rtl_opt_pass along with...
5393         (pass_data_rtl_doloop): ...new pass_data instance and...
5394         (make_pass_rtl_doloop): ...new function.
5395         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
5396         a subclass of rtl_opt_pass along with...
5397         (pass_data_lower_subreg): ...new pass_data instance and...
5398         (make_pass_lower_subreg): ...new function.
5399         (pass_lower_subreg2): Convert from a global struct to a subclass of
5400         rtl_opt_pass along with...
5401         (pass_data_lower_subreg2): ...new pass_data instance and...
5402         (make_pass_lower_subreg2): ...new function.
5403         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
5404         struct to a subclass of ipa_opt_pass_d along with...
5405         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
5406         (make_pass_ipa_lto_gimple_out): ...new function.
5407         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
5408         of ipa_opt_pass_d along with...
5409         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
5410         (make_pass_ipa_lto_finish_out): ...new function.
5411         * mode-switching.c (pass_mode_switching): Convert from a global struct
5412         to a subclass of rtl_opt_pass along with...
5413         (pass_data_mode_switching): ...new pass_data instance and...
5414         (make_pass_mode_switching): ...new function.
5415         * modulo-sched.c (pass_sms): Convert from a global struct to a
5416         subclass of rtl_opt_pass along with...
5417         (pass_data_sms): ...new pass_data instance and...
5418         (make_pass_sms): ...new function.
5419         * omp-low.c (pass_expand_omp): Convert from a global struct to a
5420         subclass of gimple_opt_pass along with...
5421         (pass_data_expand_omp): ...new pass_data instance and...
5422         (make_pass_expand_omp): ...new function.
5423         (pass_lower_omp): Convert from a global struct to a subclass of
5424         gimple_opt_pass along with...
5425         (pass_data_lower_omp): ...new pass_data instance and...
5426         (make_pass_lower_omp): ...new function.
5427         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
5428         of gimple_opt_pass along with...
5429         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
5430         (make_pass_diagnose_omp_blocks): ...new function.
5431         * passes.c (pass_early_local_passes): Convert from a global struct to
5432         a subclass of simple_ipa_opt_pass along with...
5433         (pass_data_early_local_passes): ...new pass_data instance and...
5434         (make_pass_early_local_passes): ...new function.
5435         (pass_all_early_optimizations): Convert from a global struct to a
5436         subclass of gimple_opt_pass along with...
5437         (pass_data_all_early_optimizations): ...new pass_data instance and...
5438         (make_pass_all_early_optimizations): ...new function.
5439         (pass_all_optimizations): Convert from a global struct to a subclass
5440         of gimple_opt_pass along with...
5441         (pass_data_all_optimizations): ...new pass_data instance and...
5442         (make_pass_all_optimizations): ...new function.
5443         (pass_all_optimizations_g): Convert from a global struct to a subclass
5444         of gimple_opt_pass along with...
5445         (pass_data_all_optimizations_g): ...new pass_data instance and...
5446         (make_pass_all_optimizations_g): ...new function.
5447         (pass_rest_of_compilation): Convert from a global struct to a subclass
5448         of rtl_opt_pass along with...
5449         (pass_data_rest_of_compilation): ...new pass_data instance and...
5450         (make_pass_rest_of_compilation): ...new function.
5451         (pass_postreload): Convert from a global struct to a subclass of
5452         rtl_opt_pass along with...
5453         (pass_data_postreload): ...new pass_data instance and...
5454         (make_pass_postreload): ...new function.
5455         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
5456         subclass of rtl_opt_pass along with...
5457         (pass_data_gcse2): ...new pass_data instance and...
5458         (make_pass_gcse2): ...new function.
5459         * postreload.c (pass_postreload_cse): Convert from a global struct to
5460         a subclass of rtl_opt_pass along with...
5461         (pass_data_postreload_cse): ...new pass_data instance and...
5462         (make_pass_postreload_cse): ...new function.
5463         * predict.c (pass_profile): Convert from a global struct to a subclass
5464         of gimple_opt_pass along with...
5465         (pass_data_profile): ...new pass_data instance and...
5466         (make_pass_profile): ...new function.
5467         (pass_strip_predict_hints): Convert from a global struct to a subclass
5468         of gimple_opt_pass along with...
5469         (pass_data_strip_predict_hints): ...new pass_data instance and...
5470         (make_pass_strip_predict_hints): ...new function.
5471         * recog.c (pass_peephole2): Convert from a global struct to a subclass
5472         of rtl_opt_pass along with...
5473         (pass_data_peephole2): ...new pass_data instance and...
5474         (make_pass_peephole2): ...new function.
5475         (pass_split_all_insns): Convert from a global struct to a subclass of
5476         rtl_opt_pass along with...
5477         (pass_data_split_all_insns): ...new pass_data instance and...
5478         (make_pass_split_all_insns): ...new function.
5479         (pass_split_after_reload): Convert from a global struct to a subclass
5480         of rtl_opt_pass along with...
5481         (pass_data_split_after_reload): ...new pass_data instance and...
5482         (make_pass_split_after_reload): ...new function.
5483         (pass_split_before_regstack): Convert from a global struct to a
5484         subclass of rtl_opt_pass along with...
5485         (pass_data_split_before_regstack): ...new pass_data instance and...
5486         (make_pass_split_before_regstack): ...new function.
5487         (pass_split_before_sched2): Convert from a global struct to a subclass
5488         of rtl_opt_pass along with...
5489         (pass_data_split_before_sched2): ...new pass_data instance and...
5490         (make_pass_split_before_sched2): ...new function.
5491         (pass_split_for_shorten_branches): Convert from a global struct to a
5492         subclass of rtl_opt_pass along with...
5493         (pass_data_split_for_shorten_branches): ...new pass_data instance
5494         and...
5495         (make_pass_split_for_shorten_branches): ...new function.
5496         * ree.c (pass_ree): Convert from a global struct to a subclass of
5497         rtl_opt_pass along with...
5498         (pass_data_ree): ...new pass_data instance and...
5499         (make_pass_ree): ...new function.
5500         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
5501         subclass of rtl_opt_pass along with...
5502         (pass_data_stack_regs): ...new pass_data instance and...
5503         (make_pass_stack_regs): ...new function.
5504         (pass_stack_regs_run): Convert from a global struct to a subclass of
5505         rtl_opt_pass along with...
5506         (pass_data_stack_regs_run): ...new pass_data instance and...
5507         (make_pass_stack_regs_run): ...new function.
5508         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
5509         subclass of rtl_opt_pass along with...
5510         (pass_data_cprop_hardreg): ...new pass_data instance and...
5511         (make_pass_cprop_hardreg): ...new function.
5512         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
5513         subclass of rtl_opt_pass along with...
5514         (pass_data_reginfo_init): ...new pass_data instance and...
5515         (make_pass_reginfo_init): ...new function.
5516         * regmove.c (pass_regmove): Convert from a global struct to a subclass
5517         of rtl_opt_pass along with...
5518         (pass_data_regmove): ...new pass_data instance and...
5519         (make_pass_regmove): ...new function.
5520         * regrename.c (pass_regrename): Convert from a global struct to a
5521         subclass of rtl_opt_pass along with...
5522         (pass_data_regrename): ...new pass_data instance and...
5523         (make_pass_regrename): ...new function.
5524         * reorg.c (pass_delay_slots): Convert from a global struct to a
5525         subclass of rtl_opt_pass along with...
5526         (pass_data_delay_slots): ...new pass_data instance and...
5527         (make_pass_delay_slots): ...new function.
5528         (pass_machine_reorg): Convert from a global struct to a subclass of
5529         rtl_opt_pass along with...
5530         (pass_data_machine_reorg): ...new pass_data instance and...
5531         (make_pass_machine_reorg): ...new function.
5532         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
5533         of rtl_opt_pass along with...
5534         (pass_data_sched): ...new pass_data instance and...
5535         (make_pass_sched): ...new function.
5536         (pass_sched2): Convert from a global struct to a subclass of
5537         rtl_opt_pass along with...
5538         (pass_data_sched2): ...new pass_data instance and...
5539         (make_pass_sched2): ...new function.
5540         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
5541         to a subclass of rtl_opt_pass along with...
5542         (pass_data_stack_ptr_mod): ...new pass_data instance and...
5543         (make_pass_stack_ptr_mod): ...new function.
5544         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
5545         to a subclass of rtl_opt_pass along with...
5546         (pass_data_rtl_store_motion): ...new pass_data instance and...
5547         (make_pass_rtl_store_motion): ...new function.
5548         * tracer.c (pass_tracer): Convert from a global struct to a subclass
5549         of gimple_opt_pass along with...
5550         (pass_data_tracer): ...new pass_data instance and...
5551         (make_pass_tracer): ...new function.
5552         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
5553         to a subclass of gimple_opt_pass along with...
5554         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
5555         (make_pass_diagnose_tm_blocks): ...new function.
5556         (pass_lower_tm): Convert from a global struct to a subclass of
5557         gimple_opt_pass along with...
5558         (pass_data_lower_tm): ...new pass_data instance and...
5559         (make_pass_lower_tm): ...new function.
5560         (pass_tm_init): Convert from a global struct to a subclass of
5561         gimple_opt_pass along with...
5562         (pass_data_tm_init): ...new pass_data instance and...
5563         (make_pass_tm_init): ...new function.
5564         (pass_tm_mark): Convert from a global struct to a subclass of
5565         gimple_opt_pass along with...
5566         (pass_data_tm_mark): ...new pass_data instance and...
5567         (make_pass_tm_mark): ...new function.
5568         (pass_tm_edges): Convert from a global struct to a subclass of
5569         gimple_opt_pass along with...
5570         (pass_data_tm_edges): ...new pass_data instance and...
5571         (make_pass_tm_edges): ...new function.
5572         (pass_tm_memopt): Convert from a global struct to a subclass of
5573         gimple_opt_pass along with...
5574         (pass_data_tm_memopt): ...new pass_data instance and...
5575         (make_pass_tm_memopt): ...new function.
5576         (pass_ipa_tm): Convert from a global struct to a subclass of
5577         simple_ipa_opt_pass along with...
5578         (pass_data_ipa_tm): ...new pass_data instance and...
5579         (make_pass_ipa_tm): ...new function.
5580         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
5581         subclass of gimple_opt_pass along with...
5582         (pass_data_call_cdce): ...new pass_data instance and...
5583         (make_pass_call_cdce): ...new function.
5584         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
5585         subclass of gimple_opt_pass along with...
5586         (pass_data_build_cfg): ...new pass_data instance and...
5587         (make_pass_build_cfg): ...new function.
5588         (pass_split_crit_edges): Convert from a global struct to a subclass of
5589         gimple_opt_pass along with...
5590         (pass_data_split_crit_edges): ...new pass_data instance and...
5591         (make_pass_split_crit_edges): ...new function.
5592         (pass_warn_function_return): Convert from a global struct to a
5593         subclass of gimple_opt_pass along with...
5594         (pass_data_warn_function_return): ...new pass_data instance and...
5595         (make_pass_warn_function_return): ...new function.
5596         (pass_warn_function_noreturn): Convert from a global struct to a
5597         subclass of gimple_opt_pass along with...
5598         (pass_data_warn_function_noreturn): ...new pass_data instance and...
5599         (make_pass_warn_function_noreturn): ...new function.
5600         (pass_warn_unused_result): Convert from a global struct to a subclass
5601         of gimple_opt_pass along with...
5602         (pass_data_warn_unused_result): ...new pass_data instance and...
5603         (make_pass_warn_unused_result): ...new function.
5604         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
5605         a subclass of gimple_opt_pass along with...
5606         (pass_data_merge_phi): ...new pass_data instance and...
5607         (make_pass_merge_phi): ...new function.
5608         * tree-complex.c (pass_lower_complex): Convert from a global struct to
5609         a subclass of gimple_opt_pass along with...
5610         (pass_data_lower_complex): ...new pass_data instance and...
5611         (make_pass_lower_complex): ...new function.
5612         (pass_lower_complex_O0): Convert from a global struct to a subclass of
5613         gimple_opt_pass along with...
5614         (pass_data_lower_complex_O0): ...new pass_data instance and...
5615         (make_pass_lower_complex_O0): ...new function.
5616         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
5617         subclass of gimple_opt_pass along with...
5618         (pass_data_lower_eh): ...new pass_data instance and...
5619         (make_pass_lower_eh): ...new function.
5620         (pass_refactor_eh): Convert from a global struct to a subclass of
5621         gimple_opt_pass along with...
5622         (pass_data_refactor_eh): ...new pass_data instance and...
5623         (make_pass_refactor_eh): ...new function.
5624         (pass_lower_resx): Convert from a global struct to a subclass of
5625         gimple_opt_pass along with...
5626         (pass_data_lower_resx): ...new pass_data instance and...
5627         (make_pass_lower_resx): ...new function.
5628         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
5629         of gimple_opt_pass along with...
5630         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
5631         (make_pass_lower_eh_dispatch): ...new function.
5632         (pass_cleanup_eh): Convert from a global struct to a subclass of
5633         gimple_opt_pass along with...
5634         (pass_data_cleanup_eh): ...new pass_data instance and...
5635         (make_pass_cleanup_eh): ...new function.
5636         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
5637         to a subclass of simple_ipa_opt_pass along with...
5638         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
5639         (make_pass_ipa_lower_emutls): ...new function.
5640         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
5641         a subclass of gimple_opt_pass along with...
5642         (pass_data_if_conversion): ...new pass_data instance and...
5643         (make_pass_if_conversion): ...new function.
5644         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
5645         subclass of gimple_opt_pass along with...
5646         (pass_data_build_ssa): ...new pass_data instance and...
5647         (make_pass_build_ssa): ...new function.
5648         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
5649         global struct to a subclass of gimple_opt_pass along with...
5650         (pass_data_loop_distribution): ...new pass_data instance and...
5651         (make_pass_loop_distribution): ...new function.
5652         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
5653         subclass of gimple_opt_pass along with...
5654         (pass_data_mudflap_1): ...new pass_data instance and...
5655         (make_pass_mudflap_1): ...new function.
5656         (pass_mudflap_2): Convert from a global struct to a subclass of
5657         gimple_opt_pass along with...
5658         (pass_data_mudflap_2): ...new pass_data instance and...
5659         (make_pass_mudflap_2): ...new function.
5660         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
5661         subclass of gimple_opt_pass along with...
5662         (pass_data_mudflap_1): ...new pass_data instance and...
5663         (make_pass_mudflap_1): ...new function.
5664         (pass_mudflap_2): Convert from a global struct to a subclass of
5665         gimple_opt_pass along with...
5666         (pass_data_mudflap_2): ...new pass_data instance and...
5667         (make_pass_mudflap_2): ...new function.
5668         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
5669         gimple_opt_pass along with...
5670         (pass_data_nrv): ...new pass_data instance and...
5671         (make_pass_nrv): ...new function.
5672         (pass_return_slot): Convert from a global struct to a subclass of
5673         gimple_opt_pass along with...
5674         (pass_data_return_slot): ...new pass_data instance and...
5675         (make_pass_return_slot): ...new function.
5676         * tree-object-size.c (pass_object_sizes): Convert from a global struct
5677         to a subclass of gimple_opt_pass along with...
5678         (pass_data_object_sizes): ...new pass_data instance and...
5679         (make_pass_object_sizes): ...new function.
5680         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
5681         global struct to a subclass of gimple_opt_pass along with...
5682         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
5683         and...
5684         (make_pass_cleanup_cfg_post_optimizing): ...new function.
5685         (pass_fixup_cfg): Convert from a global struct to a subclass of
5686         gimple_opt_pass along with...
5687         (pass_data_fixup_cfg): ...new pass_data instance and...
5688         (make_pass_fixup_cfg): ...new function.
5689         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
5690         (make_pass_mudflap_1): ...new function.
5691         (pass_mudflap_2): Replace declaration with that of...
5692         (make_pass_mudflap_2): ...new function.
5693         (pass_asan): Replace declaration with that of...
5694         (make_pass_asan): ...new function.
5695         (pass_asan_O0): Replace declaration with that of...
5696         (make_pass_asan_O0): ...new function.
5697         (pass_tsan): Replace declaration with that of...
5698         (make_pass_tsan): ...new function.
5699         (pass_tsan_O0): Replace declaration with that of...
5700         (make_pass_tsan_O0): ...new function.
5701         (pass_lower_cf): Replace declaration with that of...
5702         (make_pass_lower_cf): ...new function.
5703         (pass_refactor_eh): Replace declaration with that of...
5704         (make_pass_refactor_eh): ...new function.
5705         (pass_lower_eh): Replace declaration with that of...
5706         (make_pass_lower_eh): ...new function.
5707         (pass_lower_eh_dispatch): Replace declaration with that of...
5708         (make_pass_lower_eh_dispatch): ...new function.
5709         (pass_lower_resx): Replace declaration with that of...
5710         (make_pass_lower_resx): ...new function.
5711         (pass_build_cfg): Replace declaration with that of...
5712         (make_pass_build_cfg): ...new function.
5713         (pass_early_tree_profile): Replace declaration with that of...
5714         (make_pass_early_tree_profile): ...new function.
5715         (pass_cleanup_eh): Replace declaration with that of...
5716         (make_pass_cleanup_eh): ...new function.
5717         (pass_sra): Replace declaration with that of...
5718         (make_pass_sra): ...new function.
5719         (pass_sra_early): Replace declaration with that of...
5720         (make_pass_sra_early): ...new function.
5721         (pass_early_ipa_sra): Replace declaration with that of...
5722         (make_pass_early_ipa_sra): ...new function.
5723         (pass_tail_recursion): Replace declaration with that of...
5724         (make_pass_tail_recursion): ...new function.
5725         (pass_tail_calls): Replace declaration with that of...
5726         (make_pass_tail_calls): ...new function.
5727         (pass_tree_loop): Replace declaration with that of...
5728         (make_pass_tree_loop): ...new function.
5729         (pass_tree_loop_init): Replace declaration with that of...
5730         (make_pass_tree_loop_init): ...new function.
5731         (pass_lim): Replace declaration with that of...
5732         (make_pass_lim): ...new function.
5733         (pass_tree_unswitch): Replace declaration with that of...
5734         (make_pass_tree_unswitch): ...new function.
5735         (pass_predcom): Replace declaration with that of...
5736         (make_pass_predcom): ...new function.
5737         (pass_iv_canon): Replace declaration with that of...
5738         (make_pass_iv_canon): ...new function.
5739         (pass_scev_cprop): Replace declaration with that of...
5740         (make_pass_scev_cprop): ...new function.
5741         (pass_empty_loop): Replace declaration with that of...
5742         (make_pass_empty_loop): ...new function.
5743         (pass_record_bounds): Replace declaration with that of...
5744         (make_pass_record_bounds): ...new function.
5745         (pass_graphite): Replace declaration with that of...
5746         (make_pass_graphite): ...new function.
5747         (pass_graphite_transforms): Replace declaration with that of...
5748         (make_pass_graphite_transforms): ...new function.
5749         (pass_if_conversion): Replace declaration with that of...
5750         (make_pass_if_conversion): ...new function.
5751         (pass_loop_distribution): Replace declaration with that of...
5752         (make_pass_loop_distribution): ...new function.
5753         (pass_vectorize): Replace declaration with that of...
5754         (make_pass_vectorize): ...new function.
5755         (pass_slp_vectorize): Replace declaration with that of...
5756         (make_pass_slp_vectorize): ...new function.
5757         (pass_complete_unroll): Replace declaration with that of...
5758         (make_pass_complete_unroll): ...new function.
5759         (pass_complete_unrolli): Replace declaration with that of...
5760         (make_pass_complete_unrolli): ...new function.
5761         (pass_parallelize_loops): Replace declaration with that of...
5762         (make_pass_parallelize_loops): ...new function.
5763         (pass_loop_prefetch): Replace declaration with that of...
5764         (make_pass_loop_prefetch): ...new function.
5765         (pass_iv_optimize): Replace declaration with that of...
5766         (make_pass_iv_optimize): ...new function.
5767         (pass_tree_loop_done): Replace declaration with that of...
5768         (make_pass_tree_loop_done): ...new function.
5769         (pass_ch): Replace declaration with that of...
5770         (make_pass_ch): ...new function.
5771         (pass_ccp): Replace declaration with that of...
5772         (make_pass_ccp): ...new function.
5773         (pass_phi_only_cprop): Replace declaration with that of...
5774         (make_pass_phi_only_cprop): ...new function.
5775         (pass_build_ssa): Replace declaration with that of...
5776         (make_pass_build_ssa): ...new function.
5777         (pass_build_alias): Replace declaration with that of...
5778         (make_pass_build_alias): ...new function.
5779         (pass_build_ealias): Replace declaration with that of...
5780         (make_pass_build_ealias): ...new function.
5781         (pass_dominator): Replace declaration with that of...
5782         (make_pass_dominator): ...new function.
5783         (pass_dce): Replace declaration with that of...
5784         (make_pass_dce): ...new function.
5785         (pass_dce_loop): Replace declaration with that of...
5786         (make_pass_dce_loop): ...new function.
5787         (pass_cd_dce): Replace declaration with that of...
5788         (make_pass_cd_dce): ...new function.
5789         (pass_call_cdce): Replace declaration with that of...
5790         (make_pass_call_cdce): ...new function.
5791         (pass_merge_phi): Replace declaration with that of...
5792         (make_pass_merge_phi): ...new function.
5793         (pass_split_crit_edges): Replace declaration with that of...
5794         (make_pass_split_crit_edges): ...new function.
5795         (pass_pre): Replace declaration with that of...
5796         (make_pass_pre): ...new function.
5797         (pass_profile): Replace declaration with that of...
5798         (make_pass_profile): ...new function.
5799         (pass_strip_predict_hints): Replace declaration with that of...
5800         (make_pass_strip_predict_hints): ...new function.
5801         (pass_lower_complex_O0): Replace declaration with that of...
5802         (make_pass_lower_complex_O0): ...new function.
5803         (pass_lower_complex): Replace declaration with that of...
5804         (make_pass_lower_complex): ...new function.
5805         (pass_lower_vector): Replace declaration with that of...
5806         (make_pass_lower_vector): ...new function.
5807         (pass_lower_vector_ssa): Replace declaration with that of...
5808         (make_pass_lower_vector_ssa): ...new function.
5809         (pass_lower_omp): Replace declaration with that of...
5810         (make_pass_lower_omp): ...new function.
5811         (pass_diagnose_omp_blocks): Replace declaration with that of...
5812         (make_pass_diagnose_omp_blocks): ...new function.
5813         (pass_expand_omp): Replace declaration with that of...
5814         (make_pass_expand_omp): ...new function.
5815         (pass_expand_omp_ssa): Replace declaration with that of...
5816         (make_pass_expand_omp_ssa): ...new function.
5817         (pass_object_sizes): Replace declaration with that of...
5818         (make_pass_object_sizes): ...new function.
5819         (pass_strlen): Replace declaration with that of...
5820         (make_pass_strlen): ...new function.
5821         (pass_fold_builtins): Replace declaration with that of...
5822         (make_pass_fold_builtins): ...new function.
5823         (pass_stdarg): Replace declaration with that of...
5824         (make_pass_stdarg): ...new function.
5825         (pass_early_warn_uninitialized): Replace declaration with that of...
5826         (make_pass_early_warn_uninitialized): ...new function.
5827         (pass_late_warn_uninitialized): Replace declaration with that of...
5828         (make_pass_late_warn_uninitialized): ...new function.
5829         (pass_cse_reciprocals): Replace declaration with that of...
5830         (make_pass_cse_reciprocals): ...new function.
5831         (pass_cse_sincos): Replace declaration with that of...
5832         (make_pass_cse_sincos): ...new function.
5833         (pass_optimize_bswap): Replace declaration with that of...
5834         (make_pass_optimize_bswap): ...new function.
5835         (pass_optimize_widening_mul): Replace declaration with that of...
5836         (make_pass_optimize_widening_mul): ...new function.
5837         (pass_warn_function_return): Replace declaration with that of...
5838         (make_pass_warn_function_return): ...new function.
5839         (pass_warn_function_noreturn): Replace declaration with that of...
5840         (make_pass_warn_function_noreturn): ...new function.
5841         (pass_cselim): Replace declaration with that of...
5842         (make_pass_cselim): ...new function.
5843         (pass_phiopt): Replace declaration with that of...
5844         (make_pass_phiopt): ...new function.
5845         (pass_forwprop): Replace declaration with that of...
5846         (make_pass_forwprop): ...new function.
5847         (pass_phiprop): Replace declaration with that of...
5848         (make_pass_phiprop): ...new function.
5849         (pass_tree_ifcombine): Replace declaration with that of...
5850         (make_pass_tree_ifcombine): ...new function.
5851         (pass_dse): Replace declaration with that of...
5852         (make_pass_dse): ...new function.
5853         (pass_nrv): Replace declaration with that of...
5854         (make_pass_nrv): ...new function.
5855         (pass_rename_ssa_copies): Replace declaration with that of...
5856         (make_pass_rename_ssa_copies): ...new function.
5857         (pass_sink_code): Replace declaration with that of...
5858         (make_pass_sink_code): ...new function.
5859         (pass_fre): Replace declaration with that of...
5860         (make_pass_fre): ...new function.
5861         (pass_check_data_deps): Replace declaration with that of...
5862         (make_pass_check_data_deps): ...new function.
5863         (pass_copy_prop): Replace declaration with that of...
5864         (make_pass_copy_prop): ...new function.
5865         (pass_vrp): Replace declaration with that of...
5866         (make_pass_vrp): ...new function.
5867         (pass_uncprop): Replace declaration with that of...
5868         (make_pass_uncprop): ...new function.
5869         (pass_return_slot): Replace declaration with that of...
5870         (make_pass_return_slot): ...new function.
5871         (pass_reassoc): Replace declaration with that of...
5872         (make_pass_reassoc): ...new function.
5873         (pass_rebuild_cgraph_edges): Replace declaration with that of...
5874         (make_pass_rebuild_cgraph_edges): ...new function.
5875         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
5876         (make_pass_remove_cgraph_callee_edges): ...new function.
5877         (pass_build_cgraph_edges): Replace declaration with that of...
5878         (make_pass_build_cgraph_edges): ...new function.
5879         (pass_local_pure_const): Replace declaration with that of...
5880         (make_pass_local_pure_const): ...new function.
5881         (pass_tracer): Replace declaration with that of...
5882         (make_pass_tracer): ...new function.
5883         (pass_warn_unused_result): Replace declaration with that of...
5884         (make_pass_warn_unused_result): ...new function.
5885         (pass_diagnose_tm_blocks): Replace declaration with that of...
5886         (make_pass_diagnose_tm_blocks): ...new function.
5887         (pass_lower_tm): Replace declaration with that of...
5888         (make_pass_lower_tm): ...new function.
5889         (pass_tm_init): Replace declaration with that of...
5890         (make_pass_tm_init): ...new function.
5891         (pass_tm_mark): Replace declaration with that of...
5892         (make_pass_tm_mark): ...new function.
5893         (pass_tm_memopt): Replace declaration with that of...
5894         (make_pass_tm_memopt): ...new function.
5895         (pass_tm_edges): Replace declaration with that of...
5896         (make_pass_tm_edges): ...new function.
5897         (pass_split_functions): Replace declaration with that of...
5898         (make_pass_split_functions): ...new function.
5899         (pass_feedback_split_functions): Replace declaration with that of...
5900         (make_pass_feedback_split_functions): ...new function.
5901         (pass_strength_reduction): Replace declaration with that of...
5902         (make_pass_strength_reduction): ...new function.
5903         (pass_ipa_lower_emutls): Replace declaration with that of...
5904         (make_pass_ipa_lower_emutls): ...new function.
5905         (pass_ipa_function_and_variable_visibility): Replace declaration with
5906         that of...
5907         (make_pass_ipa_function_and_variable_visibility): ...new function.
5908         (pass_ipa_tree_profile): Replace declaration with that of...
5909         (make_pass_ipa_tree_profile): ...new function.
5910         (pass_early_local_passes): Replace declaration with that of...
5911         (make_pass_early_local_passes): ...new function.
5912         (pass_ipa_whole_program_visibility): Replace declaration with that
5913         of...
5914         (make_pass_ipa_whole_program_visibility): ...new function.
5915         (pass_ipa_lto_gimple_out): Replace declaration with that of...
5916         (make_pass_ipa_lto_gimple_out): ...new function.
5917         (pass_ipa_increase_alignment): Replace declaration with that of...
5918         (make_pass_ipa_increase_alignment): ...new function.
5919         (pass_ipa_inline): Replace declaration with that of...
5920         (make_pass_ipa_inline): ...new function.
5921         (pass_ipa_free_lang_data): Replace declaration with that of...
5922         (make_pass_ipa_free_lang_data): ...new function.
5923         (pass_ipa_free_inline_summary): Replace declaration with that of...
5924         (make_pass_ipa_free_inline_summary): ...new function.
5925         (pass_ipa_cp): Replace declaration with that of...
5926         (make_pass_ipa_cp): ...new function.
5927         (pass_ipa_reference): Replace declaration with that of...
5928         (make_pass_ipa_reference): ...new function.
5929         (pass_ipa_pure_const): Replace declaration with that of...
5930         (make_pass_ipa_pure_const): ...new function.
5931         (pass_ipa_pta): Replace declaration with that of...
5932         (make_pass_ipa_pta): ...new function.
5933         (pass_ipa_lto_finish_out): Replace declaration with that of...
5934         (make_pass_ipa_lto_finish_out): ...new function.
5935         (pass_ipa_tm): Replace declaration with that of...
5936         (make_pass_ipa_tm): ...new function.
5937         (pass_ipa_profile): Replace declaration with that of...
5938         (make_pass_ipa_profile): ...new function.
5939         (pass_ipa_cdtor_merge): Replace declaration with that of...
5940         (make_pass_ipa_cdtor_merge): ...new function.
5941         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
5942         of...
5943         (make_pass_cleanup_cfg_post_optimizing): ...new function.
5944         (pass_init_datastructures): Replace declaration with that of...
5945         (make_pass_init_datastructures): ...new function.
5946         (pass_fixup_cfg): Replace declaration with that of...
5947         (make_pass_fixup_cfg): ...new function.
5948         (pass_expand): Replace declaration with that of...
5949         (make_pass_expand): ...new function.
5950         (pass_instantiate_virtual_regs): Replace declaration with that of...
5951         (make_pass_instantiate_virtual_regs): ...new function.
5952         (pass_rtl_fwprop): Replace declaration with that of...
5953         (make_pass_rtl_fwprop): ...new function.
5954         (pass_rtl_fwprop_addr): Replace declaration with that of...
5955         (make_pass_rtl_fwprop_addr): ...new function.
5956         (pass_jump): Replace declaration with that of...
5957         (make_pass_jump): ...new function.
5958         (pass_jump2): Replace declaration with that of...
5959         (make_pass_jump2): ...new function.
5960         (pass_lower_subreg): Replace declaration with that of...
5961         (make_pass_lower_subreg): ...new function.
5962         (pass_cse): Replace declaration with that of...
5963         (make_pass_cse): ...new function.
5964         (pass_fast_rtl_dce): Replace declaration with that of...
5965         (make_pass_fast_rtl_dce): ...new function.
5966         (pass_ud_rtl_dce): Replace declaration with that of...
5967         (make_pass_ud_rtl_dce): ...new function.
5968         (pass_rtl_dce): Replace declaration with that of...
5969         (make_pass_rtl_dce): ...new function.
5970         (pass_rtl_dse1): Replace declaration with that of...
5971         (make_pass_rtl_dse1): ...new function.
5972         (pass_rtl_dse2): Replace declaration with that of...
5973         (make_pass_rtl_dse2): ...new function.
5974         (pass_rtl_dse3): Replace declaration with that of...
5975         (make_pass_rtl_dse3): ...new function.
5976         (pass_rtl_cprop): Replace declaration with that of...
5977         (make_pass_rtl_cprop): ...new function.
5978         (pass_rtl_pre): Replace declaration with that of...
5979         (make_pass_rtl_pre): ...new function.
5980         (pass_rtl_hoist): Replace declaration with that of...
5981         (make_pass_rtl_hoist): ...new function.
5982         (pass_rtl_store_motion): Replace declaration with that of...
5983         (make_pass_rtl_store_motion): ...new function.
5984         (pass_cse_after_global_opts): Replace declaration with that of...
5985         (make_pass_cse_after_global_opts): ...new function.
5986         (pass_rtl_ifcvt): Replace declaration with that of...
5987         (make_pass_rtl_ifcvt): ...new function.
5988         (pass_into_cfg_layout_mode): Replace declaration with that of...
5989         (make_pass_into_cfg_layout_mode): ...new function.
5990         (pass_outof_cfg_layout_mode): Replace declaration with that of...
5991         (make_pass_outof_cfg_layout_mode): ...new function.
5992         (pass_loop2): Replace declaration with that of...
5993         (make_pass_loop2): ...new function.
5994         (pass_rtl_loop_init): Replace declaration with that of...
5995         (make_pass_rtl_loop_init): ...new function.
5996         (pass_rtl_move_loop_invariants): Replace declaration with that of...
5997         (make_pass_rtl_move_loop_invariants): ...new function.
5998         (pass_rtl_unswitch): Replace declaration with that of...
5999         (make_pass_rtl_unswitch): ...new function.
6000         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
6001         (make_pass_rtl_unroll_and_peel_loops): ...new function.
6002         (pass_rtl_doloop): Replace declaration with that of...
6003         (make_pass_rtl_doloop): ...new function.
6004         (pass_rtl_loop_done): Replace declaration with that of...
6005         (make_pass_rtl_loop_done): ...new function.
6006         (pass_web): Replace declaration with that of...
6007         (make_pass_web): ...new function.
6008         (pass_cse2): Replace declaration with that of...
6009         (make_pass_cse2): ...new function.
6010         (pass_df_initialize_opt): Replace declaration with that of...
6011         (make_pass_df_initialize_opt): ...new function.
6012         (pass_df_initialize_no_opt): Replace declaration with that of...
6013         (make_pass_df_initialize_no_opt): ...new function.
6014         (pass_reginfo_init): Replace declaration with that of...
6015         (make_pass_reginfo_init): ...new function.
6016         (pass_inc_dec): Replace declaration with that of...
6017         (make_pass_inc_dec): ...new function.
6018         (pass_stack_ptr_mod): Replace declaration with that of...
6019         (make_pass_stack_ptr_mod): ...new function.
6020         (pass_initialize_regs): Replace declaration with that of...
6021         (make_pass_initialize_regs): ...new function.
6022         (pass_combine): Replace declaration with that of...
6023         (make_pass_combine): ...new function.
6024         (pass_if_after_combine): Replace declaration with that of...
6025         (make_pass_if_after_combine): ...new function.
6026         (pass_ree): Replace declaration with that of...
6027         (make_pass_ree): ...new function.
6028         (pass_partition_blocks): Replace declaration with that of...
6029         (make_pass_partition_blocks): ...new function.
6030         (pass_match_asm_constraints): Replace declaration with that of...
6031         (make_pass_match_asm_constraints): ...new function.
6032         (pass_regmove): Replace declaration with that of...
6033         (make_pass_regmove): ...new function.
6034         (pass_split_all_insns): Replace declaration with that of...
6035         (make_pass_split_all_insns): ...new function.
6036         (pass_fast_rtl_byte_dce): Replace declaration with that of...
6037         (make_pass_fast_rtl_byte_dce): ...new function.
6038         (pass_lower_subreg2): Replace declaration with that of...
6039         (make_pass_lower_subreg2): ...new function.
6040         (pass_mode_switching): Replace declaration with that of...
6041         (make_pass_mode_switching): ...new function.
6042         (pass_sms): Replace declaration with that of...
6043         (make_pass_sms): ...new function.
6044         (pass_sched): Replace declaration with that of...
6045         (make_pass_sched): ...new function.
6046         (pass_ira): Replace declaration with that of...
6047         (make_pass_ira): ...new function.
6048         (pass_reload): Replace declaration with that of...
6049         (make_pass_reload): ...new function.
6050         (pass_clean_state): Replace declaration with that of...
6051         (make_pass_clean_state): ...new function.
6052         (pass_branch_prob): Replace declaration with that of...
6053         (make_pass_branch_prob): ...new function.
6054         (pass_value_profile_transformations): Replace declaration with that
6055         of...
6056         (make_pass_value_profile_transformations): ...new function.
6057         (pass_postreload_cse): Replace declaration with that of...
6058         (make_pass_postreload_cse): ...new function.
6059         (pass_gcse2): Replace declaration with that of...
6060         (make_pass_gcse2): ...new function.
6061         (pass_split_after_reload): Replace declaration with that of...
6062         (make_pass_split_after_reload): ...new function.
6063         (pass_branch_target_load_optimize1): Replace declaration with that
6064         of...
6065         (make_pass_branch_target_load_optimize1): ...new function.
6066         (pass_thread_prologue_and_epilogue): Replace declaration with that
6067         of...
6068         (make_pass_thread_prologue_and_epilogue): ...new function.
6069         (pass_stack_adjustments): Replace declaration with that of...
6070         (make_pass_stack_adjustments): ...new function.
6071         (pass_peephole2): Replace declaration with that of...
6072         (make_pass_peephole2): ...new function.
6073         (pass_if_after_reload): Replace declaration with that of...
6074         (make_pass_if_after_reload): ...new function.
6075         (pass_regrename): Replace declaration with that of...
6076         (make_pass_regrename): ...new function.
6077         (pass_cprop_hardreg): Replace declaration with that of...
6078         (make_pass_cprop_hardreg): ...new function.
6079         (pass_reorder_blocks): Replace declaration with that of...
6080         (make_pass_reorder_blocks): ...new function.
6081         (pass_branch_target_load_optimize2): Replace declaration with that
6082         of...
6083         (make_pass_branch_target_load_optimize2): ...new function.
6084         (pass_leaf_regs): Replace declaration with that of...
6085         (make_pass_leaf_regs): ...new function.
6086         (pass_split_before_sched2): Replace declaration with that of...
6087         (make_pass_split_before_sched2): ...new function.
6088         (pass_compare_elim_after_reload): Replace declaration with that of...
6089         (make_pass_compare_elim_after_reload): ...new function.
6090         (pass_sched2): Replace declaration with that of...
6091         (make_pass_sched2): ...new function.
6092         (pass_stack_regs): Replace declaration with that of...
6093         (make_pass_stack_regs): ...new function.
6094         (pass_stack_regs_run): Replace declaration with that of...
6095         (make_pass_stack_regs_run): ...new function.
6096         (pass_df_finish): Replace declaration with that of...
6097         (make_pass_df_finish): ...new function.
6098         (pass_compute_alignments): Replace declaration with that of...
6099         (make_pass_compute_alignments): ...new function.
6100         (pass_duplicate_computed_gotos): Replace declaration with that of...
6101         (make_pass_duplicate_computed_gotos): ...new function.
6102         (pass_variable_tracking): Replace declaration with that of...
6103         (make_pass_variable_tracking): ...new function.
6104         (pass_free_cfg): Replace declaration with that of...
6105         (make_pass_free_cfg): ...new function.
6106         (pass_machine_reorg): Replace declaration with that of...
6107         (make_pass_machine_reorg): ...new function.
6108         (pass_cleanup_barriers): Replace declaration with that of...
6109         (make_pass_cleanup_barriers): ...new function.
6110         (pass_delay_slots): Replace declaration with that of...
6111         (make_pass_delay_slots): ...new function.
6112         (pass_split_for_shorten_branches): Replace declaration with that of...
6113         (make_pass_split_for_shorten_branches): ...new function.
6114         (pass_split_before_regstack): Replace declaration with that of...
6115         (make_pass_split_before_regstack): ...new function.
6116         (pass_convert_to_eh_region_ranges): Replace declaration with that
6117         of...
6118         (make_pass_convert_to_eh_region_ranges): ...new function.
6119         (pass_shorten_branches): Replace declaration with that of...
6120         (make_pass_shorten_branches): ...new function.
6121         (pass_set_nothrow_function_flags): Replace declaration with that of...
6122         (make_pass_set_nothrow_function_flags): ...new function.
6123         (pass_dwarf2_frame): Replace declaration with that of...
6124         (make_pass_dwarf2_frame): ...new function.
6125         (pass_final): Replace declaration with that of...
6126         (make_pass_final): ...new function.
6127         (pass_rtl_seqabstr): Replace declaration with that of...
6128         (make_pass_rtl_seqabstr): ...new function.
6129         (pass_release_ssa_names): Replace declaration with that of...
6130         (make_pass_release_ssa_names): ...new function.
6131         (pass_early_inline): Replace declaration with that of...
6132         (make_pass_early_inline): ...new function.
6133         (pass_inline_parameters): Replace declaration with that of...
6134         (make_pass_inline_parameters): ...new function.
6135         (pass_update_address_taken): Replace declaration with that of...
6136         (make_pass_update_address_taken): ...new function.
6137         (pass_convert_switch): Replace declaration with that of...
6138         (make_pass_convert_switch): ...new function.
6139         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
6140         to a subclass of simple_ipa_opt_pass along with...
6141         (pass_data_ipa_tree_profile): ...new pass_data instance and...
6142         (make_pass_ipa_tree_profile): ...new function.
6143         * tree-sra.c (pass_sra_early): Convert from a global struct to a
6144         subclass of gimple_opt_pass along with...
6145         (pass_data_sra_early): ...new pass_data instance and...
6146         (make_pass_sra_early): ...new function.
6147         (pass_sra): Convert from a global struct to a subclass of
6148         gimple_opt_pass along with...
6149         (pass_data_sra): ...new pass_data instance and...
6150         (make_pass_sra): ...new function.
6151         (pass_early_ipa_sra): Convert from a global struct to a subclass of
6152         gimple_opt_pass along with...
6153         (pass_data_early_ipa_sra): ...new pass_data instance and...
6154         (make_pass_early_ipa_sra): ...new function.
6155         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
6156         subclass of gimple_opt_pass along with...
6157         (pass_data_ccp): ...new pass_data instance and...
6158         (make_pass_ccp): ...new function.
6159         (pass_fold_builtins): Convert from a global struct to a subclass of
6160         gimple_opt_pass along with...
6161         (pass_data_fold_builtins): ...new pass_data instance and...
6162         (make_pass_fold_builtins): ...new function.
6163         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
6164         subclass of gimple_opt_pass along with...
6165         (pass_data_copy_prop): ...new pass_data instance and...
6166         (make_pass_copy_prop): ...new function.
6167         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
6168         global struct to a subclass of gimple_opt_pass along with...
6169         (pass_data_rename_ssa_copies): ...new pass_data instance and...
6170         (make_pass_rename_ssa_copies): ...new function.
6171         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
6172         subclass of gimple_opt_pass along with...
6173         (pass_data_dce): ...new pass_data instance and...
6174         (make_pass_dce): ...new function.
6175         (pass_dce_loop): Convert from a global struct to a subclass of
6176         gimple_opt_pass along with...
6177         (pass_data_dce_loop): ...new pass_data instance and...
6178         (make_pass_dce_loop): ...new function.
6179         (pass_cd_dce): Convert from a global struct to a subclass of
6180         gimple_opt_pass along with...
6181         (pass_data_cd_dce): ...new pass_data instance and...
6182         (make_pass_cd_dce): ...new function.
6183         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
6184         subclass of gimple_opt_pass along with...
6185         (pass_data_dominator): ...new pass_data instance and...
6186         (make_pass_dominator): ...new function.
6187         (pass_phi_only_cprop): Convert from a global struct to a subclass of
6188         gimple_opt_pass along with...
6189         (pass_data_phi_only_cprop): ...new pass_data instance and...
6190         (make_pass_phi_only_cprop): ...new function.
6191         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
6192         subclass of gimple_opt_pass along with...
6193         (pass_data_dse): ...new pass_data instance and...
6194         (make_pass_dse): ...new function.
6195         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
6196         a subclass of gimple_opt_pass along with...
6197         (pass_data_forwprop): ...new pass_data instance and...
6198         (make_pass_forwprop): ...new function.
6199         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
6200         struct to a subclass of gimple_opt_pass along with...
6201         (pass_data_tree_ifcombine): ...new pass_data instance and...
6202         (make_pass_tree_ifcombine): ...new function.
6203         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
6204         subclass of gimple_opt_pass along with...
6205         (pass_data_ch): ...new pass_data instance and...
6206         (make_pass_ch): ...new function.
6207         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
6208         subclass of gimple_opt_pass along with...
6209         (pass_data_tree_loop): ...new pass_data instance and...
6210         (make_pass_tree_loop): ...new function.
6211         (pass_tree_loop_init): Convert from a global struct to a subclass of
6212         gimple_opt_pass along with...
6213         (pass_data_tree_loop_init): ...new pass_data instance and...
6214         (make_pass_tree_loop_init): ...new function.
6215         (pass_lim): Convert from a global struct to a subclass of
6216         gimple_opt_pass along with...
6217         (pass_data_lim): ...new pass_data instance and...
6218         (make_pass_lim): ...new function.
6219         (pass_tree_unswitch): Convert from a global struct to a subclass of
6220         gimple_opt_pass along with...
6221         (pass_data_tree_unswitch): ...new pass_data instance and...
6222         (make_pass_tree_unswitch): ...new function.
6223         (pass_predcom): Convert from a global struct to a subclass of
6224         gimple_opt_pass along with...
6225         (pass_data_predcom): ...new pass_data instance and...
6226         (make_pass_predcom): ...new function.
6227         (pass_vectorize): Convert from a global struct to a subclass of
6228         gimple_opt_pass along with...
6229         (pass_data_vectorize): ...new pass_data instance and...
6230         (make_pass_vectorize): ...new function.
6231         (pass_graphite): Convert from a global struct to a subclass of
6232         gimple_opt_pass along with...
6233         (pass_data_graphite): ...new pass_data instance and...
6234         (make_pass_graphite): ...new function.
6235         (pass_graphite_transforms): Convert from a global struct to a subclass
6236         of gimple_opt_pass along with...
6237         (pass_data_graphite_transforms): ...new pass_data instance and...
6238         (make_pass_graphite_transforms): ...new function.
6239         (pass_check_data_deps): Convert from a global struct to a subclass of
6240         gimple_opt_pass along with...
6241         (pass_data_check_data_deps): ...new pass_data instance and...
6242         (make_pass_check_data_deps): ...new function.
6243         (pass_iv_canon): Convert from a global struct to a subclass of
6244         gimple_opt_pass along with...
6245         (pass_data_iv_canon): ...new pass_data instance and...
6246         (make_pass_iv_canon): ...new function.
6247         (pass_scev_cprop): Convert from a global struct to a subclass of
6248         gimple_opt_pass along with...
6249         (pass_data_scev_cprop): ...new pass_data instance and...
6250         (make_pass_scev_cprop): ...new function.
6251         (pass_record_bounds): Convert from a global struct to a subclass of
6252         gimple_opt_pass along with...
6253         (pass_data_record_bounds): ...new pass_data instance and...
6254         (make_pass_record_bounds): ...new function.
6255         (pass_complete_unroll): Convert from a global struct to a subclass of
6256         gimple_opt_pass along with...
6257         (pass_data_complete_unroll): ...new pass_data instance and...
6258         (make_pass_complete_unroll): ...new function.
6259         (pass_complete_unrolli): Convert from a global struct to a subclass of
6260         gimple_opt_pass along with...
6261         (pass_data_complete_unrolli): ...new pass_data instance and...
6262         (make_pass_complete_unrolli): ...new function.
6263         (pass_parallelize_loops): Convert from a global struct to a subclass
6264         of gimple_opt_pass along with...
6265         (pass_data_parallelize_loops): ...new pass_data instance and...
6266         (make_pass_parallelize_loops): ...new function.
6267         (pass_loop_prefetch): Convert from a global struct to a subclass of
6268         gimple_opt_pass along with...
6269         (pass_data_loop_prefetch): ...new pass_data instance and...
6270         (make_pass_loop_prefetch): ...new function.
6271         (pass_iv_optimize): Convert from a global struct to a subclass of
6272         gimple_opt_pass along with...
6273         (pass_data_iv_optimize): ...new pass_data instance and...
6274         (make_pass_iv_optimize): ...new function.
6275         (pass_tree_loop_done): Convert from a global struct to a subclass of
6276         gimple_opt_pass along with...
6277         (pass_data_tree_loop_done): ...new pass_data instance and...
6278         (make_pass_tree_loop_done): ...new function.
6279         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
6280         struct to a subclass of gimple_opt_pass along with...
6281         (pass_data_cse_reciprocals): ...new pass_data instance and...
6282         (make_pass_cse_reciprocals): ...new function.
6283         (pass_cse_sincos): Convert from a global struct to a subclass of
6284         gimple_opt_pass along with...
6285         (pass_data_cse_sincos): ...new pass_data instance and...
6286         (make_pass_cse_sincos): ...new function.
6287         (pass_optimize_bswap): Convert from a global struct to a subclass of
6288         gimple_opt_pass along with...
6289         (pass_data_optimize_bswap): ...new pass_data instance and...
6290         (make_pass_optimize_bswap): ...new function.
6291         (pass_optimize_widening_mul): Convert from a global struct to a
6292         subclass of gimple_opt_pass along with...
6293         (pass_data_optimize_widening_mul): ...new pass_data instance and...
6294         (make_pass_optimize_widening_mul): ...new function.
6295         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
6296         subclass of gimple_opt_pass along with...
6297         (pass_data_phiopt): ...new pass_data instance and...
6298         (make_pass_phiopt): ...new function.
6299         (pass_cselim): Convert from a global struct to a subclass of
6300         gimple_opt_pass along with...
6301         (pass_data_cselim): ...new pass_data instance and...
6302         (make_pass_cselim): ...new function.
6303         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
6304         subclass of gimple_opt_pass along with...
6305         (pass_data_phiprop): ...new pass_data instance and...
6306         (make_pass_phiprop): ...new function.
6307         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
6308         subclass of gimple_opt_pass along with...
6309         (pass_data_pre): ...new pass_data instance and...
6310         (make_pass_pre): ...new function.
6311         (pass_fre): Convert from a global struct to a subclass of
6312         gimple_opt_pass along with...
6313         (pass_data_fre): ...new pass_data instance and...
6314         (make_pass_fre): ...new function.
6315         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
6316         subclass of gimple_opt_pass along with...
6317         (pass_data_reassoc): ...new pass_data instance and...
6318         (make_pass_reassoc): ...new function.
6319         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
6320         subclass of gimple_opt_pass along with...
6321         (pass_data_sink_code): ...new pass_data instance and...
6322         (make_pass_sink_code): ...new function.
6323         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
6324         subclass of gimple_opt_pass along with...
6325         (pass_data_strlen): ...new pass_data instance and...
6326         (make_pass_strlen): ...new function.
6327         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
6328         struct to a subclass of gimple_opt_pass along with...
6329         (pass_data_build_alias): ...new pass_data instance and...
6330         (make_pass_build_alias): ...new function.
6331         (pass_build_ealias): Convert from a global struct to a subclass of
6332         gimple_opt_pass along with...
6333         (pass_data_build_ealias): ...new pass_data instance and...
6334         (make_pass_build_ealias): ...new function.
6335         (pass_ipa_pta): Convert from a global struct to a subclass of
6336         simple_ipa_opt_pass along with...
6337         (pass_data_ipa_pta): ...new pass_data instance and...
6338         (make_pass_ipa_pta): ...new function.
6339         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
6340         subclass of gimple_opt_pass along with...
6341         (pass_data_uncprop): ...new pass_data instance and...
6342         (make_pass_uncprop): ...new function.
6343         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
6344         global struct to a subclass of gimple_opt_pass along with...
6345         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
6346         (make_pass_late_warn_uninitialized): ...new function.
6347         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
6348         to a subclass of gimple_opt_pass along with...
6349         (pass_data_init_datastructures): ...new pass_data instance and...
6350         (make_pass_init_datastructures): ...new function.
6351         (pass_early_warn_uninitialized): Convert from a global struct to a
6352         subclass of gimple_opt_pass along with...
6353         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
6354         (make_pass_early_warn_uninitialized): ...new function.
6355         (pass_update_address_taken): Convert from a global struct to a
6356         subclass of gimple_opt_pass along with...
6357         (pass_data_update_address_taken): ...new pass_data instance and...
6358         (make_pass_update_address_taken): ...new function.
6359         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
6360         struct to a subclass of gimple_opt_pass along with...
6361         (pass_data_release_ssa_names): ...new pass_data instance and...
6362         (make_pass_release_ssa_names): ...new function.
6363         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
6364         subclass of gimple_opt_pass along with...
6365         (pass_data_stdarg): ...new pass_data instance and...
6366         (make_pass_stdarg): ...new function.
6367         * tree-switch-conversion.c (pass_convert_switch): Convert from a
6368         global struct to a subclass of gimple_opt_pass along with...
6369         (pass_data_convert_switch): ...new pass_data instance and...
6370         (make_pass_convert_switch): ...new function.
6371         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
6372         to a subclass of gimple_opt_pass along with...
6373         (pass_data_tail_recursion): ...new pass_data instance and...
6374         (make_pass_tail_recursion): ...new function.
6375         (pass_tail_calls): Convert from a global struct to a subclass of
6376         gimple_opt_pass along with...
6377         (pass_data_tail_calls): ...new pass_data instance and...
6378         (make_pass_tail_calls): ...new function.
6379         * tree-vect-generic.c (pass_lower_vector): Convert from a global
6380         struct to a subclass of gimple_opt_pass along with...
6381         (pass_data_lower_vector): ...new pass_data instance and...
6382         (make_pass_lower_vector): ...new function.
6383         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
6384         gimple_opt_pass along with...
6385         (pass_data_lower_vector_ssa): ...new pass_data instance and...
6386         (make_pass_lower_vector_ssa): ...new function.
6387         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
6388         to a subclass of gimple_opt_pass along with...
6389         (pass_data_slp_vectorize): ...new pass_data instance and...
6390         (make_pass_slp_vectorize): ...new function.
6391         (pass_ipa_increase_alignment): Convert from a global struct to a
6392         subclass of simple_ipa_opt_pass along with...
6393         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
6394         (make_pass_ipa_increase_alignment): ...new function.
6395         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
6396         gimple_opt_pass along with...
6397         (pass_data_vrp): ...new pass_data instance and...
6398         (make_pass_vrp): ...new function.
6399         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
6400         subclass of simple_ipa_opt_pass along with...
6401         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
6402         (make_pass_ipa_free_lang_data): ...new function.
6403         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
6404         gimple_opt_pass along with...
6405         (pass_data_tsan): ...new pass_data instance and...
6406         (make_pass_tsan): ...new function.
6407         (pass_tsan_O0): Convert from a global struct to a subclass of
6408         gimple_opt_pass along with...
6409         (pass_data_tsan_O0): ...new pass_data instance and...
6410         (make_pass_tsan_O0): ...new function.
6411         * var-tracking.c (pass_variable_tracking): Convert from a global
6412         struct to a subclass of rtl_opt_pass along with...
6413         (pass_data_variable_tracking): ...new pass_data instance and...
6414         (make_pass_variable_tracking): ...new function.
6415         * web.c (pass_web): Convert from a global struct to a subclass of
6416         rtl_opt_pass along with...
6417         (pass_data_web): ...new pass_data instance and...
6418         (make_pass_web): ...new function.
6419         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
6420         declaration with that of...
6421         (make_pass_mode_switch_use): ...new function.
6422         (pass_resolve_sw_modes): Replace declaration with that of...
6423         (make_pass_resolve_sw_modes): ...new function.
6424         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
6425         from a global struct to a subclass of rtl_opt_pass along with...
6426         (pass_data_mode_switch_use): ...new pass_data instance and...
6427         (make_pass_mode_switch_use): ...new function.
6428         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
6429         from a global struct to a subclass of rtl_opt_pass along with...
6430         (pass_data_resolve_sw_modes): ...new pass_data instance and...
6431         (make_pass_resolve_sw_modes): ...new function.
6432         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
6433         struct to a subclass of rtl_opt_pass along with...
6434         (pass_data_insert_vzeroupper): ...new pass_data instance and...
6435         (make_pass_insert_vzeroupper): ...new function.
6436         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
6437         global struct to a subclass of rtl_opt_pass along with...
6438         (pass_data_work_around_errata): ...new pass_data instance and...
6439         (make_pass_work_around_errata): ...new function.
6440         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
6441         struct to a subclass of rtl_opt_pass along with...
6442         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
6443         (make_pass_mips_machine_reorg2): ...new function.
6445 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
6447         * passes.c (pass_manager::operator new): New.
6449 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
6451         Handwritten part of conversion of passes to C++ classes.
6453         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
6454         (toplev.o): Add dep on PASS_MANAGER_H.
6455         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
6456         of early local pases to reflect this moving from a global to a
6457         member of gcc::pass_manager.
6458         (cgraph_add_new_function): Likewise.
6459         * lto-cgraph.c (lto_output_node): Update for conversion of
6460         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
6461         * passes.c (opt_pass::clone): New.
6462         (opt_pass::gate): New.
6463         (opt_pass::execute): New.
6464         (opt_pass::opt_pass): New.
6465         (pass_manager::execute_early_local_passes): New.
6466         (pass_manager::execute_pass_mode_switching): new.
6467         (finish_optimization_passes): Convert to...
6468         (pass_manager::finish_optimization_passes): ...this.
6469         (finish_optimization_passes): Update for conversion of passes to
6470         C++ classes.
6471         (register_dump_files_1): Use has_gate since we cannot portably
6472         check a vtable entry against NULL.
6473         (dump_one_pass): Likewise.
6474         (ipa_write_summaries_2): Likewise.
6475         (ipa_write_optimization_summaries_1): Likewise.
6476         (ipa_read_summaries_1): Likewise.
6477         (ipa_read_optimization_summaries_1): Likewise.
6478         (execute_ipa_stmt_fixups): Likewise.
6479         (pass_manager::pass_manager): Rewrite pass-creation, invoking
6480         pass-creation functions rather than wiring up globals, and
6481         storing the results in fields of pass_manager generated using
6482         pass-instances.def.
6483         (pass_manager::dump_profile_report): Update for conversion of
6484         passes to C++ classes.
6485         (pass_manager::execute_ipa_summary_passes): Likewise.
6486         (execute_one_ipa_transform_pass): Likewise.
6487         (execute_one_pass): Use has_gate and has_execute since we cannot
6488         portably check a vtable entry against NULL.
6489         * pass_manager.h (pass_manager::finish_optimization_passes): New.
6490         (pass_manager): Use pass-instances.def to add fields for the
6491         various pass instances.
6492         * toplev.c (finalize): Update for move of
6493         finish_optimization_passes to a method of gcc::pass_manager.
6494         * toplev.h (finish_optimization_passes): Move to method of class
6495         pass_manager.
6496         * tree-pass.h (struct pass_data): New.
6497         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
6498         (opt_pass::gate): Convert to virtual function.
6499         (opt_pass::~opt_pass): New.
6500         (opt_pass::clone): New.
6501         (opt_pass::execute): Convert to virtual function.
6502         (opt_pass::opt_pass): New.
6503         (opt_pass::ctxt_): new.
6504         (gimple_opt_pass): Convert to subclass of opt_pass.
6505         (gimple_opt_pass::gimple_opt_pass): New.
6506         (rtl_opt_pass): Convert to subclass of opt_pass.
6507         (rtl_opt_pass::rtl_opt_pass): New.
6508         (ipa_opt_pass_d): Convert to subclass of opt_pass.
6509         (ipa_opt_pass_d::ipa_opt_pass_d): New.
6510         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
6511         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
6512         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
6513         invocation of pass_mode_switching to reflect this moving from a
6514         global to a member of gcc::pass_manager.
6515         (ix86_option_override): Rework how pass_insert_vzeroupper is
6516         added to the pass_manager to reflect autogenerated changes.
6517         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
6519 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
6521         PR rtl-optimization/57708
6522         * recog.c (peep2_find_free_register): Validate all regs in a
6523         multi-reg mode.
6525 2013-08-05  Jan Hubicka  <jh@suse.cz>
6527         PR lto/57602
6528         * cgraph.c (verify_cgraph_node): Accept local flags from other
6529         partitions.
6530         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
6531         (function_and_variable_visibility): Likewise.
6532         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
6534 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
6536         * graph.c (init_graph_slim_pretty_print): Remove.
6537         (print_graph_cfg): Do not call it.  Use local pretty printer.
6538         (start_graph_dump): Likewise.
6540 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
6542         * gimple-pretty-print.c (buffer): Remove.
6543         (initialized): Likewise.
6544         (maybe_init_pretty_print): Likewise.
6545         (print_gimple_stmt): Do not call it.  Use non-static local
6546         pretty_printer variable.
6547         (print_gimple_expr): Likewise.
6548         (print_gimple_seq): Likewise.
6549         (gimple_dump_bb): Likewise.
6551 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
6553         * asan.c (asan_pp): Remove.
6554         (asan_pp_initialized): Likewise.
6555         (asan_pp_initialize): Likewise.
6556         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
6557         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
6558         (asan_add_global): Likewise.
6560 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
6562         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
6563         * diagnostic.h (diagnostic_flush_buffer): Adjust.
6564         * pretty-print.c (pp_formatted_text_data): Likewise.
6565         (pp_indent): Rename from pp_base_indent.
6566         (pp_format): Rename from pp_base_format.
6567         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
6568         (pp_format_verbatim): Rename from pp_base_format_verbatim.
6569         (pp_flush): Rename from pp_base_flush.
6570         (pp_set_line_maximum_length): Rename from
6571         pp_base_set_line_maximum_length.
6572         (pp_clear_output_area): Rename from pp_base_clear_output_area.
6573         (pp_set_prefix): Rename from pp_base_set_prefix.
6574         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
6575         (pp_emit_prefix): Rename from pp_base_emit_prefix.
6576         (pp_append_text): Rename from pp_base_append_text.
6577         (pp_formatted_text): Rename from pp_base_formatted_text.
6578         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
6579         (pp_remaining_character_count_for_line): Rename from
6580         pp_base_remaining_character_count_for_line.
6581         (pp_newline): Rename from pp_base_newline.
6582         (pp_character): Rename from pp_base_character.
6583         (pp_string): Rename from pp_base_string.
6584         (pp_maybe_space): Rename from pp_base_maybe_space.
6585         * asan.c (asan_pp_string): Adjust.
6586         (asan_emit_stack_protection): Likewise.
6587         (asan_add_global): Likewise.
6588         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
6589         * tree-mudflap.c (mf_varname_tree): Likewise.
6590         * tree-pretty-print.c (pp_tree_identifier): Rename from
6591         pp_base_tree_identifier.
6592         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
6593         Declare as function.
6595 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
6597         * pretty-print.h (pp_bar_bar): New.
6598         (pp_ampersand_ampersand): Likewise.
6599         (pp_less_equal): Likewise.
6600         (pp_greater_equal): Likewise.
6601         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
6602         printer functions instead of pp_string or operators and punctuators.
6603         (dump_gimple_call): Likewise.
6604         (dump_gimple_omp_for): Likewise.
6605         (dump_gimple_transaction): Likewise.
6606         (dump_gimple_phi): Likewise.
6607         (pp_gimple_stmt_1): Likewise.
6608         * sched-vis.c (print_insn): Likewise.
6609         * tree-mudflap.c (mf_varname_tree): Likewise.
6610         * tree-pretty-print.c (dump_block_node): Likewise.
6611         (dump_generic_node): Likewise.
6613 2013-08-02  Jan Hubicka  <jh@suse.cz>
6615         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
6616         boundaries.
6617         * lto-streamer-out.c (tree_is_indexable): Results decls and
6618         parm decls are not indexable.
6619         (DFS_write_tree_body): Do not follow args and results.
6620         (hash_tree): Likewise.
6621         (output_functions): Rearrange so struct function is needed
6622         only when real body is output; be able to also ouptut abstract
6623         functions; output DECL_ARGUMENTS and DECL_RESULT.
6624         (lto_output): When not in WPA, ale store abstract functions.
6625         (write_symbol): Do not care about RESULT_DECL.
6626         (output_symbol_p): Handle correctly sbtract decls.
6627         * lto-streamer-in.c (input_function): Rearrange so struct
6628         function can be NULL at entry; allow streaming of
6629         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
6630         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
6631         sanity check during LTO.
6632         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
6633         RESULT_DECl and DECL_ARGUMENTS.
6634         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
6635         Likewise.
6637 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
6639         * pretty-print.h (pp_underscore): New.
6640         (pp_comma): Tidy.
6641         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
6642         printer functions instead of pp_character.
6643         (dump_binary_rhs): Likewise.
6644         (dump_ternary_rhs): Likewise.
6645         (dump_gimple_call_args): Likewise.
6646         (pp_points_to_solution): Likewise.
6647         (dump_gimple_call): Likewise.
6648         (dump_gimple_switch): Likewise.
6649         (dump_gimple_cond): Likewise.
6650         (dump_gimple_bind): Likewise.
6651         (dump_gimple_try): Likewise.
6652         (dump_gimple_omp_for): Likewise.
6653         (dump_gimple_omp_continue): Likewise.
6654         (dump_gimple_omp_single): Likewise.
6655         (dump_gimple_omp_sections): Likewise.
6656         (dump_gimple_omp_block): Likewise.
6657         (dump_gimple_omp_critical): Likewise.
6658         (dump_gimple_transaction): Likewise.
6659         (dump_gimple_asm): Likewise.
6660         (dump_gimple_phi): Likewise.
6661         (dump_gimple_omp_parallel): Likewise.
6662         (dump_gimple_omp_task): Likewise.
6663         (dump_gimple_omp_atomic_load): Likewise.
6664         (dump_gimple_omp_atomic_store): Likewise.
6665         (dump_gimple_mem_ops): Likewise.
6666         (pp_gimple_stmt_1): Likewise.
6667         (pp_cfg_jump): Likewise.
6668         (dump_implicit_edges): Likewise.
6669         (gimple_dump_bb_for_graph): Likewise.
6670         * graph.c (draw_cfg_node): Likewise.
6671         * langhooks.c (lhd_print_error_function): Likewise.
6672         * sched-vis.c (print_exp): Likewise.
6673         (print_value): Likewise.
6674         (print_pattern): Likewise.
6675         (print_insn): Likewise.
6676         (rtl_dump_bb_for_graph): Likewise.
6677         * tree-pretty-print.c (dump_function_declaration): Likewise.
6678         (dump_array_domain): Likewise.
6679         (dump_omp_clause): Likewise.
6680         (dump_location): Likewise.
6681         (dump_generic_node): Likewise.
6682         (print_struct_decl): Likewise.
6683         * diagnostic.c (diagnostic_show_locus): Use pp_space.
6685 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
6687         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
6688         candidate table when replacing a candidate statement.
6689         (replace_rhs_if_not_dup): Likewise.
6690         (replace_one_candidate): Likewise.
6692 2013-08-02  Jan Hubicka  <jh@suse.cz>
6693             Martin Liska  <marxin.liska@gmail.com>
6695         * cgraphunit.c (add_new_function): Fix logic when adding from
6696         late IPA pass.
6697         (assemble_thunk): Rename to ...
6698         (expand_thunk); .. this one; export; get it working with
6699         general functions; make produced gimple valid.
6700         * cgraph.h (expand_thunk): Declare.
6702 2013-08-02  Jan Hubicka  <jh@suse.cz>
6704         * ipa-cp.c (gather_context_independent_values): Use
6705         ipa_get_param_move_cost.
6706         (get_replacement_map): Remove PARAM; move parameter folding
6707         into tree-inline.c
6708         (create_specialized_node): Update.
6709         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
6710         assert that we have gimple body; update move_cost.
6711         (count_formal_params): Assert that we have gimple body.
6712         (ipa_dump_param): New function.
6713         (ipa_alloc_node_params): Break out from ...
6714         (ipa_initialize_node_params): ... here.
6715         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
6716         (ipa_write_node_info): Stream move costs.
6717         (ipa_read_node_info): Read move costs.
6718         (ipa_update_after_lto_read): Do not recompute node params.
6719         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
6720         (ipa_get_param): Check we are not in WPA.
6721         (ipa_get_param_move_cost): New.
6722         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
6723         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
6724         parm numbers to be present.
6726 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
6728         PR rtl-optimization/58048
6729         * lra-constraints.c (process_alt_operands): Don't check asm
6730         operand on register.
6732 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
6734         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
6735         the implied StoreLoad barrier for atomic operations if before.
6737 2013-08-02  Jan Hubicka  <jh@suse.cz>
6738             Martin Liska  <marxin.liska@gmail.com>
6740         * cgraph.c (cgraph_function_body_availability): Do not check
6741         cgraph flags.
6742         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
6743         symtab_node_availability): Declare.
6744         * ipa.c (can_replace_by_local_alias): New.
6745         (function_and_variable_visibility): Use it.
6746         * symtab.c (symtab_for_node_and_aliases,
6747         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
6749 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
6751         PR rtl-optimization/57963
6752         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
6753         (lra_constraints): Use them.
6755 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
6757         * config/arm/types.md (define_attr "type"): Add "load_acq"
6758         and "store_rel".
6759         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
6760         changes.
6761         (cortex_a53_store1): Likewise.
6763 2013-08-01  Jan Hubicka  <jh@suse.cz>
6765         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
6766         partitions are not needed.
6768 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
6770         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
6771         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
6772         MAYBE_NON_Q_CLASS_P where appropriate.
6774 2013-08-01  Jan Hubicka  <jh@suse.cz>
6776         * cgraph.h (release_function_body): Declare.
6777         * tree.c (free_lang_data_in_decl): Free, parameters and return values
6778         of unused delcarations.
6780 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6782         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
6783         RTL form when subtracting a constant.
6785 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6787         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
6788         Generate canonical plus rtx with negated immediate instead of minus
6789         where appropriate.
6790         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
6792 2013-08-01  Jan Hubicka  <jh@suse.cz>
6794         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
6795         (cgraph_release_function_body): Likewise.
6796         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
6797         * cgraph.h (cgrpah_node): Rename abstract_and_needed
6798         to used_as_abstract_origin.
6799         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
6800         symbols used as abstract origins.
6801         * cgraphunit.c (analyze_functions): Update.
6802         * ipa.c (symtab_remove_unreachable_nodes): Recompute
6803         used_as_abstract_origin.
6804         * tree-inline.c (tree_function_versioning): Update
6805         used_as_abstract_origin; be ready for DECL_RESULT and
6806         DECL_ARGUMENTS to be NULL.
6808         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
6809         for abstract functions.
6810         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
6811         real symbols.
6813 2013-08-01  Jan Hubicka  <jh@suse.cz>
6815         * profile.c (compute_value_histograms): Fix thinko.
6817 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
6819         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
6820         aarch-common-protos.h to extra_headers.
6821         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
6822         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
6823         * config/aarch64/t-aarch64 (aarch-common.o): Define.
6825 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
6827         * config/aarch64/aarch64.md (define_attr "type"): Delete.
6828         Include "../arm/types.md".  Define "type" attribute for all patterns.
6829         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
6830         attribute changes.
6832 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
6834         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
6835         to support power8 load fusion.
6836         (fusion_gpr_mem_load): Likewise.
6838         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
6840         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
6841         declarations for power8 load fusion.
6842         (emit_fusion_gpr_load): Likewise.
6844         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
6845         tuning for power8, turn on fusion mode by default.  Turn on sign
6846         extending fusion mode if normal fusion mode is on, and we are at
6847         -O2 or -O3.
6848         (fusion_gpr_load_p): New function, return true if we can fuse an
6849         addis instruction with a dependent load to a GPR.
6850         (emit_fusion_gpr_load): Emit the instructions for power8 load
6851         fusion to GPRs.
6853         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
6854         (VSX load fusion peepholes): New peepholes to fuse together an
6855         addi instruction with a VSX load instruction.
6857         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
6858         peepholes to fuse an addis instruction with a load to a GPR base
6859         register.  If we are supporting sign extending fusions, convert
6860         sign extending loads to zero extending loads and add an explicit
6861         sign extension.
6863 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
6865         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
6866         aarch-common-protos.h to extra_headers.
6867         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
6868         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
6869         (arm_early_store_addr_dep): Likewise.
6870         (arm_no_early_alu_shift_dep): Likewise.
6871         (arm_no_early_alu_shift_value_dep): Likewise.
6872         (arm_no_early_mul_dep): Likewise.
6873         (arm_no_early_store_addr_dep): Likewise.
6874         (arm_mac_accumulator_is_mul_result): Likewise.
6875         (arm_mac_accumulator_is_result): Likewise.
6876         * config/arm/aarch-common.c: ... here.  New file.
6877         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
6878         here to ...
6879         (arm_early_store_addr_dep): Likewise.
6880         (arm_no_early_alu_shift_dep): Likewise.
6881         (arm_no_early_alu_shift_value_dep): Likewise.
6882         (arm_no_early_mul_dep): Likewise.
6883         (arm_no_early_store_addr_dep): Likewise.
6884         (arm_mac_accumulator_is_mul_result): Likewise.
6885         (arm_mac_accumulator_is_result): Likewise.
6886         * config/arm/aarch-common-protos.h: ... here.  New file.
6887         * config/arm/t-arm (aarch-common.o): Define.
6889 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
6891         * config/arm/arm.md: Include new file "types.md".
6892         (define_attr "type"): Move from here to ...
6893         (define_attr "mul32"): Likewise.
6894         (define_attr "mul64"): Likewise.
6895         * config/arm/types.md: ... here.  New file.
6897 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
6899         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
6900         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
6902 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
6904         * gen-pass-instances.awk: Fix offset of substr().
6906 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
6908         * Makefile.in (pass-instances.def): New.
6909         (passes.o): Replace dependency on passes.def with one on
6910         pass-instances.def
6912         * gen-pass-instances.awk: New.
6914         * passes.c (pass_manager::pass_manager): Use pass-instances.def
6915         rather than passes.def, updating local definition of NEXT_PASS
6916         macro to add an extra NUM parameter (currently unused).
6918 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
6920         * Makefile.in (PASS_MANAGER_H): New.
6921         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
6922         (passes.o): Likewise.
6923         (statistics.o): Likewise.
6924         (cgraphunit.o): Likewise.
6925         (context.o): Depend on PASS_MANAGER_H.
6927         * pass_manager.h: New.
6929         * cgraphunit.c (cgraph_add_new_function): Update for moves
6930         of globals to fields of pass_manager.
6931         (analyze_function): Likewise.
6932         (expand_function): Likewise.
6933         (ipa_passes): Likewise.
6934         (compile): Likewise.
6936         * context.c (context::context): New.
6937         * context.h  (context::context): New.
6938         (context::get_passes): New.
6939         (context::passes_): New.
6941         * lto-cgraph.c (input_node): Update for moves of globals to
6942         fields of pass_manager.
6944         * passes.c (all_passes): Remove, in favor of a field of the
6945         same name within the new class pass_manager.
6946         (all_small_ipa_passes): Likewise.
6947         (all_lowering_passes): Likewise.
6948         (all_regular_ipa_passes): Likewise.
6949         (all_late_ipa_passes): Likewise.
6950         (all_lto_gen_passes): Likewise.
6951         (passes_by_id): Likewise.
6952         (passes_by_id_size): Likewise.
6953         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
6954         the new class pass_manager.
6955         (set_pass_for_id): Convert to...
6956         (pass_manager::set_pass_for_id): ...method.
6957         (get_pass_for_id): Convert to...
6958         (pass_manager::get_pass_for_id): ...method.
6959         (register_one_dump_file): Move body of implementation into...
6960         (pass_manager::register_one_dump_file): ...here.
6961         (register_dump_files_1): Convert to...
6962         (pass_manager::register_dump_files_1): ...method.
6963         (register_dump_files): Convert to...
6964         (pass_manager::register_dump_files): ...method.
6965         (create_pass_tab): Update for moves of globals to fields of
6966         pass_manager.
6967         (dump_passes): Move body of implementation into...
6968         (pass_manager::dump_passes): ...here.
6969         (register_pass): Move body of implementation into...
6970         (pass_manager::register_pass): ...here.
6971         (init_optimization_passes): Convert into...
6972         (pass_manager::pass_manager): ...constructor for new
6973         pass_manager class, and initialize the pass_lists array.
6974         (check_profile_consistency): Update for moves of globals to
6975         fields of pass_manager.
6976         (dump_profile_report): Move body of implementation into...
6977         (pass_manager::dump_profile_report): ...here.
6978         (ipa_write_summaries_1): Update for moves of pass lists from
6979         being globals to fields of pass_manager.
6980         (ipa_write_optimization_summaries): Likewise.
6981         (ipa_read_summaries):  Likewise.
6982         (ipa_read_optimization_summaries): Likewise.
6983         (execute_all_ipa_stmt_fixups): Likewise.
6985         * statistics.c (statistics_fini): Update for moves of globals to
6986         fields of pass_manager.
6988         * toplev.c (general_init): Replace call to
6989         init_optimization_passes with construction of the pass_manager
6990         instance.
6992         * tree-pass.h (all_passes): Remove, in favor of a field of the
6993         same name within the new class pass_manager.
6994         (all_small_ipa_passes): Likewise.
6995         (all_lowering_passes): Likewise.
6996         (all_regular_ipa_passes): Likewise.
6997         (all_lto_gen_passes): Likewise.
6998         (all_late_ipa_passes): Likewise.
6999         (passes_by_id): Likewise.
7000         (passes_by_id_size): Likewise.
7001         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
7002         the new class pass_manager.
7003         (get_pass_for_id): Remove.
7005 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
7007         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
7008         configs.
7010 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
7012         * arm.md (mulhi3): New expand pattern.
7014 2013-07-30  Jan Hubicka  <jh@suse.cz>
7015             Martin Liska  <marxin.liska@gmail.com>
7017         * profile.c (compute_value_histograms): Do not ICE when
7018         there is mismatch only on some counters.
7020 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
7022         PR rtl-optimization/57637
7023         * function.c (move_insn_for_shrink_wrap): Also check the
7024         GEN set of the LIVE problem for the liveness analysis
7025         if it exists, otherwise give up.
7027 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
7029         PR tree-optimization/57993
7030         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
7031         replaced statement in the candidate table.
7032         (phi_add_costs): Return infinite cost when the hidden basis does
7033         not dominate all phis on which the candidate is dependent.
7034         (replace_one_candidate): Record replaced statement in the
7035         candidate table.
7037 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
7039         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
7040         (ashlv2si3): New expander.
7041         (*ashlv2si3_i): New define_insn_and_split.
7042         * predicates.md (float_operation): Allow patterns with three
7043         basic sub-patterns.
7045         PR rtl-optimization/58021
7046         * mode-switching.c (create_pre_exit): Always split off preceding
7047         insns if we are not at the basic block head.
7049 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
7051         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
7052         (UCLIBC_DYNAMIC_LINKER): New macro.
7053         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
7054         `-mnan=2008'.
7055         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
7056         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
7057         `-mnan=2008'.
7058         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
7059         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
7060         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
7061         for SF and DF modes.  Use ieee_quad_format for TF mode.
7062         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
7063         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
7064         (mips_option_override): Handle `-mnan=legacy'.
7065         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
7066         `-mabs=2008' and `-mnan=2008'.
7067         (OPTION_DEFAULT_SPECS): Add "nan" default.
7068         (ASM_SPEC): Handle `-mnan='.
7069         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
7070         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
7071         comment accordingly.
7072         (neg<mode>2): Likewise.
7073         * config/mips/mips.opt (mabs, mnan): New options.
7074         * doc/install.texi (Configuration): Document `--with-nan=' option.
7075         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
7076         `-mnan=' options.
7077         (MIPS Options): Document them.
7078         * config.gcc <mips*-*-*>: Handle `--with-nan='.
7079         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
7080         * configure: Regenerate.
7081         * config.in: Regenerate.
7083 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
7085         * config/i386/i386.md (float post-reload splitters): Do not check
7086         for subregs of SSE registers.
7088 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
7089             H.J. Lu  <hongjiu.lu@intel.com>
7091         PR target/57954
7092         PR target/57988
7093         * config/i386/i386.md (post-reload splitter
7094         to avoid partial SSE reg dependency stalls): New pattern.
7096 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
7098         * config/s390/s390.md ("movcc"): Swap load and store instructions.
7100 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
7102         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
7103         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
7105 2013-07-26  Cary Coutant  <ccoutant@google.com>
7107         * dwarf2out.c (die_checksum_ordered): Don't include template
7108         instantiations in signature.
7109         (is_template_parameter): New function.
7110         (is_template_instantiation): New function.
7111         (generate_skeleton_bottom_up): Don't include template instantiations
7112         in type unit DIE.
7113         (generate_skeleton): Likewise.
7114         (break_out_comdat_types): Move recursive call to break out nested
7115         types earlier.
7116         (prune_unused_types_mark_generic_parms_dies): Call
7117         is_template_parameter.
7119 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
7121         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
7122         uses vector registers.
7123         * config/aarch64/iterators.md: Add attributes rtn and vas.
7125 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7126             Richard Earnshaw  <richard.earnshaw@arm.com>
7128         * combine.c (simplify_comparison): Re-canonicalize operands
7129         where appropriate.
7130         * config/arm/arm.md (movcond_addsi): New splitter.
7132 2013-07-25  Sterling Augustine  <saugustine@google.com>
7134         * dwarf2out.c (size_of_pubnames): Move code to...
7135         (include_pubname_in_output): ...here.  New.
7136         (want_pubnames): Rearrange.
7137         (output_pubnames): Call include_pubname_in_output.  Move assertion.
7139 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
7141         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
7143 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
7145         PR target/38836
7146         * doc/extend.texi: Remove obsolete builtins. Fix
7147         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
7149 2013-07-25  Jan Hubicka  <jh@suse.cz>
7151         * cgraph.c (release_function_body): Break out from ...
7152         (cgraph_release_function_body): ... this one; also release DECL_RESULT
7153         and DECL_ARGUMENTS.
7154         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
7155         old_tree in the map.
7156         (create_specialized_node): Update.
7157         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
7158         into index.
7159         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
7160         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
7161         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
7162         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
7163         DECL_RESULT.
7165 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7167         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
7168         addsi3_carryin_alt2_<optab>): Correct output template.
7170 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7172         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
7173         Adjust for arm_restrict_it.
7174         Remove trailing whitespace.
7176 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
7178         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
7179         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
7181         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
7183 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
7185         PR rtl-optimization/57960
7186         * lra-constraints.c (process_alt_operands): Use the right mode
7187         when checking strict_low.
7189 2013-07-25  Jan Hubicka  <jh@suse.cz>
7191         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
7192         * cgraph.c (cgraph_remove_node): Do not release function body
7193         when in cgraph streaming.
7194         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
7195         in other partitions are not considered reachable; fix handling of
7196         clones.
7198 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7200         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
7202 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
7204         PR target/19599
7205         PR target/57731
7206         PR target/57837
7207         * config/arm/arm.md ("*sibcall_insn): Replace use of
7208         Ss with US. Adjust output for v5 and v4t.
7209         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
7211         * config/arm/constraints.md ("Ss"): Rename to US.
7213 2013-07-25  Terry Guo  <terry.guo@arm.com>
7215         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
7216         shift_add/shift_sub0/shift_sub1 RTXs.
7218 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
7219             Anton Blanchard  <anton@au1.ibm.com>
7221         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
7222         (altivec_vpks<VI_char>ss): Likewise.
7223         (altivec_vpks<VI_char>us): Likewise.
7224         (altivec_vpku<VI_char>us): Likewise.
7225         (altivec_vpku<VI_char>um): Likewise.
7227 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
7229         Introduce context class.
7231         * Makefile.in (CONTEXT_H): New.
7232         (OBJS): Add context.o.
7233         (toplev.o): Add CONTEXT_H to dependencies.
7234         (context.o): New.
7236         * toplev.c (general_init): Create the singleton gcc::context instance.
7238         * context.c: New.
7240         * context.h: New.
7242 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
7244         PR rtl-optimization/57968
7245         * mode-switching.c (create_pre_exit): Allow instructions that
7246         don't set a return register to need a non-exit mode.
7248 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
7249             Anton Blanchard  <anton@au1.ibm.com>
7251         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
7252         operands to vperm for little endian.
7253         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
7254         of lvsl to create the control mask for a vperm for little endian.
7256 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7257             Anton Blanchard  <anton@au1.ibm.com>
7259         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
7260         two operands for little-endian.
7262 2013-07-23  Steve Ellcey  <sellcey@mips.com>
7264         * config/mips/mips.c (mips_case_values_threshold): New.
7265         (TARGET_CASE_VALUES_THRESHOLD): Define.
7267 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7268             Anton Blanchard  <anton@au1.ibm.com>
7270         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
7271         selection of field for vector splat in little endian mode.
7273 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
7275         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
7276         expanders to rs6000.md.
7277         (ior<mode>3): Likewise.
7278         (and<mode>3): Likewise.
7279         (one_cmpl<mode>2): Likewise.
7280         (nor<mode>3): Likewise.
7281         (andc<mode>3): Likewise.
7282         (eqv<mode>3): Likewise.
7283         (nand<mode>3): Likewise.
7284         (orc<mode>3): Likewise.
7286         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
7287         declaration.
7289         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
7290         to split multi-word logical operations.
7291         (rs6000_split_logical_di): Likewise.
7292         (rs6000_split_logical): Likewise.
7294         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
7295         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
7296         and allow TImode operations in 32-bit.
7297         (vsx_and<mode>3_64bit): Likewise.
7298         (vsx_ior<mode>3_32bit): Likewise.
7299         (vsx_ior<mode>3_64bit): Likewise.
7300         (vsx_xor<mode>3_32bit): Likewise.
7301         (vsx_xor<mode>3_64bit): Likewise.
7302         (vsx_one_cmpl<mode>2_32bit): Likewise.
7303         (vsx_one_cmpl<mode>2_64bit): Likewise.
7304         (vsx_nor<mode>3_32bit): Likewise.
7305         (vsx_nor<mode>3_64bit): Likewise.
7306         (vsx_andc<mode>3_32bit): Likewise.
7307         (vsx_andc<mode>3_64bit): Likewise.
7308         (vsx_eqv<mode>3_32bit): Likewise.
7309         (vsx_eqv<mode>3_64bit): Likewise.
7310         (vsx_nand<mode>3_32bit): Likewise.
7311         (vsx_nand<mode>3_64bit): Likewise.
7312         (vsx_orc<mode>3_32bit): Likewise.
7313         (vsx_orc<mode>3_64bit): Likewise.
7315         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
7316         logical types in GPRs.
7318         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
7319         logical insns to rs6000.md, and allow TImode operations in
7320         32-bit.
7321         (altivec_ior<mode>3): Likewise.
7322         (altivec_xor<mode>3): Likewise.
7323         (altivec_one_cmpl<mode>2): Likewise.
7324         (altivec_nor<mode>3): Likewise.
7325         (altivec_andc<mode>3): Likewise.
7327         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
7328         attributes for moving the 128-bit logical operations into
7329         rs6000.md.
7330         (BOOL_REGS_OUTPUT): Likewise.
7331         (BOOL_REGS_OP1): Likewise.
7332         (BOOL_REGS_OP2): Likewise.
7333         (BOOL_REGS_UNARY): Likewise.
7334         (BOOL_REGS_AND_CR0): Likewise.
7335         (one_cmpl<mode>2): Add support for DI logical operations on
7336         32-bit, splitting the operations to 32-bit.
7337         (anddi3): Likewise.
7338         (iordi3): Likewise.
7339         (xordi3): Likewise.
7340         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
7341         changes to combine the 32/64-bit code, allow logical operations on
7342         TI mode in 32-bit, and to use similar match_operator patterns like
7343         scalar mode uses.  Combine the Altivec and VSX code for logical
7344         operations, and move it here.
7345         (ior<mode>3, 128-bit types): Likewise.
7346         (xor<mode>3, 128-bit types): Likewise.
7347         (one_cmpl<mode>3, 128-bit types): Likewise.
7348         (nor<mode>3, 128-bit types): Likewise.
7349         (andc<mode>3, 128-bit types): Likewise.
7350         (eqv<mode>3, 128-bit types): Likewise.
7351         (nand<mode>3, 128-bit types): Likewise.
7352         (orc<mode>3, 128-bit types): Likewise.
7353         (and<mode>3_internal): Likewise.
7354         (bool<mode>3_internal): Likewise.
7355         (boolc<mode>3_internal1): Likewise.
7356         (boolc<mode>3_internal2): Likewise.
7357         (boolcc<mode>3_internal1): Likewise.
7358         (boolcc<mode>3_internal2): Likewise.
7359         (eqv<mode>3_internal1): Likewise.
7360         (eqv<mode>3_internal2): Likewise.
7361         (one_cmpl1<mode>3_internal): Likewise.
7363 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
7365         * config/microblaze/microblaze.c (microblaze_expand_prologue):
7366         Rename flag_stack_usage to flag_stack_usage_info.
7368 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
7370         * config/microblaze/sync.md: New file.
7371         * config/microblaze/microblaze.md: Include sync.md
7372         * config/microblaze/microblaze.c: Add print_operand 'y'.
7373         * config/microblaze/constraints.md: Add memory_contraint
7374         'Q' which is a single register.
7376 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
7378         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
7380 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
7382         * reload.c (find_reloads): Exit loop once we find this operand
7383         cannot be reloaded somehow for this alternative.
7385         * reload.c (find_reloads): Exit loop once we find a hard register.
7387         * rtlanal.c (computed_jump_p): Exit loop once we find label
7388         reference is used.
7390         * i386.c (ix86_pad_returns): Exit loop after setting replace.
7392         * cfgloopmanip.c (remove_path): Exit loop after setting
7393         irred_invalidated.
7395         * gensupport.c (subst_dup): Avoid loop if code is not
7396         MATCH_DUP nor MATCH_OP_DUP.
7398 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
7400         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
7402 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
7404         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
7405         true for SP_REGNUM if mode == ptr_mode.
7406         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
7407         with value R0_REGNUM + 31.
7409 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
7411         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
7412         pad pointer-typed argument downward.
7414 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
7416         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
7417         and __ILP32__ when the ILP32 model is in use.
7419 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
7421         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
7422         (aarch64_load_symref_appropriately): In the case of
7423         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
7424         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
7425         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
7426         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
7427         if 'mode' doesn't equal to 'ptr_mode'.
7428         (aarch64_output_mi_thunk): Add an assertion on the alignment of
7429         'vcall_offset'; change to call aarch64_emit_move differently depending
7430         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
7431         to calculate the upper bound of 'vcall_offset'.
7432         (aarch64_cannot_force_const_mem): Change to also return true if
7433         mode != ptr_mode.
7434         (aarch64_legitimize_reload_address): In the case of large
7435         displacements, add new local variable 'xmode' and an assertion
7436         based on it; change to use 'xmode' to generate the new rtx and
7437         reload.
7438         (aarch64_asm_trampoline_template): Change to generate the template
7439         differently depending on TARGET_ILP32 or not; change to use
7440         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
7441         (aarch64_trampoline_size): Removed.
7442         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
7443         and replace immediate literals with it.  Change to use 'ptr_mode'
7444         instead of 'DImode' and call convert_memory_address if the mode
7445         of 'fnaddr' doesn't equal to 'ptr_mode'.
7446         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
7447         to output symbol.
7448         (aarch64_elf_asm_destructor): Likewise.
7449         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
7450         on TARGET_ILP32 instead of aarch64_trampoline_size.
7451         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
7452         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
7453         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
7454         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
7455         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
7456         (storewb_pair<GPI:mode>_<P:mode>): ... this.
7457         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
7458         depending on the value of 'mode'.
7459         (add_losym_<mode>): New.
7460         (ldr_got_small_<mode>): New, based on ldr_got_small.
7461         (ldr_got_small): Remove.
7462         (ldr_got_small_sidi): New.
7463         * config/aarch64/iterators.md (P): New.
7464         (PTR): Change to 'ptr_mode' in the condition.
7466 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
7468         * config.gcc (aarch64*-*-*): Support --with-abi.
7469         (aarch64*-*-elf): Support --with-multilib-list.
7470         (aarch64*-*-linux*): Likewise.
7471         (supported_defaults): Add abi to aarch64*-*-*.
7472         * configure.ac: Mention AArch64 for --with-multilib-list.
7473         * configure: Re-generated.
7474         * config/aarch64/biarchilp32.h: New file.
7475         * config/aarch64/biarchlp64.h: New file.
7476         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
7477         (ABI_SPEC): Ditto.
7478         (MULTILIB_DEFAULTS): Ditto.
7479         (DRIVER_SELF_SPECS): Ditto.
7480         (ASM_SPEC): Update to also substitute -mabi.
7481         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
7482         file whose name depends on -mabi= and -mbig-endian.
7483         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
7484         TARGET_ILP32.
7485         (POINTER_SIZE): New define.
7486         (POINTERS_EXTEND_UNSIGNED): Ditto.
7487         (enum aarch64_abi_type): New enumeration tag.
7488         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
7489         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
7490         (TARGET_ILP32): New define.
7491         * config/aarch64/aarch64.opt (mabi): New.
7492         (aarch64_abi): New.
7493         (ilp32, lp64): New values for -mabi.
7494         * config/aarch64/t-aarch64 (comma): New define.
7495         (MULTILIB_OPTIONS): Ditto.
7496         (MULTILIB_DIRNAMES): Ditto.
7497         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
7498         * doc/invoke.texi: Document -mabi for AArch64.
7500 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
7502         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
7504 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
7505             Anton Blanchard  <anton@au1.ibm.com>
7507         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
7508         endianness when selecting field to splat.
7510 2013-07-22  Eric Christopher  <echristo@gmail.com>
7512         * dwarf2out.c (die_odr_checksum): New function to use
7513         CHECKSUM_ macros and ULEB128 for DIE tag.
7514         (generate_type_signature): Use.
7516 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
7518         * config.gcc (sparc*-*-*): Accept leon3 processor.
7519         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
7520         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
7521         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
7522         * config/sparc/sparc.opt (enum processor_type): Add leon3.
7523         (mfix-ut699): Adjust comment.
7524         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
7525         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
7526         (CPP_CPU_SPEC): Likewise.
7527         (ASM_CPU_SPEC): Likewise.
7528         * config/sparc/sparc.c (leon3_cost): New constant.
7529         (sparc_option_override): Add leon3 support.
7530         (mem_ref): New function.
7531         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
7532         (sparc_do_work_around_errata): Look into the instruction in the delay
7533         slot and adjust accordingly.  Add fix for the data cache nullify issues
7534         of the UT699.  Change insertion position for the NOP.
7535         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
7536         (leon3_load): New reservation.
7537         (leon_store): Bump latency to 2.
7538         (grfpu): New automaton.
7539         (grfpu_alu): New unit.
7540         (grfpu_ds): Likewise.
7541         (leon_fp_alu): Adjust.
7542         (leon_fp_mult): Delete.
7543         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
7544         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
7545         * config/sparc/sparc.md (cpu): Add leon3.
7546         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
7547         (swapsi): Likewise.
7548         (atomic_test_and_set): Likewise.
7549         (ldstub): Likewise.
7551 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
7553         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
7554         default for R5900 targets.
7555         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
7556         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
7557         * config/mips/mips.c (mips_option_override): Report an error for
7558         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
7559         for -march=r5900 -mhard-float.
7561 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
7563         * df-problems.c (can_move_insns_across): Exit loop once we
7564         find a non-fixed, non-global register.
7566         * ipa-pure-const.c (propagate_nothrow): Exit loop after
7567         setting can_throw.
7569         * omega.c (omega_eliminate_red): Break after setting red_found.
7570         (omega_problem_has_red_equations): Similarly after setting found.
7571         (omega_query_variable): Similarly after setting coupled.
7573 2013-07-22  Marek Polacek  <polacek@redhat.com>
7575         * gimplify.c: Don't include gimple.h twice.
7577 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7579         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
7580         instead of TARGET_THUMB1.
7581         (Pz): New constraint.
7582         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
7583         encodings.
7584         (compare_negsi_si): Likewise.
7585         (compare_addsi2_op0): Likewise.
7586         (compare_addsi2_op1): Likewise.
7587         (addsi3_carryin_<optab>): Likewise.
7588         (addsi3_carryin_alt2_<optab>): Likewise.
7589         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
7590         for arm_restrict_it.
7591         (subsi3_carryin): Likewise.
7592         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
7593         (minmax_arithsi): Disable for arm_restrict_it.
7594         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
7595         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
7596         (satsi_<SAT:code>_shift): Likewise.
7597         (arm_shiftsi3): Add alternative for 16-bit encoding.
7598         (arm32_movhf): Disable for arm_restrict_it.
7599         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
7600         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
7602 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
7604         * config/arm/arm.md (attribute "insn"): Delete.
7605         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
7606         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
7607         (not_shiftsi): Update for attribute change.
7608         (not_shiftsi_compare0): Likewise.
7609         (not_shiftsi_compare0_scratch): Likewise.
7610         (arm_one_cmplsi2): Likewise.
7611         (thumb1_one_cmplsi2): Likewise.
7612         (notsi_compare0): Likewise.
7613         (notsi_compare0_scratch): Likewise.
7614         (thumb1_movdi_insn): Likewise.
7615         (arm_movsi_insn): Likewise.
7616         (movhi_insn_arch4): Likewise.
7617         (movhi_bytes): Likewise.
7618         (arm_movqi_insn): Likewise.
7619         (thumb1_movqi_insn): Likewise.
7620         (arm32_movhf): Likewise.
7621         (thumb1_movhf): Likewise.
7622         (arm_movsf_soft_insn): Likewise.
7623         (thumb1_movsf_insn): Likewise.
7624         (thumb_movdf_insn): Likewise.
7625         (movsicc_insn): Likewise.
7626         (movsfcc_soft_insn): Likewise.
7627         (and_scc): Likewise.
7628         (cond_move): Likewise.
7629         (if_move_not): Likewise.
7630         (if_not_move): Likewise.
7631         (if_shift_move): Likewise.
7632         (if_move_shift): Likewise.
7633         (if_shift_shift): Likewise.
7634         (if_not_arith): Likewise.
7635         (if_arith_not): Likewise.
7636         (cond_move_not): Likewise.
7637         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
7638         (neon_mov<mode>): Likewise.
7639         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
7640         (thumb2_movsi_vfp): Likewise.
7641         (movsf_vfp): Likewise.
7642         (thumb2_movsf_vfp): Likewise.
7643         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
7644         change.
7645         (cortexa7_older_only): Likewise.
7646         (cortexa7_younger): Likewise.
7647         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
7648         (1020alu_shift_op): Likewise.
7649         (1020alu_shift_reg_op): Likewise.
7650         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
7651         (alu_shift_op): Likewise.
7652         (alu_shift_reg_op): Likewise.
7653         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
7654         (11_alu_shift_op): Likewise.
7655         (11_alu_shift_reg_op): Likewise.
7656         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
7657         (9_alu_shift_reg_op): Likewise.
7658         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7659         change.
7660         (cortex_a15_alu_shift): Likewise.
7661         (cortex_a15_alu_shift_reg): Likewise.
7662         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
7663         (cortex_a5_alu_shift): Likewise.
7664         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
7665         change.
7666         (cortex_a53_alu_shift): Likewise.
7667         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
7668         change.
7669         (cortex_a7_alu_reg): Likewise.
7670         (cortex_a7_alu_shift): Likewise.
7671         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
7672         (cortex_a8_alu_shift): Likewise.
7673         (cortex_a8_alu_shift_reg): Likewise.
7674         (cortex_a8_mov): Likewise.
7675         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
7676         (cortex_a9_dp_shift): Likewise.
7677         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
7678         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
7679         (cortex_r4_mov): Likewise.
7680         (cortex_r4_alu_shift): Likewise.
7681         (cortex_r4_alu_shift_reg): Likewise.
7682         * config/arm/fa526.md (526_alu_op): Update for attribute change.
7683         (526_alu_shift_op): Likewise.
7684         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
7685         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
7686         (626te_alu_shift_op): Likewise.
7687         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
7688         (726te_alu_op): Likewise.
7689         (726te_alu_shift_op): Likewise.
7690         (726te_alu_shift_reg_op): Likewise.
7691         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
7692         (mp626_alu_shift_op): Likewise.
7693         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
7694         (pj4_alu_e1_conds): Likewise.
7695         (pj4_alu): Likewise.
7696         (pj4_alu_conds): Likewise.
7697         (pj4_shift): Likewise.
7698         (pj4_shift_conds): Likewise.
7699         (pj4_alu_shift): Likewise.
7700         (pj4_alu_shift_conds): Likewise.
7702 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7704         * config/arm/predicates.md (shiftable_operator_strict_it):
7705         New predicate.
7706         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
7707         Disable cond_exec version for arm_restrict_it.
7708         (thumb2_smaxsi3): Convert to generate cond_exec.
7709         (thumb2_sminsi3): Likewise.
7710         (thumb32_umaxsi3): Likewise.
7711         (thumb2_uminsi3): Likewise.
7712         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
7713         (thumb2_neg_abssi2): Likewise.
7714         (thumb2_mov_scc): Add alternative for 16-bit encoding.
7715         (thumb2_movsicc_insn): Adjust alternatives.
7716         (thumb2_mov_negscc): Disable for arm_restrict_it.
7717         (thumb2_mov_negscc_strict_it): New pattern.
7718         (thumb2_mov_notscc_strict_it): New pattern.
7719         (thumb2_mov_notscc): Disable for arm_restrict_it.
7720         (thumb2_ior_scc): Likewise.
7721         (thumb2_ior_scc_strict_it): New pattern.
7722         (thumb2_cond_move): Adjust for arm_restrict_it.
7723         (thumb2_cond_arith): Disable for arm_restrict_it.
7724         (thumb2_cond_arith_strict_it): New pattern.
7725         (thumb2_cond_sub): Adjust for arm_restrict_it.
7726         (thumb2_movcond): Likewise.
7727         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
7728         (thumb2_zero_extendhisi2_v6): Likewise.
7729         (thumb2_zero_extendqisi2_v6): Likewise.
7730         (orsi_notsi_si): Likewise.
7731         (orsi_not_shiftsi_si): Likewise.
7733 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
7735         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
7736         instruction sequence is 1 byte shorter.
7738 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
7740         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
7741         it is not needed after split.
7743 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
7745         PR target/51784
7746         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
7747         second label for nonlocal goto receivers. Don't output pic base labels
7748         unless we're producing PIC; mark that action unreachable().
7749         (ix86_save_reg): If the function contains a nonlocal label, save the
7750         PIC base reg.
7751         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
7752         * config/darwin.c (emitted_pic_label_num): New GTY.
7753         (update_pic_label_number_if_needed): New.
7754         (machopic_output_function_base_name): Adjust for nonlocal receiver
7755         case.
7756         (machopic_should_output_picbase_label): New.
7757         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
7758         (nonlocal_goto_receiver): New insn and split.
7760 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
7762         * config/aarch64/aarch64-builtins.c
7763         (aarch64_fold_builtin): Fold abs in all modes.
7764         * config/aarch64/aarch64-simd-builtins.def
7765         (abs): Enable for all modes.
7766         * config/aarch64/arm_neon.h
7767         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
7768         (vabs_f64): Add missing intrinsic.
7770 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
7772         * config/aarch64/arm_neon.h (vabs_s64): New function
7774 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
7776         PR target/57516
7777         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
7778         * config/avr/avr.md (adjust_len): Add `round'.
7779         * config/avr/avr-protos.h (avr_out_round): New prototype.
7780         (avr_out_plus): Add `out_label' argument.
7781         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
7782         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
7783         Handle the case where `insn' is just a pattern.
7784         (avr_out_bitop): Handle the case where `insn' is just a pattern.
7785         (avr_out_round): New function.
7786         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
7788 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
7790         * config/microblaze/microblaze.c (microblaze_expand_prologue):
7791         Add check for flag_stack_usage to handle -fstack-usage support
7793 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
7795         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
7796         interaction for new Power8 flags and VSX.
7798 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
7800         PR middle-end/57698
7801         * tree-inline.c (expand_call_inline): Emit errors during
7802         early_inlining only if optimization is not turned on.
7804 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
7806         * passes.def: New.
7808         * passes.c (init_optimization_passes): Move the construction of
7809         the pass hierarchy into a new passes.def file.
7811         * Makefile.in (passes.o): Add dependency on passes.def.
7813 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
7815         * passes.c (init_optimization_passes): Introduce macros for
7816         constructing the tree of passes (INSERT_PASSES_AFTER,
7817         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
7818         TERMINATE_PASS_LIST).
7820 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
7821             Wei Mi  <wmi@google.com>
7823         PR rtl-optimization/57878
7824         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
7825         top.
7826         (reload_pseudo_compare_func): Check nregs first for reload
7827         pseudos.
7829 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
7831         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
7833 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
7835         * read-rtl.c (validate_const_int): Once an invalid character is
7836         seen, quit the loop.
7838         * gengtype.c (write_roots): Similarly once we find the "deletable"
7839         or "if_marked" option.
7841 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
7843         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
7844         "xtab" and "sat".  Move value "clz" from here to ...
7845         (attriubte "type"): ... here.
7846         (satsi_<SAT:code>): Delete "insn" attribute.
7847         (satsi_<SAT:code>_shift): Likewise.
7848         (arm_zero_extendqisi2addsi): Likewise.
7849         (arm_extendqisi2addsi): Likewise.
7850         (clzsi2): Update for attribute changes.
7851         (rbitsi2): Likewise.
7852         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
7853         attribute.
7854         (arm_usatsihi): Likewise.
7855         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
7857 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
7859         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
7860         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
7861         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
7862         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
7863         in alphabetical order.
7864         (attribute "core_cycles"): Update for attribute changes.
7865         (arm_addsi3): Likewise.
7866         (addsi3_compare0): Likewise.
7867         (addsi3_compare0_scratch): Likewise.
7868         (addsi3_compare_op1): Likewise.
7869         (addsi3_compare_op2): Likewise.
7870         (compare_addsi2_op0): Likewise.
7871         (compare_addsi2_op1): Likewise.
7872         (addsi3_carryin_shift_<optab>): Likewise.
7873         (subsi3_carryin_shift): Likewise.
7874         (rsbsi3_carryin_shift): Likewise.
7875         (arm_subsi3_insn): Likewise.
7876         (subsi3_compare0): Likewise.
7877         (subsi3_compare): Likewise.
7878         (arm_andsi3_insn): Likewise.
7879         (thumb1_andsi3_insn): Likewise.
7880         (andsi3_compare0): Likewise.
7881         (andsi3_compare0_scratch): Likewise.
7882         (zeroextractsi_compare0_scratch
7883         (andsi_not_shiftsi_si): Likewise.
7884         (iorsi3_insn): Likewise.
7885         (iorsi3_compare0): Likewise.
7886         (iorsi3_compare0_scratch): Likewise.
7887         (arm_xorsi3): Likewise.
7888         (thumb1_xorsi3_insn): Likewise.
7889         (xorsi3_compare0): Likewise.
7890         (xorsi3_compare0_scratch): Likewise.
7891         (satsi_<SAT:code>_shift): Likewise.
7892         (rrx): Likewise.
7893         (arm_shiftsi3): Likewise.
7894         (shiftsi3_compare0): Likewise.
7895         (not_shiftsi): Likewise.
7896         (not_shiftsi_compare0): Likewise.
7897         (not_shiftsi_compare0_scratch): Likewise.
7898         (arm_one_cmplsi2): Likewise.
7899         (thumb_one_complsi2): Likewise.
7900         (notsi_compare0): Likewise.
7901         (notsi_compare0_scratch): Likewise.
7902         (thumb1_zero_extendhisi2): Likewise.
7903         (arm_zero_extendhisi2): Likewise.
7904         (arm_zero_extendhisi2_v6): Likewise.
7905         (arm_zero_extendhisi2addsi): Likewise.
7906         (thumb1_zero_extendqisi2): Likewise.
7907         (thumb1_zero_extendqisi2_v6): Likewise.
7908         (arm_zero_extendqisi2): Likewise.
7909         (arm_zero_extendqisi2_v6): Likewise.
7910         (arm_zero_extendqisi2addsi): Likewise.
7911         (thumb1_extendhisi2): Likewise.
7912         (arm_extendhisi2): Likewise.
7913         (arm_extendhisi2_v6): Likewise.
7914         (arm_extendqisi): Likewise.
7915         (arm_extendqisi_v6): Likewise.
7916         (arm_extendqisi2addsi): Likewise.
7917         (thumb1_extendqisi2): Likewise.
7918         (thumb1_movdi_insn): Likewise.
7919         (arm_movsi_insn): Likewise.
7920         (movsi_compare0): Likewise.
7921         (movhi_insn_arch4): Likewise.
7922         (movhi_bytes): Likewise.
7923         (arm_movqi_insn): Likewise.
7924         (thumb1_movqi_insn): Likewise.
7925         (arm32_movhf): Likewise.
7926         (thumb1_movhf): Likewise.
7927         (arm_movsf_soft_insn): Likewise.
7928         (thumb1_movsf_insn): Likewise.
7929         (movdf_soft_insn): Likewise.
7930         (thumb_movdf_insn): Likewise.
7931         (arm_cmpsi_insn): Likewise.
7932         (cmpsi_shiftsi): Likewise.
7933         (cmpsi_shiftsi_swp): Likewise.
7934         (arm_cmpsi_negshiftsi_si): Likewise.
7935         (movsicc_insn): Likewise.
7936         (movsfcc_soft_insn): Likewise.
7937         (arith_shiftsi): Likewise.
7938         (arith_shiftsi_compare0
7939         (arith_shiftsi_compare0_scratch
7940         (sub_shiftsi): Likewise.
7941         (sub_shiftsi_compare0
7942         (sub_shiftsi_compare0_scratch
7943         (and_scc): Likewise.
7944         (cond_move): Likewise.
7945         (if_plus_move): Likewise.
7946         (if_move_plus): Likewise.
7947         (if_move_not): Likewise.
7948         (if_not_move): Likewise.
7949         (if_shift_move): Likewise.
7950         (if_move_shift): Likewise.
7951         (if_shift_shift): Likewise.
7952         (if_not_arith): Likewise.
7953         (if_arith_not): Likewise.
7954         (cond_move_not): Likewise.
7955         (thumb1_ashlsi3): Set type attribute.
7956         (thumb1_ashrsi3): Likewise.
7957         (thumb1_lshrsi3): Likewise.
7958         (thumb1_rotrsi3): Likewise.
7959         (shiftsi3_compare0_scratch): Likewise.
7960         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
7961         (neon_mov<mode>): Likewise.
7962         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
7963         attribute changes.
7964         (thumb2_movsi_insn): Likewise.
7965         (thumb2_cmpsi_neg_shiftsi): Likewise.
7966         (thumb2_extendqisi_v6): Likewise.
7967         (thumb2_zero_extendhisi2_v6): Likewise.
7968         (thumb2_zero_extendqisi2_v6): Likewise.
7969         (thumb2_shiftsi3_short): Likewise.
7970         (thumb2_addsi3_compare0_scratch): Likewise.
7971         (orsi_not_shiftsi_si): Likewise.
7972         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
7973         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
7974         changes.
7975         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
7976         (1020alu_shift_op): Likewise.
7977         (1020alu_shift_reg_op): Likewise.
7978         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
7979         (alu_shift_op): Likewise.
7980         (alu_shift_reg_op): Likewise.
7981         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
7982         (11_alu_shift_op): Likewise.
7983         (11_alu_shift_reg_op): Likewise.
7984         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
7985         (9_alu_shift_reg_op): Likewise.
7986         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7987         changes.
7988         (cortex_a15_alu_shift): Likewise.
7989         (cortex_a15_alu_shift_reg): Likewise.
7990         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
7991         changes.
7992         (cortex_a5_alu_shift): Likewise.
7993         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
7994         changes.
7995         (cortex_a53_alu_shift): Likewise.
7996         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
7997         changes.
7998         (cortex_a7_alu_reg): Likewise.
7999         (cortex_a7_alu_shift): Likewise.
8000         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
8001         changes.
8002         (cortex_a8_alu_shift): Likewise.
8003         (cortex_a8_alu_shift_reg): Likewise.
8004         (cortex_a8_mov): Likewise.
8005         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
8006         (cortex_a9_dp_shift): Likewise.
8007         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
8008         changes.
8009         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
8010         changes.
8011         (cortex_r4_mov): Likewise.
8012         (cortex_r4_alu_shift): Likewise.
8013         (cortex_r4_alu_shift_reg): Likewise.
8014         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
8015         (526_alu_shift_op): Likewise.
8016         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
8017         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
8018         (626te_alu_shift_op): Likewise.
8019         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
8020         (726te_alu_op): Likewise.
8021         (726te_alu_shift_op): Likewise.
8022         (726te_alu_shift_reg_op): Likewise.
8023         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
8024         (mp626_alu_shift_op): Likewise.
8025         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
8026         (pj4_alu_e1_conds): Likewise.
8027         (pj4_alu): Likewise.
8028         (pj4_alu_conds): Likewise.
8029         (pj4_shift): Likewise.
8030         (pj4_shift_conds): Likewise.
8031         (pj4_alu_shift): Likewise.
8032         (pj4_alu_shift_conds): Likewise.
8033         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
8034         changes.
8035         (cortexa7_older_only): Likewise.
8036         (cortexa7_younger): Likewise.
8038 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
8040         * ipa-pure-const.c (generate_summary): Rename to...
8041         (pure_const_generate_summary): ... this.
8043 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
8045         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
8047 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
8049         PR target/57909
8050         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
8051         usage in HI mode.
8053 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8055         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
8056         enabled without -march=zEC12.
8057         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
8058         flags to be set.
8060 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
8062         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
8063         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
8064         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
8065         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
8066         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
8067         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
8068         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
8069         ISA_HAS_FP4.
8070         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
8071         and ISA_HAS_NMADD3_NMSUB3.
8072         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
8073         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
8074         (nmsub4<mode>, nmsub3<mode>): Likewise.
8075         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
8077 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
8079         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
8080         TARGET_MIPS5400 checking.
8082 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
8083             Peter Bergner  <bergner@vnet.ibm.com>
8085         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
8086         registers in the comment.
8087         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
8088         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
8089         rather than FIRST_PSEUDO_REGISTERS.
8091 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
8093         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
8094         enable extra ISA flags with TARGET_HTM.
8096 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
8098         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
8099         Fix comment typos.
8101 2013-07-15  Cong Hou  <congh@google.com>
8103         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
8104         in compare function for sorting.
8106 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
8108         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
8109         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
8110         * config/rs6000/rs6000.opt: Add -mhtm option.
8111         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
8112         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
8113         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
8114         __HTM__ if the HTM instructions are available.
8115         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
8116         htm_spr_reg_operand): New define_predicates.
8117         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
8118         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
8119         Include htm.md.
8120         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
8121         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
8122         HTM builtin functions.
8123         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
8124         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
8125         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
8126         (rs6000_builtin_mask_calculate): Likewise.
8127         (rs6000_option_override_internal): Likewise.
8128         (bdesc_htm): Add new HTM builtin support.
8129         (htm_spr_num): New function.
8130         (htm_spr_regno): Likewise.
8131         (rs6000_htm_spr_icode): Likewise.
8132         (htm_expand_builtin): Likewise.
8133         (htm_init_builtins): Likewise.
8134         (rs6000_expand_builtin): Add support for HTM builtin functions.
8135         (rs6000_init_builtins): Likewise.
8136         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
8137         option.
8138         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
8139         (TARGET_HTM, MASK_HTM): Define macros.
8140         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
8141         (FIXED_REGISTERS): Likewise.
8142         (CALL_USED_REGISTERS): Likewise.
8143         (CALL_REALLY_USED_REGISTERS): Likewise.
8144         (REG_ALLOC_ORDER): Likewise.
8145         (enum reg_class): Likewise.
8146         (REG_CLASS_NAMES): Likewise.
8147         (REG_CLASS_CONTENTS): Likewise.
8148         (REGISTER_NAMES): Likewise.
8149         (ADDITIONAL_REGISTER_NAMES): Likewise.
8150         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
8151         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
8152         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
8153         * config/rs6000/htm.md: New file.
8154         * config/rs6000/htmintrin.h: New file.
8155         * config/rs6000/htmxlintrin.h: New file.
8157 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8159         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
8160         Define SYMBOL_TINY_GOT, update comment.
8161         * config/aarch64/aarch64.c
8162         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
8163         (aarch64_expand_mov_immediate): Likewise.
8164         (aarch64_print_operand): Likewise.
8165         (aarch64_classify_symbol): Likewise.
8166         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
8167         (ldr_got_tiny): Define.
8169 2013-07-13  Tobias Grosser  <tobias@grosser.es>
8171         PR tree-optimization/54094
8172         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
8173           scheduling dimension for the parallelism check from the polyhedral
8174           information in the AST.
8175         * graphite-dependences.c (carries_deps): Do not assume the schedule is
8176           in 2D + 1 form.
8178 2013-07-13  Jason Merrill  <jason@redhat.com>
8180         * print-tree.c (debug_vec_tree): Use debug_raw.
8181         (debug_raw (vec<tree, va_gc> &)): New.
8182         (debug_raw (vec<tree, va_gc> *)): New.
8183         * tree.h: Declare them.
8185 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
8187         * ifcvt.c (ifcvt_after_combine): New static variable.
8188         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
8189         for size.
8190         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
8191         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
8192         rest_of_handle_if_after_reload): Pass new argument for if_convert.
8194 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
8196         * config/mips/mips.c (mips_expand_call): Remove empty statement.
8198 2013-07-12  Michael Matz  <matz@suse.de>
8200         PR middle-end/55771
8201         * convert.c (convert_to_real): Reject non-float inner types.
8203 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
8205         * config/aarch64/aarch64-protos.h
8206         (aarch64_simd_immediate_valid_for_move): Remove.
8207         * config/aarch64/aarch64.c (simd_immediate_info): New member.
8208         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
8209         cases.
8210         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
8212 2013-07-11  Steve Ellcey  <sellcey@mips.com>
8214         * config/mips/mips.c (mips_conditional_register_usage): Do not
8215         use t[0-7] registers in MIPS16 mode when optimizing for size.
8217 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
8219         * config/i386/i386.c (dispatch_function_versions): Fix array
8220         indexing of function_version_info to match actual_versions.
8222 2013-07-11  Teresa Johnson  <tejohnson@google.com>
8224         * vec.h (struct va_gc): Move release out-of-line.
8225         (va_gc::release): Call ggc_free on released vec.
8227 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
8229         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
8230         Require GOT register as additional operand in UNSPEC.
8231         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
8232         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
8233         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
8234         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
8235         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
8236         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
8237         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
8239 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
8241         PR target/57631
8242         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
8243         name seen by assembler/linker rather if available.
8245 2013-07-11  Andreas Schwab  <schwab@suse.de>
8247         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
8249 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
8251         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
8253 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
8255         * doc/tm.texi.in: Move hook documentation to ....
8256         * target.def: ... here.
8258         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
8259         text on @hook line.
8260         * doc/tm.texi: Regenerate.
8262 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
8264         PR c++/57869
8265         * doc/invoke.texi: Document -Wconditionally-supported.
8267 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
8269         PR target/57844
8270         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
8271         of my_fp.
8273 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
8275         PR target/57506
8276         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
8277         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
8278         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
8279         Remove duplicate devices.
8280         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
8281         * config/avr/t-multilib: Regenerate.
8282         * config/avr/avr-tables.opt: Regenerate.
8283         * doc/avr-mmcu.texi: Regenerate.
8285 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
8287         PR target/56987
8288         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
8290 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
8292         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
8293         the cost of MULT when optimizing for size.
8295 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
8297         * config/cr16/cr16-protos.h: Don't include target.h.
8299 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
8301         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
8302         adjust register size for TDmode and TFmode for VSX registers.
8304 2013-07-08  Kai Tietz  <ktietz@redhat.com>
8306         PR target/56892
8307         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
8308         hook_bool_const_tree_true.
8310 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8312         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
8313         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
8314         * config/s390/s390.md: Define FPR*_REGNUM constants.
8315         Fix FPR2_REGNUM constant (18 -> 17).
8316         ("*trunc<BFP:mode><DFP_ALL:mode>2")
8317         ("*trunc<DFP_ALL:mode><BFP:mode>2")
8318         ("trunc<BFP:mode><DFP_ALL:mode>2")
8319         ("trunc<DFP_ALL:mode><BFP:mode>2")
8320         ("*extend<BFP:mode><DFP_ALL:mode>2")
8321         ("*extend<DFP_ALL:mode><BFP:mode>2")
8322         ("extend<BFP:mode><DFP_ALL:mode>2")
8323         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
8324         FPR4_REGNUM.
8326 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
8328         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
8330 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8332         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
8333         and cfun_fpr_bit_p to cfun_fpr_save_p.
8334         (s390_frame_area, s390_register_info, s390_frame_info)
8335         (s390_emit_prologue, s390_emit_epilogue)
8336         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
8337         register numbers.
8338         * config/s390/s390.h: Define *_REGNUM macros for floating point
8339         register numbers.
8341 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
8343         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
8345 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
8347         PR rtl-optimization/57786
8348         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
8349         and break out of the loop when it is set to false.
8351 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
8353         PR target/57819
8354         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
8355         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
8356         (const_int 63)) 0)).
8357         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
8358         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
8359         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
8361         PR rtl-optimization/57829
8362         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
8363         mask bits outside of mode are just sign-extension from mode to HWI.
8365 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
8367         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
8368         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
8369         adjust_address instead of change_address to keep info about alignment.
8370         (emit_strmov): Remove.
8371         (emit_memmov): New function.
8372         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
8373         (expand_movmem_epilogue): Likewise and return updated rtx for
8374         destination.
8375         (expand_constant_movmem_prologue): Likewise and return updated rtx for
8376         destination and source.
8377         (decide_alignment): Refactor, handle vector_loop.
8378         (ix86_expand_movmem): Likewise.
8379         (ix86_expand_setmem): Likewise.
8380         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
8382 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
8384         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
8385         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
8387 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
8389         * config/i386/sse.md (sse_movlhps): Change alternative 3
8390         of operand 2 to "m".
8392 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
8394         PR target/57807
8395         * config/i386/sse.md (iptr): New mode attribute.
8396         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
8397         (<sse>_vm<plusminus_insn><mode>3): Ditto.
8398         (<sse>_vmmul<mode>3): Ditto.
8399         (<sse>_vmdiv<mode>3): Ditto.
8400         (sse_vmrcpv4sf2): Ditto.
8401         (<sse>_vmsqrt<mode>2): Ditto.
8402         (sse_vmrsqrtv4sf2): Ditto.
8403         (<sse>_vm<code><mode>3): Ditto.
8404         (avx_vmcmp<mode>3): Ditto.
8405         (<sse>_vmmaskcmp<mode>3): Ditto.
8406         (<sse>_comi): Ditto.
8407         (<sse>_ucomi): Ditto.
8408         (*xop_vmfrcz_<mode>): Ditto.
8409         (*fmai_fmadd_<mode>): Ditto.
8410         (*fmai_fmsub_<mode>): Ditto.
8411         (*fmai_fnmadd_<mode>): Ditto.
8412         (*fmai_fnmsub_<mode>): Ditto.
8413         (*fma4i_vmfmadd_<mode>): Ditto.
8414         (*fma4i_vmfmsub_<mode>): Ditto.
8415         (*fma4i_vmfnmadd_<mode>): Ditto.
8416         (*fma4i_vmfnmsub_<mode>): Ditto.
8417         (*xop_vmfrcz_<mode>): Ditto.
8418         (sse_cvtps2pi): Ditto.
8419         (sse_cvttps2pi): Ditto.
8420         (sse_cvtss2si): Ditto.
8421         (sse_cvtss2si_2): Ditto.
8422         (sse_cvtss2siq_2): Ditto.
8423         (sse_cvttss2si): Ditto.
8424         (sse_cvttss2siq): Ditto.
8425         (sse_cvtsd2si): Ditto.
8426         (sse_cvtsd2si_2): Ditto.
8427         (sse_cvtsd2siq_2): Ditto.
8428         (sse_cvttsd2si): Ditto.
8429         (sse_cvttsd2siq): Ditto.
8430         (sse_cvtsd2ss): Ditto.
8431         (sse_cvtss2sd): Ditto.
8432         (avx2_pbroadcast<mode>): Ditto.
8433         (avx2_pbroadcast<mode>_1): Ditto.
8434         (*avx_vperm_broadcast_v4sf): Ditto.
8436         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
8437         (sse_movlhps): Ditto.
8438         (sse_storehps): Ditto.
8439         (sse_loadhps): Ditto.
8440         (sse_storelps): Ditto.
8441         (sse_loadlps): Ditto.
8442         (*vec_concatv4sf): Ditto.
8443         (*vec_interleave_highv2df): Ditto.
8444         (*vec_interleave_lowv2df): Ditto.
8445         (*vec_extractv2df_1_sse): Ditto.
8446         (*vec_extractv2df_0_sse): Ditto.
8447         (sse2_storelpd): Ditto.
8448         (sse2_loadlpd): Ditto.
8449         (sse2_movsd): Ditto.
8450         (*vec_concatv4si): Ditto.
8451         (vec_concatv2di): Ditto.
8453         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
8454         for Intel asm dialect.
8455         (mmx_punpcklwd): Ditto.
8456         (mmx_punpckldq): Ditto.
8458         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
8459         for intel assembler dialect.
8461 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
8463         PR target/29776
8464         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
8465         for BUILT_IN_C{LZ,LRSB}*.
8466         * tree.h (CASE_INT_FN): Add FN##IMAX case.
8467         * tree-vrp.c (extract_range_basic): Handle
8468         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
8469         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
8470         fall thru to code calling set_value*.
8471         * builtins.c (expand_builtin): Remove *IMAX cases.
8472         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
8473         if width is bigger than 2*HWI.
8475 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
8477         PR rtl-optimization/55342
8478         * lra-int.h (lra_subreg_reload_pseudos): New.
8479         * lra.c: Add undoing optional reloads to the block diagram.
8480         (lra_subreg_reload_pseudos): New.
8481         (lra_optional_reload_pseudos): Change comments.
8482         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
8483         lra_optional_reload_pseudos after undo transformations.
8484         * lra-assigns.c (pseudo_prefix_title): New.
8485         (lra_setup_reg_renumber): Use it.
8486         (spill_for): Ditto.  Check subreg reload pseudos too.
8487         (assign_by_spills): Consider subreg reload pseudos too.
8488         * lra-constraints.c (simplify_operand_subreg): Use
8489         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
8490         (curr_insn_transform): Recognize and do optional reloads.
8491         (undo_optional_reloads): New.
8492         (lra_undo_inheritance): Call undo_optional_reloads.
8494 2013-07-05  Thomas Quinot  <quinot@adacore.com>
8496         * tree-complex.c (expand_complex_operations_1): Fix typo.
8498 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
8500         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
8501         (tune_params): New member 'const vec_costs'.
8502         * config/aarch64/aarch64.c (generic_vector_cost): New.
8503         (generic_tunings): New member 'generic_vector_cost'.
8504         (aarch64_builtin_vectorization_cost): New.
8505         (aarch64_add_stmt_cost): New.
8506         (TARGET_VECTORIZE_ADD_STMT_COST): New.
8507         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
8509 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
8511         PR target/57777
8512         * config/i386/predicates.md (vsib_address_operand): Disallow
8513         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
8515 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
8517         PR middle-end/55030
8518         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
8519         expand_builtin_setjmp_receiver.
8520         (expand_label): Adjust, call expand_builtin_setjmp_receiver
8521         with NULL for the label parameter.
8522         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
8523         the frame-pointer.  Adjust comments.
8524         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
8525         only if LABEL is non-NULL.
8527 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
8529         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
8530         (ARM_ABI_AAPCS64): Ditto.
8531         (arm_abi): Ditto.
8532         (ARM_DEFAULT_ABI): Ditto.
8534 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
8536         * config/aarch64/aarch64-builtins.c
8537         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
8538         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
8539         (st1): Likewise.
8540         * config/aarch64/aarch64-simd.md
8541         (aarch64_ld1<VALL:mode>): New.
8542         (aarch64_st1<VALL:mode>): Likewise.
8543         * config/aarch64/arm_neon.h
8544         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
8546 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
8548         * config/i386/i386.c (gate_insert_vzeroupper): Check if
8549         target ISA is AVX.
8550         (ix86_option_override_internal):Turn on all -mavx target flags by
8551         default as they are dependent on AVX anyway.
8553 2013-07-02  Cary Coutant  <ccoutant@google.com>
8555         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
8556         deterministic hash.
8557         (loc_checksum_ordered): Likewise.
8558         (hash_loc_operands): Remove inline keyword.
8560 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
8562         PR tree-optimization/57741
8563         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
8564         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
8565         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
8566         Allow REAL_CST step_exprs if flag_associative_math.
8567         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
8569 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
8571         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
8573 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
8575         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
8577 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8579         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
8580         encoding.
8581         (iorsi3_insn): Likewise.
8582         (arm_xorsi3): Likewise.
8584 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
8586         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
8587         to ...
8588         (attribute "type"): ... here, and prefix with "wmmx_".
8589         (attribute "core_cycles"): Update for attribute changes.
8590         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
8591         (tbcstv4hi): Likewise.
8592         (tbcstv2si): Likewise.
8593         (iwmmxt_iordi3): Likewise.
8594         (iwmmxt_xordi3): Likewise.
8595         (iwmmxt_anddi3): Likewise.
8596         (iwmmxt_nanddi3): Likewise.
8597         (iwmmxt_arm_movdi): Likewise.
8598         (iwmmxt_movsi_insn): Likewise.
8599         (mov<mode>_internal): Likewise.
8600         (and<mode>3_iwmmxt): Likewise.
8601         (ior<mode>3_iwmmxt): Likewise.
8602         (xor<mode>3_iwmmxt): Likewise.
8603         (add<mode>3_iwmmxt): Likewise.
8604         (ssaddv8qi3): Likewise.
8605         (ssaddv4hi3): Likewise.
8606         (ssaddv2si3): Likewise.
8607         (usaddv8qi3): Likewise.
8608         (usaddv4hi3): Likewise.
8609         (usaddv2si3): Likewise.
8610         (sub<mode>3_iwmmxt): Likewise.
8611         (sssubv8qi3): Likewise.
8612         (sssubv4hi3): Likewise.
8613         (sssubv2si3): Likewise.
8614         (ussubv8qi3): Likewise.
8615         (ussubv4hi3): Likewise.
8616         (ussubv2si3): Likewise.
8617         (mulv4hi3_iwmmxt): Likewise.
8618         (smulv4hi3_highpart): Likewise.
8619         (umulv4hi3_highpart): Likewise.
8620         (iwmmxt_wmacs): Likewise.
8621         (iwmmxt_wmacsz): Likewise.
8622         (iwmmxt_wmacu): Likewise.
8623         (iwmmxt_wmacuz): Likewise.
8624         (iwmmxt_clrdi): Likewise.
8625         (iwmmxt_clrv8qi): Likewise.
8626         (iwmmxt_clr4hi): Likewise.
8627         (iwmmxt_clr2si): Likewise.
8628         (iwmmxt_uavgrndv8qi3): Likewise.
8629         (iwmmxt_uavgrndv4hi3): Likewise.
8630         (iwmmxt_uavgv8qi3): Likewise.
8631         (iwmmxt_uavgv4hi3): Likewise.
8632         (iwmmxt_tinsrb): Likewise.
8633         (iwmmxt_tinsrh): Likewise.
8634         (iwmmxt_tinsrw): Likewise.
8635         (iwmmxt_textrmub): Likewise.
8636         (iwmmxt_textrmsb): Likewise.
8637         (iwmmxt_textrmuh): Likewise.
8638         (iwmmxt_textrmsh): Likewise.
8639         (iwmmxt_textrmw): Likewise.
8640         (iwmxxt_wshufh): Likewise.
8641         (eqv8qi3): Likewise.
8642         (eqv4hi3): Likewise.
8643         (eqv2si3): Likewise.
8644         (gtuv8qi3): Likewise.
8645         (gtuv4hi3): Likewise.
8646         (gtuv2si3): Likewise.
8647         (gtv8qi3): Likewise.
8648         (gtv4hi3): Likewise.
8649         (gtv2si3): Likewise.
8650         (smax<mode>3_iwmmxt): Likewise.
8651         (umax<mode>3_iwmmxt): Likewise.
8652         (smin<mode>3_iwmmxt): Likewise.
8653         (umin<mode>3_iwmmxt): Likewise.
8654         (iwmmxt_wpackhss): Likewise.
8655         (iwmmxt_wpackwss): Likewise.
8656         (iwmmxt_wpackdss): Likewise.
8657         (iwmmxt_wpackhus): Likewise.
8658         (iwmmxt_wpackwus): Likewise.
8659         (iwmmxt_wpackdus): Likewise.
8660         (iwmmxt_wunpckihb): Likewise.
8661         (iwmmxt_wunpckihh): Likewise.
8662         (iwmmxt_wunpckihw): Likewise.
8663         (iwmmxt_wunpckilb): Likewise.
8664         (iwmmxt_wunpckilh): Likewise.
8665         (iwmmxt_wunpckilw): Likewise.
8666         (iwmmxt_wunpckehub): Likewise.
8667         (iwmmxt_wunpckehuh): Likewise.
8668         (iwmmxt_wunpckehuw): Likewise.
8669         (iwmmxt_wunpckehsb): Likewise.
8670         (iwmmxt_wunpckehsh): Likewise.
8671         (iwmmxt_wunpckehsw): Likewise.
8672         (iwmmxt_wunpckelub): Likewise.
8673         (iwmmxt_wunpckeluh): Likewise.
8674         (iwmmxt_wunpckeluw): Likewise.
8675         (iwmmxt_wunpckelsb): Likewise.
8676         (iwmmxt_wunpckelsh): Likewise.
8677         (iwmmxt_wunpckelsw): Likewise.
8678         (ror<mode>3): Likewise.
8679         (ashr<mode>3_iwmmxt): Likewise.
8680         (lshr<mode>3_iwmmxt): Likewise.
8681         (ashl<mode>3_iwmmxt): Likewise.
8682         (ror<mode>3_di): Likewise.
8683         (ashr<mode>3_di): Likewise.
8684         (lshr<mode>3_di): Likewise.
8685         (ashl<mode>3_di): Likewise.
8686         (iwmmxt_wmadds): Likewise.
8687         (iwmmxt_wmaddu): Likewise.
8688         (iwmmxt_tmia): Likewise.
8689         (iwmmxt_tmiaph): Likewise.
8690         (iwmmxt_tmiabb): Likewise.
8691         (iwmmxt_tmiatb): Likewise.
8692         (iwmmxt_tmiabt): Likewise.
8693         (iwmmxt_tmiatt): Likewise.
8694         (iwmmxt_tmovmskb): Likewise.
8695         (iwmmxt_tmovmskh): Likewise.
8696         (iwmmxt_tmovmskw): Likewise.
8697         (iwmmxt_waccb): Likewise.
8698         (iwmmxt_wacch): Likewise.
8699         (iwmmxt_waccw): Likewise.
8700         (iwmmxt_waligni): Likewise.
8701         (iwmmxt_walignr): Likewise.
8702         (iwmmxt_walignr0): Likewise.
8703         (iwmmxt_walignr1): Likewise.
8704         (iwmmxt_walignr2): Likewise.
8705         (iwmmxt_walignr3): Likewise.
8706         (iwmmxt_wsadb): Likewise.
8707         (iwmmxt_wsadh): Likewise.
8708         (iwmmxt_wsadbz): Likewise.
8709         (iwmmxt_wsadhz): Likewise.
8710         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
8711         (iwmmxt_wabsdiffb): Likewise.
8712         (iwmmxt_wabsdiffh): Likewise.
8713         (iwmmxt_wabsdiffw): Likewise.
8714         (iwmmxt_waddsubhx): Likewise
8715         (iwmmxt_wsubaddhx): Likewise.
8716         (addc<mode>3): Likewise.
8717         (iwmmxt_avg4): Likewise.
8718         (iwmmxt_avg4r): Likewise.
8719         (iwmmxt_wmaddsx): Likewise.
8720         (iwmmxt_wmaddux): Likewise.
8721         (iwmmxt_wmaddsn): Likewise.
8722         (iwmmxt_wmaddun): Likewise.
8723         (iwmmxt_wmulwsm): Likewise.
8724         (iwmmxt_wmulwum): Likewise.
8725         (iwmmxt_wmulsmr): Likewise.
8726         (iwmmxt_wmulumr): Likewise.
8727         (iwmmxt_wmulwsmr): Likewise.
8728         (iwmmxt_wmulwumr): Likewise.
8729         (iwmmxt_wmulwl): Likewise.
8730         (iwmmxt_wqmulm): Likewise.
8731         (iwmmxt_wqmulwm): Likewise.
8732         (iwmmxt_wqmulmr): Likewise.
8733         (iwmmxt_wqmulwmr): Likewise.
8734         (iwmmxt_waddbhusm): Likewise.
8735         (iwmmxt_waddbhusl): Likewise.
8736         (iwmmxt_wqmiabb): Likewise.
8737         (iwmmxt_wqmiabt): Likewise.
8738         (iwmmxt_wqmiatb): Likewise.
8739         (iwmmxt_wqmiatt): Likewise.
8740         (iwmmxt_wqmiabbn): Likewise.
8741         (iwmmxt_wqmiabtn): Likewise.
8742         (iwmmxt_wqmiatbn): Likewise.
8743         (iwmmxt_wqmiattn): Likewise.
8744         (iwmmxt_wmiabb): Likewise.
8745         (iwmmxt_wmiabt): Likewise.
8746         (iwmmxt_wmiatb): Likewise.
8747         (iwmmxt_wmiatt): Likewise.
8748         (iwmmxt_wmiabbn): Likewise.
8749         (iwmmxt_wmiabtn): Likewise.
8750         (iwmmxt_wmiatbn): Likewise.
8751         (iwmmxt_wmiattn): Likewise.
8752         (iwmmxt_wmiawbb): Likewise.
8753         (iwmmxt_wmiawbt): Likewise.
8754         (iwmmxt_wmiawtb): Likewise.
8755         (iwmmxt_wmiawtt): Likewise.
8756         (iwmmxt_wmiawbbn): Likewise.
8757         (iwmmxt_wmiawbtn): Likewise.
8758         (iwmmxt_wmiawtbn): Likewise.
8759         (iwmmxt_wmiawttn): Likewise.
8760         (iwmmxt_wmerge): Likewise.
8761         (iwmmxt_tandc<mode>3): Likewise.
8762         (iwmmxt_torc<mode>3): Likewise.
8763         (iwmmxt_torvsc<mode>3): Likewise.
8764         (iwmmxt_textrc<mode>3): Likewise.
8765         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
8766         (wmmxt_pack): Likewise.
8767         (wmmxt_mult_c1): Likewise.
8768         (wmmxt_mult_c2): Likewise.
8769         (wmmxt_alu_c1): Likewise.
8770         (wmmxt_alu_c2): Likewise.
8771         (wmmxt_alu_c3): Likewise.
8772         (wmmxt_transfer_c1): Likewise.
8773         (wmmxt_transfer_c2): Likewise.
8774         (wmmxt_transfer_c3): Likewise.
8775         (marvell_f_iwmmxt_wstr): Likewise.
8776         (marvell_f_iwmmxt_wldr): Likewise.
8778 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
8780         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
8782 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
8784         Revert:
8785         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
8786         * lra-constraints.c (need_for_split_p): Check call used hard regs
8787         living through calls.
8789         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
8790         call used regs for call insn.
8792 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
8794         PR target/57736
8795         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
8796         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
8798 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
8800         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
8802 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
8804         * lra-constraints.c (need_for_split_p): Check call used hard regs
8805         living through calls.
8807 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
8809         PR target/57744
8810         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
8811         to tie with any other modes.  Eliminate Altivec vector mode tests,
8812         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
8813         code, to return 0 if testing MODE2 for a condition, if we've
8814         already tested MODE1 for the same condition.
8816 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8818         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
8819         layout.
8821 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8823         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
8824         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
8826 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
8828         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
8829         Define.
8830         (aarch64_symbolic_constant_p): Remove.
8831         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
8832         static.  Fix line length and white space.
8833         (aarch64_symbolic_constant_p): Remove.
8834         * config/aarch64/predicates.md (aarch64_valid_symref):
8835         Use aarch64_classify_symbol_expression.
8837 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8839         * config/arm/constraints.md (Ts): New constraint.
8840         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
8841         16-bit encodings.
8842         (compare_scc): Use "Ts" constraint for operand 0.
8843         (ior_scc_scc): Likewise.
8844         (and_scc_scc): Likewise.
8845         (and_scc_scc_nodom): Likewise.
8846         (ior_scc_scc_cmp): Likewise for operand 7.
8847         (and_scc_scc_cmp): Likewise.
8848         * config/arm/thumb2.md (thumb2_movsi_insn):
8849         Add alternatives for 16-bit encodings.
8850         (thumb2_movhi_insn): Likewise.
8851         (thumb2_movsicc_insn): Likewise.
8852         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
8853         (thumb2_negscc): Use "Ts" constraint.
8854         Move mvn instruction outside cond_exec block.
8855         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
8856         for 16-bit encodings.
8858 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8860         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
8861         encoding.
8862         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
8863         (mulsi3subsi): Likewise.
8864         (mulsidi3adddi): Likewise.
8865         (mulsidi3_v6): Likewise.
8866         (umulsidi3_v6): Likewise.
8867         (umulsidi3adddi_v6): Likewise.
8868         (smulsi3_highpart_v6): Likewise.
8869         (umulsi3_highpart_v6): Likewise.
8870         (mulhisi3tb): Likewise.
8871         (mulhisi3bt): Likewise.
8872         (mulhisi3tt): Likewise.
8873         (maddhisi4): Likewise.
8874         (maddhisi4tb): Likewise.
8875         (maddhisi4tt): Likewise.
8876         (maddhidi4): Likewise.
8877         (maddhidi4tb): Likewise.
8878         (maddhidi4tt): Likewise.
8879         (zeroextractsi_compare0_scratch): Likewise.
8880         (insv_zero): Likewise.
8881         (insv_t2): Likewise.
8882         (anddi_notzesidi_di): Likewise.
8883         (anddi_notsesidi_di): Likewise.
8884         (andsi_notsi_si): Likewise.
8885         (iordi_zesidi_di): Likewise.
8886         (xordi_zesidi_di): Likewise.
8887         (andsi_iorsi3_notsi): Likewise.
8888         (smax_0): Likewise.
8889         (smax_m1): Likewise.
8890         (smin_0): Likewise.
8891         (not_shiftsi): Likewise.
8892         (unaligned_loadsi): Likewise.
8893         (unaligned_loadhis): Likewise.
8894         (unaligned_loadhiu): Likewise.
8895         (unaligned_storesi): Likewise.
8896         (unaligned_storehi): Likewise.
8897         (extv_reg): Likewise.
8898         (extzv_t2): Likewise.
8899         (divsi3): Likewise.
8900         (udivsi3): Likewise.
8901         (arm_zero_extendhisi2addsi): Likewise.
8902         (arm_zero_extendqisi2addsi): Likewise.
8903         (compareqi_eq0): Likewise.
8904         (arm_extendhisi2_v6): Likewise.
8905         (arm_extendqisi2addsi): Likewise.
8906         (arm_movt): Likewise.
8907         (thumb2_ldrd): Likewise.
8908         (thumb2_ldrd_base): Likewise.
8909         (thumb2_ldrd_base_neg): Likewise.
8910         (thumb2_strd): Likewise.
8911         (thumb2_strd_base): Likewise.
8912         (thumb2_strd_base_neg): Likewise.
8913         (arm_negsi2): Add alternative for 16-bit encoding.
8914         (arm_one_cmplsi2): Likewise.
8916 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8918         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
8919         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
8920         (movdfcc): Likewise.
8921         * config/arm/vfp.md (*thumb2_movsf_vfp):
8922         Disable predication for arm_restrict_it.
8923         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
8924         (*thumb2_movdfcc_vfp): Likewise.
8925         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
8926         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
8927         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
8928         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
8929         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
8930         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
8931         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
8932         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
8933         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
8934         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
8935         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
8936         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
8937         Disable predication for arm_restrict_it.
8939 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
8941         * config/i386/bmiintrin.h (_bextr_u32): New.
8942         (_bextr_u64): Ditto.
8944 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
8946         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
8947         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
8948         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
8949         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
8950         * config/mips/n32-elf.h: ...this new file.
8952 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
8954         PR target/57224
8955         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
8956         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
8958 2013-06-27  Catherine Moore  <clm@codesourcery.com>
8960         * config/mips/mips-tables.opt: Regenerate.
8961         * config/mips/mips-cpus.def: Add m14ke and m14kec.
8962         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
8963         * doc/invoke.texi: Add -m14kc.
8965 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
8967         PR target/57623
8968         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
8969         constraints of operand 1 and 2.
8971         PR target/57623
8972         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
8973         to match RTL canonicalization.  Swap predicates and
8974         constraints of operand 1 and 2.
8976 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
8978         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
8979         Process OP_INOUT regs for splitting too.
8981 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
8983         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
8984         decl before the loop, initialize to NULL.
8985         (vectorizable_load): Initialize ptr_incr to NULL.
8987 2013-06-27  Martin Jambor  <mjambor@suse.cz>
8989         PR lto/57208
8990         * ipa-ref.h (ipa_maybe_record_reference): Declare.
8991         * ipa-ref.c (ipa_maybe_record_reference): New function.
8992         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
8993         * ipa-cp.c (create_specialized_node): Record potential references from
8994         aggvals.
8995         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
8997 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8999         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
9000         parameter 'mode' of type 'enum machine_mode mode'; change to pass
9001         'mode' to force_reg.
9002         (aarch64_add_offset): Update calls to aarch64_force_temporary.
9003         (aarch64_expand_mov_immediate): Likewise.
9005 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
9007         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
9008         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
9010 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9012         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
9013         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
9014         (struct machine_function): Add tbegin_p.
9015         (s390_canonicalize_comparison): Fold CC mode compares to
9016         conditional jump if possible.
9017         (s390_emit_jump): Return the emitted jump.
9018         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
9019         Handle CCRAWmode compares.
9020         (s390_option_override): Default to -mhtm if available.
9021         (s390_reg_clobbered_rtx): Handle floating point regs as well.
9022         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
9023         FPRs instead of df_regs_ever_live_p.
9024         (s390_optimize_nonescaping_tx): New function.
9025         (s390_init_frame_layout): Extend clobbered_regs array to cover
9026         FPRs as well.
9027         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
9028         (s390_expand_tbegin): New function.
9029         (enum s390_builtin): New enum definition.
9030         (code_for_builtin): New array definition.
9031         (s390_init_builtins): New function.
9032         (s390_expand_builtin): New function.
9033         (TARGET_INIT_BUILTINS): Define.
9034         (TARGET_EXPAND_BUILTIN): Define.
9035         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
9036         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
9037         (s390_alc_comparison): Likewise.
9038         * config/s390/s390-modes.def: Add CCRAWmode.
9039         * config/s390/s390.h (processor_flags): Add PF_TX.
9040         (TARGET_CPU_HTM): Define macro.
9041         (TARGET_HTM): Define macro.
9042         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
9043         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
9044         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
9045         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
9046         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
9047         (TBEGIN_MASK, TBEGINC_MASK): New constants.
9048         ("*cc_to_int"): Move up.
9049         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
9050         constants other than 0.
9051         ("*ccraw_to_int"): New insn and splitter definition.
9052         ("tbegin", "tbegin_nofloat", "tbegin_retry")
9053         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
9054         ("tx_assist"): New expander.
9055         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
9056         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
9057         * config/s390/s390.opt: Add -mhtm option.
9058         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
9059         * config/s390/htmxlintrin.h: New file.
9060         * config/s390/htmintrin.h: New file.
9061         * config/s390/s390intrin.h: New file.
9062         * doc/extend.texi: Document htm builtins.
9063         * config.gcc: Add the new header files to extra_headers.
9065 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
9067         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
9068         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
9070 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
9071             Pat Haugen  <pthaugen@us.ibm.com>
9072             Peter Bergner  <bergner@vnet.ibm.com>
9074         * config/rs6000/power8.md: New.
9075         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
9076         setting for power8 entry.
9077         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
9078         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
9079         test for Power4/Power5 only.
9080         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
9081         support.
9082         (force_new_group): Adjust comment.
9083         * config/rs6000/rs6000.md: Include power8.md.
9085 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
9087         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
9088         * config/arm/arm-protos.h (arm_max_conditional_execute): New
9089         declaration.
9090         (tune_params): Update comment.
9091         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
9092         (arm_max_conditional_execute): New function.
9093         (thumb2_final_prescan_insn): Use max_insn_skipped and
9094         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
9096 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
9098         PR tree-optimization/57705
9099         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
9100         SSA_NAME step, provided that it is not defined inside the loop.
9101         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
9102         (get_initial_def_for_induction): Handle SSA_NAME IV step.
9104 2013-06-25  Martin Jambor  <mjambor@suse.cz>
9106         PR middle-end/57670
9107         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
9108         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
9109         calls in the dump.
9110         (ipa_note_param_call): Initialize member_ptr flag.
9111         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
9112         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
9113         (ipa_write_indirect_edge_info): Stream member_ptr flag.
9114         (ipa_read_indirect_edge_info): Likewise.
9116 2013-06-25  Richard Biener  <rguenther@suse.de>
9118         PR middle-end/56977
9119         * passes.c (init_optimization_passes): Move pass_fold_builtins
9120         and pass_dce earlier with -Og.
9122 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
9124         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
9125         <BIT_FIELD_REF>: Remove trailing TAB.
9126         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
9127         remove blank line.
9129 2013-06-24  Martin Jambor  <mjambor@suse.cz>
9131         PR tree-optimization/57358
9132         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
9133         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
9134         (ipa_analyze_params_uses): Generate pessimistic info when true.
9136 2013-06-24  Martin Jambor  <mjambor@suse.cz>
9138         PR tree-optimization/57539
9139         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
9140         global.inlined_to of the new node to it.  All callers changed.
9141         * ipa-inline-transform.c (clone_inlined_nodes): New variable
9142         inlining_into, pass it to cgraph_clone_node.
9143         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
9144         ipa_free_edge_args_substructures.
9145         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
9146         rdesc linked list.  Do not assert rdesc edges have inlined caller.
9147         Assert we have found an rdesc in the rdesc list.
9149 2013-06-24  Richard Biener  <rguenther@suse.de>
9151         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
9152         (pointer_set_lookup): Declare.
9153         (class pointer_map): New template class implementing a
9154         generic pointer to T map.
9155         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
9156         pointer_map<T>::contains, pointer_map<T>::insert,
9157         pointer_map<T>::traverse): New functions.
9158         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
9159         (pointer_set_lookup): New function.
9160         (pointer_set_contains): Use pointer_set_lookup.
9161         (pointer_set_insert): Likewise.
9162         (insert_aux): Remove.
9163         (struct pointer_map_t): Embed a pointer_set_t.
9164         (pointer_map_create): Adjust.
9165         (pointer_map_destroy): Likewise.
9166         (pointer_map_contains): Likewise.
9167         (pointer_map_insert): Likewise.
9168         (pointer_map_traverse): Likewise.
9169         * tree-streamer.h (struct streamer_tree_cache_d): Use a
9170         pointer_map<unsigned> instead of a pointer_map_t.
9171         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
9172         (streamer_tree_cache_lookup): Likewise.
9173         (streamer_tree_cache_create): Likewise.
9174         (streamer_tree_cache_delete): Likewise.
9175         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
9176         pointer_map<unsigned> instead of a pointer_map_t.
9177         (lto_init_tree_ref_encoder): Adjust.
9178         (lto_destroy_tree_ref_encoder): Likewise.
9179         * lto-section-out.c (lto_output_decl_index): Likewise.
9180         (lto_record_function_out_decl_state): Likewise.
9181         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
9183 2013-06-24  Richard Biener  <rguenther@suse.de>
9185         PR tree-optimization/57488
9186         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
9188 2013-06-24  Alan Modra  <amodra@gmail.com>
9190         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
9191         (gen_easy_altivec_constant): Likewise.
9192         * config/rs6000/predicates.md (easy_vector_constant_add_self,
9193         easy_vector_constant_msb): Likewise.
9195 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
9197         PR target/57688
9198         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
9199         add missing return true.
9201 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
9203         PR target/52483
9204         * config/sh/predicates.md (general_extend_operand): Invoke
9205         general_movsrc_operand for memory operands.
9206         (general_movsrc_operand): Allow reg+reg addressing, do not use
9207         general_operand for memory operands.
9209 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
9211         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
9212         when current target options does not apply.
9213         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
9214         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
9215         * config/i386/bmiintrin.h: Pass appropriate target
9216         attributes to header.
9217         * config/i386/mmintrin.h: Ditto.
9218         * config/i386/nmmintrin.h: Ditto.
9219         * config/i386/avx2intrin.h: Ditto.
9220         * config/i386/fxsrintrin.h: Ditto.
9221         * config/i386/tbmintrin.h: Ditto.
9222         * config/i386/xsaveintrin.h: Ditto.
9223         * config/i386/f16cintrin.h: Ditto.
9224         * config/i386/xtestintrin.h: Ditto.
9225         * config/i386/xsaveoptintrin.h: Ditto.
9226         * config/i386/bmi2intrin.h: Ditto.
9227         * config/i386/lzcntintrin.h: Ditto.
9228         * config/i386/smmintrin.h: Ditto.
9229         * config/i386/wmmintrin.h: Ditto.
9230         * config/i386/x86intrin.h: Remove all header include guards.
9231         * config/i386/prfchwintrin.h: Ditto.
9232         * config/i386/pmmintrin.h: Ditto.
9233         * config/i386/tmmintrin.h: Ditto.
9234         * config/i386/xmmintrin.h: Ditto.
9235         * config/i386/popcntintrin.h: Ditto.
9236         * config/i386/rdseedintrin.h: Ditto.
9237         * config/i386/ammintrin.h: Ditto.
9238         * config/i386/emmintrin.h: Ditto.
9239         * config/i386/immintrin.h: Remove all header include guards.
9240         * config/i386/fma4intrin.h: Ditto.
9241         * config/i386/lwpintrin.h: Ditto.
9242         * config/i386/xopintrin.h: Ditto.
9243         * config/i386/ia32intrin.h: Ditto.
9244         * config/i386/avxintrin.h: Ditto.
9245         * config/i386/rtmintrin.h: Ditto.
9246         * config/i386/fmaintrin.h: Ditto.
9247         * config/i386/mm3dnow.h: Ditto.
9249 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
9251         * common/config/i386/i386-common.c: Handle LZCNT.
9253 2013-06-22  Andi Kleen  <ak@linux.intel.com>
9255         * doc/extend.texi: Use __atomic_store_n instead of
9256         __atomic_store in HLE example.
9258 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
9260         * config/sh/sh.c: Remove <cstdlib> workaround.
9262 2013-06-21  Andi Kleen  <ak@linux.intel.com>
9264         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
9266 2013-06-21  Andi Kleen  <ak@linux.intel.com>
9268         * doc/extend.texi: Document that __atomic_clear and
9269           __atomic_test_and_set should only be used with bool.
9271 2013-06-20  Jan Hubicka  <jh@suse.cz>
9273         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
9274         types_same_for_odr.
9275         * tree.c (decls_same_for_odr): New function.
9276         (same_for_edr): New function.
9277         (types_same_for_odr): New function.
9278         (get_binfo_at_offset): Use it.
9279         * tree.h (types_same_for_odr): Declare.
9281 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
9282             Jason Merrill  <jason@redhat.com>
9284         * system.h: Include <cstdlib> as well as <stdlib.h>.
9286 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
9288         PR target/57655
9289         * config/i386/i386.c (construct_container): Report error if
9290         long double is used with disabled x87 float returns.
9292 2013-06-20  Jan Hubicka  <jh@suse.cz>
9294         * lto-cgraph.c (input_symtab): Do not set cgraph state.
9296 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
9298         PR rtl-optimization/57425
9299         PR rtl-optimization/57569
9300         * alias.c (write_dependence_p): Remove parameters mem_mode and
9301         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
9302         Changed all callers.
9303         (canon_anti_dependence): Get comments and semantics in sync.
9304         Add parameter mem_canonicalized.  Changed all callers.
9305         * rtl.h (canon_anti_dependence): Update prototype.
9307 2013-06-20  Richard Biener  <rguenther@suse.de>
9309         * data-streamer-in.c (streamer_read_uhwi): Optimize single
9310         byte case, inline streamer_read_uchar and defer section
9311         overrun check.
9313 2013-06-20  Richard Biener  <rguenther@suse.de>
9315         PR tree-optimization/57584
9316         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
9317         SSA names into the expanded expression that take part in
9318         abnormal coalescing.
9320 2013-06-19  Sharad Singhai  <singhai@google.com>
9322         * gcov.c (print_usage): Handle new option.
9323         (process_args): Ditto.
9324         (get_gcov_intermediate_filename): New function.
9325         (output_intermediate_file): New function.
9326         (output_gcov_file): New function
9327         (generate_results): Handle new option.
9328         (release_function): Relase demangled name.
9329         (read_graph_file): Handle demangled name.
9330         (output_lines): Ditto.
9331         * doc/gcov.texi: Document gcov intermediate format.
9333 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
9335         PR bootstrap/57604
9336         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
9337         (lra_emit_add): Use the functions.  Add comment about Y as an
9338         address segment.
9340 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
9342         PR driver/57652
9343         * collect2.c (collect_atexit): New.
9344         (collect_exit): Delete.
9345         (main): Register collect_atexit with atexit.
9346         (collect_wait): Change collect_exit to exit.
9347         (do_wait): Same.
9348         * collect2.h (collect_exit): Delete.
9349         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
9351 2013-06-19  Wei Mi  <wmi@google.com>
9353         PR rtl-optimization/57518
9354         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
9355         if regno is used in paradoxical subreg.
9356         (update_equiv_regs): Check pdx_subregs[regno] before
9357         set a reg to be equivalent with a mem.
9359 2013-06-19  Matthias Klose  <doko@ubuntu.com>
9361         PR driver/57651
9362         * file-find.h (find_a_file): Add a mode parameter.
9363         * file-find.c (find_a_file): Likewise.
9364         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
9365         with X_OK for the executables.
9366         * collect2.c (main): Call find_a_file with X_OK.
9368 2013-06-19  Steve Ellcey  <sellcey@mips.com>
9370         PR target/56942
9371         * config/mips/mips.md (casesi_internal_mips16_<mode>):
9372         Use NEXT_INSN instead of next_real_insn.
9374 2013-06-19  Jan Hubicka  <jh@suse.cz>
9376         * cgraph.h (const_value_known_p): Replace by ...
9377         (ctor_for_folding): .. this one.
9378         * cgraphunit.c (process_function_and_variable_attributes): Use it.
9379         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
9380         * expr.c (expand_expr_real_1): Likewise.
9381         (string_constant): Likewise.
9382         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
9383         * ipa.c (process_references): Likewise.
9384         (symtab_remove_unreachable_nodes): Likewise.
9385         * ipa-inline-analysis.c (param_change_prob): Likewise.
9386         * gimple-fold.c (canonicalize_constructor_val): Likewise.
9387         (get_base_constructor): Likwise.
9388         * varpool.c (varpool_remove_node): Likewise.
9389         (varpool_remove_initializer): LIkewise.
9390         (dump_varpool_node): LIkwise.
9391         (const_value_known_p): Rewrite to ...
9392         (ctor_for_folding): ... this one.
9394 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
9396         PR driver/57651
9397         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
9398         PERSONALITY in $PATH derived prefixes.
9400 2013-06-19  Jeff Law  <law@redhat.com>
9402         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
9403         in comment.
9405         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
9406         (simplify_bitwise_binary): Use it to simpify certain binary ops on
9407         booleans.
9409 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
9411         * config/arm/vfp.md: Move VFP instruction classification documentation
9412         to ...
9413         * config/arm/arm.md: ... here.  Update instruction classification
9414         documentation.
9416 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
9418         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
9419         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
9420         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
9422 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
9424         PR target/57609
9425         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
9426         with NEXT_INSN.  Use tablejump_p to check for jump table data
9427         insns.
9429 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
9431         PR c++/56544
9432         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
9433         that now in C++ the value is correct per the C++ standards.
9435 2013-06-19  Richard Biener  <rguenther@suse.de>
9437         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
9438         for global context.
9440 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9442         Revert:
9443         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9445         PR target/57609
9446         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
9447         with next_active_insn.
9449 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
9451         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
9452         functions are inlined during failures to flag an error.
9453         * tree-inline.c (expand_call_inline): Allow the error to be flagged
9454         in early inline pass.
9456 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
9458         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
9459         in comments.
9461 2013-06-18  Julian Brown  <julian@codesourcery.com>
9463         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
9464         Permit virtual register pre-reload if !strict.
9465         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
9466         change.
9467         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
9468         prototype.
9469         * config/arm/neon.md (movmisalign<mode>): Use
9470         neon_perm_struct_or_reg_operand instead of
9471         neon_struct_or_register_operand.
9472         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
9473         neon_permissive_struct_operand instead of neon_struct_operand.
9474         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
9475         neon_vector_mem_operand.
9476         * config/arm/predicates.md (neon_struct_operand): Adjust call to
9477         neon_vector_mem_operand.
9478         (neon_permissive_struct_operand): New.
9479         (neon_struct_or_register_operand): Rename to...
9480         (neon_perm_struct_or_reg_operand): This. Adjust call to
9481         neon_vector_mem_operand.
9483 2013-06-18  Richard Biener  <rguenther@suse.de>
9485         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
9486         * lto-streamer.h: Include pointer-set.h.
9487         (struct lto_decl_slot): Remove.
9488         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
9489         Remove next_index entry.
9490         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
9491         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
9492         (lto_init_tree_ref_encoder): Adjust.
9493         (lto_destroy_tree_ref_encoder): Likewise.
9494         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
9495         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
9496         (lto_output_decl_index): Adjust.
9497         (lto_new_out_decl_state): Likewise.
9498         (lto_record_function_out_decl_state): Likewise.
9499         * lto-streamer-out.c (copy_function): Likewise.
9501 2013-06-18  Richard Biener  <rguenther@suse.de>
9503         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
9504         * cgraphunit.c: Include cfgloop.h.
9505         (init_lowered_empty_function): Initialize the loop tree.
9506         (assemble_thunk): Insert new BBs into loops.
9508 2013-06-18  Richard Biener  <rguenther@suse.de>
9510         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
9511         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
9512         the map from cache entry to cache index optional.
9513         (streamer_tree_cache_replace_tree): Adjust accordingly.
9514         (streamer_tree_cache_append): Likewise.
9515         (streamer_tree_cache_delete): Likewise.
9516         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
9517         streamer cache map from cache entry to cache index.
9518         * lto-streamer-out.c (create_output_block): Adjust.
9520 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
9522         * config/arm/arm.md (attribute "insn"): Move multiplication and
9523         division attributes to...
9524         (attribute "type"): ... here.  Remove mult.
9525         (attribute "mul32"): New attribute.
9526         (attribute "mul64"): Add umaal.
9527         (*arm_mulsi3): Update attributes.
9528         (*arm_mulsi3_v6): Likewise.
9529         (*thumb_mulsi3): Likewise.
9530         (*thumb_mulsi3_v6): Likewise.
9531         (*mulsi3_compare0): Likewise.
9532         (*mulsi3_compare0_v6): Likewise.
9533         (*mulsi_compare0_scratch): Likewise.
9534         (*mulsi_compare0_scratch_v6): Likewise.
9535         (*mulsi3addsi): Likewise.
9536         (*mulsi3addsi_v6): Likewise.
9537         (*mulsi3addsi_compare0): Likewise.
9538         (*mulsi3addsi_compare0_v6): Likewise.
9539         (*mulsi3addsi_compare0_scratch): Likewise.
9540         (*mulsi3addsi_compare0_scratch_v6): Likewise.
9541         (*mulsi3subsi): Likewise.
9542         (*mulsidi3adddi): Likewise.
9543         (*mulsi3addsi_v6): Likewise.
9544         (*mulsidi3adddi_v6): Likewise.
9545         (*mulsidi3_nov6): Likewise.
9546         (*mulsidi3_v6): Likewise.
9547         (*umulsidi3_nov6): Likewise.
9548         (*umulsidi3_v6): Likewise.
9549         (*umulsidi3adddi): Likewise.
9550         (*umulsidi3adddi_v6): Likewise.
9551         (*smulsi3_highpart_nov6): Likewise.
9552         (*smulsi3_highpart_v6): Likewise.
9553         (*umulsi3_highpart_nov6): Likewise.
9554         (*umulsi3_highpart_v6): Likewise.
9555         (mulhisi3): Likewise.
9556         (*mulhisi3tb): Likewise.
9557         (*mulhisi3bt): Likewise.
9558         (*mulhisi3tt): Likewise.
9559         (maddhisi4): Likewise.
9560         (*maddhisi4tb): Likewise.
9561         (*maddhisi4tt): Likewise.
9562         (maddhidi4): Likewise.
9563         (*maddhidi4tb): Likewise.
9564         (*maddhidi4tt): Likewise.
9565         (divsi3): Likewise.
9566         (udivsi3): Likewise.
9567         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
9568         (thumb2_mulsi_short_compare0): Likewise.
9569         (thumb2_mulsi_short_compare0_scratch): Likewise.
9570         * config/arm/arm1020e.md (1020mult1): Update attribute change.
9571         (1020mult2): Likewise.
9572         (1020mult3): Likewise.
9573         (1020mult4): Likewise.
9574         (1020mult5): Likewise.
9575         (1020mult6): Likewise.
9576         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
9577         change.
9578         (cortex_a15_mult64): Likewise.
9579         (cortex_a15_sdiv): Likewise.
9580         (cortex_a15_udiv): Likewise.
9581         * config/arm/arm1026ejs.md (mult1): Update attribute change.
9582         (mult2): Likewise.
9583         (mult3): Likewise.
9584         (mult4): Likewise.
9585         (mult5): Likewise.
9586         (mult6): Likewise.
9587         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
9588         (pj4_ir_div): Likewise.
9589         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
9590         (11_mult2): Likewise.
9591         (11_mult3): Likewise.
9592         (11_mult4): Likewise.
9593         (11_mult5): Likewise.
9594         (11_mult6): Likewise.
9595         (11_mult7): Likewise.
9596         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
9597         (cortex_a8_mla): Likewise.
9598         (cortex_a8_mull): Likewise.
9599         (cortex_a8_smulwy): Likewise.
9600         (cortex_a8_smlald): Likewise.
9601         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
9602         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
9603         (cortex_r4_mul_3): Likewise.
9604         (cortex_r4_mla_4): Likewise.
9605         (cortex_r4_mla_3): Likewise.
9606         (cortex_r4_smlald): Likewise.
9607         (cortex_r4_mull): Likewise.
9608         (cortex_r4_sdiv): Likewise.
9609         (cortex_r4_udiv): Likewise.
9610         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
9611         (cortex_a7_idiv): Likewise.
9612         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
9613         (9_mult2): Likewise.
9614         (9_mult3): Likewise.
9615         (9_mult4): Likewise.
9616         (9_mult5): Likewise.
9617         (9_mult6): Likewise.
9618         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
9619         (cortex_a53_sdiv): Likewise.
9620         (cortex_a53_udiv): Likewise.
9621         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
9622         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
9623         (mp626_mult2): Likewise.
9624         (mp626_mult3): Likewise.
9625         (mp626_mult4): Likewise.
9626         * config/arm/fa526.md (526_mult1): Update attribute change.
9627         (526_mult2): Likewise.
9628         * config/arm/arm-generic.md (mult): Update attribute change.
9629         (mult_ldsched_strongarm): Likewise.
9630         (mult_ldsched): Likewise.
9631         (multi_cycle): Likewise.
9632         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
9633         * config/arm/fa606te.md (606te_mult1): Update attribute change.
9634         (606te_mult2): Likewise.
9635         (606te_mult3): Likewise.
9636         (606te_mult4): Likewise.
9637         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
9638         (cortex_a9_mac16): Likewise.
9639         (cortex_a9_multiply): Likewise.
9640         (cortex_a9_mac): Likewise.
9641         (cortex_a9_multiply_long): Likewise.
9642         * config/arm/fa626te.md (626te_mult1): Update attribute change.
9643         (626te_mult2): Likewise.
9644         (626te_mult3): Likewise.
9645         (626te_mult4): Likewise.
9647 2013-06-18  Richard Biener  <rguenther@suse.de>
9649         PR lto/57334
9650         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
9652 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9654         PR target/57609
9655         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
9656         with next_active_insn.
9658 2013-06-18  Alan Modra  <amodra@gmail.com>
9660         * config/rs6000/rs6000.h (enum data_align): New.
9661         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
9662         (DATA_ABI_ALIGNMENT): Define.
9663         (CONSTANT_ALIGNMENT): Correct comment.
9664         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
9665         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
9667 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
9669         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
9670         ATTRIBUTE_UNUSED marking.
9672 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
9674         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
9675         alternative and update.
9676         (aarch64_dup_lanedi): Delete.
9677         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
9678         * config/aarch64/aarch64-simd-builtins.def: Update.
9680 2013-06-17  Richard Biener  <rguenther@suse.de>
9682         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
9683         (lto_input_scc): Declare.
9684         (lto_input_tree_1): Likewise.
9685         (struct lto_stats_d): Add num_tree_bodies_output and
9686         num_pickle_refs_output.
9687         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
9688         (lto_read_tree_1): Split out from ...
9689         (lto_read_tree): ... this.
9690         (lto_input_scc): New function.
9691         (lto_input_tree_1): Split out from ...
9692         (lto_input_tree): ... this.  Handle LTO_tree_scc.
9693         (lto_data_in_create): Create the streamer cache without hashes.
9694         * lto-streamer-out.c (create_output_block): Create the streamer
9695         cache with hashes when not doing WPA.
9696         (lto_write_tree_1): Split out from ...
9697         (lto_write_tree): ... this.
9698         (get_symbol_initial_value): New function.
9699         (lto_output_tree_1): Split out from ...
9700         (lto_output_tree): ... this.  Write trees as series of SCCs
9701         using a DFS walk via DFS_write_tree.
9702         (struct sccs, struct scc_entry): New types.
9703         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
9704         (DFS_write_tree_body): New function.
9705         (DFS_write_tree): Likewise.
9706         (hash_tree): Likewise.
9707         (scc_entry_compare): Likewise.
9708         (hash_scc): Likewise.
9709         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
9710         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
9711         TREE_CHAIN as regular reference.
9712         (streamer_read_integer_cst): Remove.
9713         (streamer_get_pickled_tree): Adjust.
9714         * tree-streamer-out.c (streamer_write_chain): Disable streaming
9715         of DECL_EXTERNALs in BLOCK_VARS for now.
9716         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
9717         reference.
9718         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
9719         Add hash value argument and record that if hashes are recorded
9720         in the cache.
9721         (streamer_tree_cache_insert_1): Adjust.
9722         (streamer_tree_cache_insert): Likewise.
9723         (streamer_tree_cache_insert_at): Rename to ...
9724         (streamer_tree_cache_replace_tree): ... this and adjust.
9725         (streamer_tree_cache_append): Adjust.
9726         (record_common_node): Likewise.
9727         (streamer_tree_cache_create): Add argument whether to
9728         record hash values together with trees.
9729         (streamer_tree_cache_delete): Adjust.
9730         * tree-streamer.h (struct streamer_tree_cache_d): Add
9731         vector of hashes.
9732         (streamer_read_integer_cst): Remove.
9733         (streamer_tree_cache_insert): Adjust.
9734         (streamer_tree_cache_append): Likewise.
9735         (streamer_tree_cache_insert_at): Rename to ...
9736         (streamer_tree_cache_replace_tree): ... this and adjust.
9737         (streamer_tree_cache_create): Add argument whether to record hashes.
9738         (streamer_tree_cache_get): Rename to ...
9739         (streamer_tree_cache_get_tree): ... this.
9740         (streamer_tree_cache_get_hash): New function.
9741         * tree.c (cache_integer_cst): New function.
9742         * tree.h (cache_integer_cst): Declare.
9743         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
9744         * lto-symtab.c (lto_varpool_replace_node): Only release
9745         DECL_INITIAL of non-prevailing decls.
9746         * varpool.c (varpool_remove_initializer): Do not release
9747         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
9749 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
9751         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
9752         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
9753         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
9754         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
9755         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
9756         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
9757         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
9758         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
9759         instead of TARGET_64BIT.
9760         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
9761         Require ISA_HAS_<D>DIV.
9763 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
9765         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
9766         (mips*-*-linux*): Move default with_llsc setting to where other
9767         defaults are set.
9768         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
9769         with_arch block.
9770         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
9771         Likewise.  Remove default with_tune setting.  Move default float
9772         setting to its own block.  Handle with_llsc in the same block as above.
9774 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
9776         PR rtl-optimization/57425
9777         PR rtl-optimization/57569
9778         * alias.c (write_dependence_p): Add new parameters mem_mode,
9779         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
9780         Changed all callers.
9781         (canon_anti_dependence): New function.
9782         * cse.c (check_dependence): Use canon_anti_dependence.
9783         * cselib.c (cselib_invalidate_mem): Likewise.
9784         * rtl.h (canon_anti_dependence): Declare.
9786 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
9788         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
9789         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
9790         ".set mips3" for 64-bit targets.
9792 2013-06-15  Dehao Chen  <dehao@google.com>
9794         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
9795         * gimple-low.c (gimple_check_call_matching_types): Likewise.
9796         (gimple_check_call_args): Likewise.
9797         * value-prof.c (check_ic_target): Likewise.
9798         * ipa-inline.c (early_inliner): Likewise.
9799         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
9800         * cgraph.c (cgraph_create_edge_1): Likewise.
9801         (cgraph_make_edge_direct): Likewise.
9803 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
9805         PR target/57615
9806         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
9807         rs6000_output_move_128bit to handle emitting quad memory
9808         operations.  Set attribute length to 8 bytes.
9810 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
9812         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
9813         New pattern.
9814         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
9815         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
9816         (aarch64_<su>mlsl<mode>): Likewise.
9818 2013-06-14  Mike Stump  <mikestump@comcast.net>
9820         * Makefile.in (TARGET_H): Add insn-codes.h.
9822 2013-06-14  Alan Modra  <amodra@gmail.com>
9824         PR middle-end/57134
9825         PR middle-end/57586
9826         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
9827         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
9828         bitfield expansion when EXPAND_MEMORY.
9829         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
9831 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
9833         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
9834         test for clearing quad memory on 32-bit later.
9836 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
9838         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
9839         (fold_negate_expr): Likewise.
9840         (fold_real_zero_addition_p): Handle vectors.
9841         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
9843 2013-06-14  Alan Modra  <amodra@gmail.com>
9845         * varasm.c (force_const_mem): Revert 2013-06-07 change.
9847 2013-06-13  Jan Hubicka  <jh@suse.cz>
9849         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
9850         Local comdats are not externally visible.
9851         * symtab.c (dump_symtab_base): Dump externally visible.
9852         (verify_symtab_base): Verify back links in the symtab hash.
9854 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
9856         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
9857         CONVERT_EXPR as equal nodes.
9859 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
9861         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
9863 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
9865         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
9866         Generalize to complex and vector.
9867         * tree.c (build_all_ones_cst): New function.
9868         * tree.h (build_all_ones_cst): Declare it.
9870 2013-06-13  Alan Modra  <amodra@gmail.com>
9872         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
9873         * config/rs6000/rs6000.md (signbittf2): New insn.
9874         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
9875         (abstf2_internal, cmptf_internal2): Likewise.
9876         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
9878 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
9879             Pat Haugen  <pthaugen@us.ibm.com>
9880             Peter Bergner  <bergner@vnet.ibm.com>
9882         * config/rs6000/rs6000.c (emit_load_locked): Add support for
9883         power8 byte, half-word, and quad-word atomic instructions.
9884         (emit_store_conditional): Likewise.
9885         (rs6000_expand_atomic_compare_and_swap): Likewise.
9886         (rs6000_expand_atomic_op): Likewise.
9888         * config/rs6000/sync.md (larx): Add new modes for power8.
9889         (stcx): Likewise.
9890         (AINT): New mode iterator to include TImode as well as normal
9891         integer modes on power8.
9892         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
9893         that VSX registers are not considered.  Use AINT mode iterator
9894         instead of INT1 to allow inclusion of quad word atomic operations
9895         on power8.
9896         (load_locked<mode>): Likewise.
9897         (store_conditional<mode>): Likewise.
9898         (atomic_compare_and_swap<mode>): Likewise.
9899         (atomic_exchange<mode>): Likewise.
9900         (atomic_nand<mode>): Likewise.
9901         (atomic_fetch_<fetchop_name><mode>): Likewise.
9902         (atomic_nand_fetch<mode>): Likewise.
9903         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
9904         each type.
9905         (ATOMIC): On power8, add QImode, HImode modes.
9906         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
9907         modes that promote to SImode.
9908         (load_lockedti): Convert TImode arguments to PTImode, so that we
9909         get a guaranteed even/odd register pair.
9910         (load_lockedpti): Likewise.
9911         (store_conditionalti): Likewise.
9912         (store_conditionalpti): Likewise.
9914         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
9915         atomic load/store instructions.
9916         (HSI): Likewise.
9918 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
9920         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
9921         loads.
9922         (insn_count): New attribute, with most cases extracted from...
9923         (length): ...here.  Redefine most cases in terms of insn_count.
9924         (single_insn): Delete.
9925         (can_delay): Use insn_count to check for single instructions.
9926         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
9927         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
9928         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
9929         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
9930         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
9931         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
9932         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
9933         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
9934         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
9935         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
9936         rather than "length".
9937         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
9938         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
9939         Use "insn_count" rather than "length".
9940         * config/mips/mips-dsp.md
9941         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
9942         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
9943         length attributes.
9945 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
9947         PR tree-optimization/57361
9948         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
9950 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
9952         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
9953         to split.
9954         (aarch64_simd_combine<mode>): New instruction expansion.
9955         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
9956         function prototype.
9957         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
9958         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
9960 2013-06-12  Jan Hubicka  <jh@suse.cz>
9962         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
9963         decl has when in streaming stage.
9964         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
9965         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
9967 2013-06-12  Roland Stigge  <stigge@antcom.de>
9969         PR target/57578
9970         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
9972 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
9974         PR tree-optimization/57537
9975         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
9976         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
9978 2013-06-12  Richard Biener  <rguenther@suse.de>
9980         * data-streamer.h (streamer_write_char_stream): CSE
9981         obs->current_pointer.
9982         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
9983         streamer_write_char_stream manually and optimize the resulting loop.
9984         (streamer_write_hwi_stream): Likewise.
9986 2013-06-12  Jan Hubicka  <jh@suse.cz>
9988         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
9989         * cgraph.h (varpool_create_empty_node): Declare.
9990         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
9991         duplicated nodes.
9992         * symtab.c (symtab_unregister_node): Be lax about missin entries
9993         in node hash.
9994         (symtab_get_node): Update comment.
9995         * varpool.c (varpool_create_empty_node): Break out from ...
9996         (varpool_node_for_decl): ... here.
9997         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
9999 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
10001         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
10002         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
10003         part.  Use straight-line flow at the end.
10004         <COMPONENT_REF>: Remove superfluous else.
10005         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
10007 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
10009         PR target/56564
10010         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
10011         target hook even for !TREE_PUBLIC decls.  If no resolution info
10012         is available, return false for common and external decls.
10014 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
10016         * config/rl78/constraints.md (U): New constraint.
10017         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
10018         valloc attribute.
10020 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
10022         PR target/57589
10023         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
10024         to allow returning address to AT_PLATFORM name.
10026 2013-06-11  Jan Hubicka  <jh@suse.cz>
10028         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
10029         * cgraph.h (symtab_node_base): Add weakref flag.
10030         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
10031         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
10032         (output_weakrefs): Use weakref flag.
10033         * fold-const.c (simple_operand_p): Handle WEAK.
10034         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
10035         * ipa.c (varpool_externally_visible_p): Drop weakref.
10036         (function_and_variable_visibility): Update comment; fix weakref
10037         sanity checks; do not clear DECL_WEAK on them.
10038         * lto-cgraph.c (lto_output_node): update.
10039         (lto_output_varpool_node): Update.
10040         (input_overwrite_node): Update.
10041         (input_node): Update.
10042         (input_varpool_node): Update.
10043         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
10044         (lto_symtab_merge_symbols): Add sanity check.
10045         (lto_symtab_prevailing_decl): Do not special case weakrefs.
10046         * passes.c (rest_of_decl_compilation): Set static flag, too.
10047         * symtab.c (dump_symtab_base): Dump weakref.
10048         (verify_symtab_base): Sanity check weakrefs.
10049         (symtab_make_decl_local): Remove duplicated code.
10050         (symtab_alias_ultimate_target): Simplify.
10051         * varpool.c (varpool_create_variable_alias): Set weakref flag.
10053 2013-06-11  Tom de Vries  <tom@codesourcery.com>
10055         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
10056         sequence_vect == NULL.
10058 2013-06-11  DJ Delorie  <dj@redhat.com>
10060         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
10061         (rl78_unwind_word_mode): New.
10063 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
10065         * final.c (debug_prefix_maps): Make static.
10067 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
10069         * function.c (initial_trampoline): Remove stray copy.
10071 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
10073         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
10075 2013-06-11  Martin Jambor  <mjambor@suse.cz>
10077         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
10078         within bounds at the beginning of the function.
10080 2013-06-11  Alan Modra  <amodra@gmail.com>
10082         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
10083         reporting.
10084         (get_named_section): Don't NULL !DECL_P decl.
10086 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
10088         * doc/invoke.texi (core-avx2): Document.
10089         (slm): Likewise.
10090         (atom): Updated with MOVBE.
10092 2013-06-11  Richard Biener  <rguenther@suse.de>
10094         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
10096 2013-06-11  Anton Blanchard  <anton@samba.org>
10098         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
10099         correct shift value in little-endian mode.
10101 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
10103         PR target/56564
10104         * varasm.c (get_variable_align): Move #endif to the right place.
10106 2013-06-10  Cary Coutant  <ccoutant@google.com>
10108         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
10109         for hash so that hash table traversal order is deterministic.
10111 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
10112             Pat Haugen  <pthaugen@us.ibm.com>
10113             Peter Bergner  <bergner@vnet.ibm.com>
10115         * config/rs6000/vector.md (GPR move splitter): Do not split moves
10116         of vectors in GPRS if they are direct moves or quad word load or
10117         store moves.
10119         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
10120         declaration.
10121         (direct_move_p): Likewise.
10122         (quad_load_store_p): Likewise.
10124         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
10125         classes into bins based on the physical register type.
10126         (reg_class_to_reg_type): Likewise.
10127         (IS_STD_REG_TYPE): Likewise.
10128         (IS_FP_VECT_REG_TYPE): Likewise.
10129         (reload_fpr_gpr): Arrays to determine what insn to use if we can
10130         use direct move instructions.
10131         (reload_gpr_vsx): Likewise.
10132         (reload_vsx_gpr): Likewise.
10133         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
10134         information that is a simplification of register classes.  Also
10135         precalculate direct move reload helpers.
10136         (direct_move_p): New function to return true if the operation can
10137         be done as a direct move instruciton.
10138         (quad_load_store_p): New function to return true if the operation
10139         is a quad memory operation.
10140         (rs6000_legitimize_address): If quad memory, only allow register
10141         indirect for TImode addresses.
10142         (rs6000_legitimate_address_p): Likewise.
10143         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
10144         (rs6000_reload_register_type): Likewise.
10145         (register_to_reg_type): Return register type.
10146         (rs6000_secondary_reload_simple_move): New helper function for
10147         secondary reload and secondary memory needed to identify anything
10148         that is a simple move, and does not need reloading.
10149         (rs6000_secondary_reload_direct_move): New helper function for
10150         secondary reload to identify cases that can be done with several
10151         instructions via the direct move instructions.
10152         (rs6000_secondary_reload_move): New helper function for secondary
10153         reload to identify moves between register types that can be done.
10154         (rs6000_secondary_reload): Add support for quad memory operations
10155         and for direct move.
10156         (rs6000_secondary_memory_needed): Likewise.
10157         (rs6000_debug_secondary_memory_needed): Change argument names.
10158         (rs6000_output_move_128bit): New function to return the move to
10159         use for 128-bit moves, including knowing about the various
10160         limitations of quad memory operations.
10162         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
10163         memory operations.  call rs6000_output_move_128bit for the actual
10164         instruciton(s) to generate.
10165         (vsx_movti_64bit): Likewise.
10167         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
10168         (UNSPEC_P8V_MTVSRWZ): Likewise.
10169         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
10170         (UNSPEC_P8V_MTVSRD): Likewise.
10171         (UNSPEC_P8V_XXPERMDI): Likewise.
10172         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
10173         (UNSPEC_FUSION_GPR): Likewise.
10174         (FMOVE128_GPR): New iterator for direct move.
10175         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
10176         (f32_sv): Likewise.
10177         (f32_dm): Likewise.
10178         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
10179         loads and direct move instructions.
10180         (zero_extendsidi2_lfiwzx): Likewise.
10181         (extendsidi2_lfiwax): Likewise.
10182         (extendsidi2_nocell): Likewise.
10183         (floatsi<mode>2_lfiwax): Likewise.
10184         (lfiwax): Likewise.
10185         (floatunssi<mode>2_lfiwzx): Likewise.
10186         (lfiwzx): Likewise.
10187         (fix_trunc<mode>_stfiwx): Likewise.
10188         (fixuns_trunc<mode>_stfiwx): Likewise.
10189         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
10190         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
10191         (parity<mode>2_cmpb): Set length/type attr.
10192         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
10193         for 'mr.' to fast_compare.
10194         (bpermd_<mode>): Change type attr to popcnt.
10195         (p8_fmrgow_<mode>): New insns for power8 direct move support.
10196         (p8_mtvsrwz_1): Likewise.
10197         (p8_mtvsrwz_2): Likewise.
10198         (reload_fpr_from_gpr<mode>): Likewise.
10199         (p8_mtvsrd_1): Likewise.
10200         (p8_mtvsrd_2): Likewise.
10201         (p8_xxpermdi_<mode>): Likewise.
10202         (reload_vsx_from_gpr<mode>): Likewise.
10203         (reload_vsx_from_gprsf): Likewise.
10204         (p8_mfvsrd_3_<mode>): LIkewise.
10205         (reload_gpr_from_vsx<mode>): Likewise.
10206         (reload_gpr_from_vsxsf): Likewise.
10207         (p8_mfvsrd_4_disf): Likewise.
10208         (multi-word GPR splits): Do not split direct moves or quad memory
10209         operations.
10211 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
10213         * tree-into-ssa.c (interesting_blocks): Make static.
10215 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
10217         PR target/56564
10218         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
10219         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
10220         Use DATA_ABI_ALIGNMENT for that case instead if defined.
10221         (get_variable_align): New function.
10222         (get_variable_section, emit_bss, emit_common,
10223         assemble_variable_contents, place_block_symbol): Use
10224         get_variable_align instead of DECL_ALIGN.
10225         (assemble_noswitch_variable): Add align argument, use it
10226         instead of DECL_ALIGN.
10227         (assemble_variable): Adjust caller.  Use get_variable_align
10228         instead of DECL_ALIGN.
10229         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
10230         caller.
10231         (DATA_ABI_ALIGNMENT): Define.
10232         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
10233         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
10234         opt is false, only return the psABI mandated alignment increase.
10235         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
10236         (DATA_ABI_ALIGNMENT): ... this.
10237         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
10238         (DATA_ABI_ALIGNMENT): ... this.
10239         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
10240         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
10241         (DATA_ABI_ALIGNMENT): ... this.
10242         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
10243         * doc/tm.texi: Regenerated.
10245 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
10247         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
10248         cmp_code to construct REG_EQUAL note.
10250 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
10252         PR target/57568
10253         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
10254         that operands[2] doesn't overlap with operands[0].
10256 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
10257             Jan Hubicka  <jh@suse.cz>
10259         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
10260         hack to mark symbols as used.
10262 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
10264         PR rtl-optimization/57559
10265         * lra-constraints.c (process_alt_operands): Don't discourage
10266         memory with known offset for offsetable memory constraint.
10267         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
10269 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
10271         * varasm.c (struct oc_local_state): Reorder fields.
10272         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
10273         and adjust accordingly.
10274         (output_constructor): Reorder initialization code and adjust call to
10275         output_constructor_bitfield.
10277 2013-06-07  Jan Hubicka  <jh@suse.cz>
10279         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
10281 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
10283         * tree-object-size.c (unknown): Make const.
10285 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
10287         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
10288         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
10289         for last alternative in the cpu_facility attribute.
10291 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10293         PR target/56315
10294         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
10295         (xordi3): Change operand 2 constraint to arm_xordi_operand.
10296         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
10297         * config/arm/constraints.md (Dg): New constraint.
10298         * config/arm/neon.md (xordi3_neon): Remove.
10299         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
10300         * config/arm/predicates.md (arm_xordi_operand): New predicate.
10302 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10304         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
10305         Clean up alternatives.
10307 2013-06-07  Alan Modra  <amodra@gmail.com>
10309         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
10310         va_list_gpr_size.
10312 2013-06-07  Alan Modra  <amodra@gmail.com>
10314         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
10316 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10318         * config/arm/constraints.md (Df): New constraint.
10319         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
10320         Correct length attribute for last two alternatives.
10322 2013-06-07  Alan Modra  <amodra@gmail.com>
10324         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
10325         override user -mfp-in-toc.
10326         (offsettable_ok_by_alignment): Consider just the current access
10327         rather than the whole object, unless BLKmode.  Handle
10328         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
10329         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
10330         for -mcmodel=medium.
10331         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
10332         override user -mfp-in-toc or -msum-in-toc.  Default to
10333         -mno-fp-in-toc for -mcmodel=medium.
10335 2013-06-06  DJ Delorie  <dj@redhat.com>
10337         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
10338         TARGET_VALID_POINTER_MODE.
10340 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
10341             Pat Haugen  <pthaugen@us.ibm.com>
10342             Peter Bergner  <bergner@vnet.ibm.com>
10344         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
10345         Document new power8 builtins.
10347         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
10348         condition code register, to allow 128-bit logical operations to be
10349         done in the VSX or GPR registers.
10350         (nor<mode>3): Use the canonical form for nor.
10351         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
10352         vclz*, and vpopcnt* vector instructions.
10353         (nand<mode>3): Likewise.
10354         (orc<mode>3): Likewise.
10355         (clz<mode>2): LIkewise.
10356         (popcount<mode>2): Likewise.
10358         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
10359         that only the GPRs are recognized.
10361         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10362         support for new power8 builtins.
10364         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
10365         builtin functions.
10366         (xscvdpspn): Likewise.
10367         (vclz): Likewise.
10368         (vclzb): Likewise.
10369         (vclzh): Likewise.
10370         (vclzw): Likewise.
10371         (vclzd): Likewise.
10372         (vpopcnt): Likewise.
10373         (vpopcntb): Likewise.
10374         (vpopcnth): Likewise.
10375         (vpopcntw): Likewise.
10376         (vpopcntd): Likewise.
10377         (vgbbd): Likewise.
10378         (vmrgew): Likewise.
10379         (vmrgow): Likewise.
10380         (eqv): Likewise.
10381         (eqv_v16qi3): Likewise.
10382         (eqv_v8hi3): Likewise.
10383         (eqv_v4si3): Likewise.
10384         (eqv_v2di3): Likewise.
10385         (eqv_v4sf3): Likewise.
10386         (eqv_v2df3): Likewise.
10387         (nand): Likewise.
10388         (nand_v16qi3): Likewise.
10389         (nand_v8hi3): Likewise.
10390         (nand_v4si3): Likewise.
10391         (nand_v2di3): Likewise.
10392         (nand_v4sf3): Likewise.
10393         (nand_v2df3): Likewise.
10394         (orc): Likewise.
10395         (orc_v16qi3): Likewise.
10396         (orc_v8hi3): Likewise.
10397         (orc_v4si3): Likewise.
10398         (orc_v2di3): Likewise.
10399         (orc_v4sf3): Likewise.
10400         (orc_v2df3): Likewise.
10402         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
10403         allow power8 quad mode in 64-bit.
10404         (rs6000_builtin_vectorized_function): Add support to vectorize
10405         ISA 2.07 count leading zeros, population count builtins.
10406         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
10407         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
10408         (builtin_function_type): Add vgbbd builtin function which takes an
10409         unsigned argument.
10410         (altivec_expand_vec_perm_const): Add support for new power8 merge
10411         instructions.
10413         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
10414         that does not include TImdoe for use with 32-bit.
10415         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
10416         instructions.
10417         (UNSPEC_VSX_CVDPSPN): Likewise.
10418         (vsx_xscvdpspn): Likewise.
10419         (vsx_xscvspdpn): Likewise.
10420         (vsx_xscvdpspn_scalar): Likewise.
10421         (vsx_xscvspdpn_directmove): Likewise.
10422         (vsx_and<mode>3): Split logical operations into 32-bit and
10423         64-bit. Add support to do logical operations on TImode as well as
10424         VSX vector types.  Allow logical operations to be done in either
10425         VSX registers or in general purpose registers in 64-bit mode.  Add
10426         splitters if GPRs were used. For AND, add clobber of CCmode to
10427         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
10428         encoding.
10429         (vsx_and<mode>3_32bit): Likewise.
10430         (vsx_and<mode>3_64bit): Likewise.
10431         (vsx_ior<mode>3): Likewise.
10432         (vsx_ior<mode>3_32bit): Likewise.
10433         (vsx_ior<mode>3_64bit): Likewise.
10434         (vsx_xor<mode>3): Likewise.
10435         (vsx_xor<mode>3_32bit): Likewise.
10436         (vsx_xor<mode>3_64bit): Likewise.
10437         (vsx_one_cmpl<mode>2): Likewise.
10438         (vsx_one_cmpl<mode>2_32bit): Likewise.
10439         (vsx_one_cmpl<mode>2_64bit): Likewise.
10440         (vsx_nor<mode>3): Likewise.
10441         (vsx_nor<mode>3_32bit): Likewise.
10442         (vsx_nor<mode>3_64bit): Likewise.
10443         (vsx_andc<mode>3): Likewise.
10444         (vsx_andc<mode>3_32bit): Likewise.
10445         (vsx_andc<mode>3_64bit): Likewise.
10446         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
10447         and xxlorc instructions.
10448         (vsx_eqv<mode>3_64bit): Likewise.
10449         (vsx_nand<mode>3_32bit): Likewise.
10450         (vsx_nand<mode>3_64bit): Likewise.
10451         (vsx_orc<mode>3_32bit): Likewise.
10452         (vsx_orc<mode>3_64bit): Likewise.
10454         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
10456         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
10457         instruction.
10458         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
10459         (p8_vmrgow): Likewise.
10460         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
10461         GPRs to be split under VSX.
10462         (p8v_clz<mode>2): Add power8 count leading zero support.
10463         (p8v_popcount<mode>2): Add power8 population count support.
10464         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
10465         support.
10467         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
10468         instruction.
10470         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
10471         builtin functions.
10472         (vec_nand): Likewise.
10473         (vec_vclz): Likewise.
10474         (vec_vclzb): Likewise.
10475         (vec_vclzd): Likewise.
10476         (vec_vclzh): Likewise.
10477         (vec_vclzw): Likewise.
10478         (vec_vgbbd): Likewise.
10479         (vec_vmrgew): Likewise.
10480         (vec_vmrgow): Likewise.
10481         (vec_vpopcnt): Likewise.
10482         (vec_vpopcntb): Likewise.
10483         (vec_vpopcntd): Likewise.
10484         (vec_vpopcnth): Likewise.
10485         (vec_vpopcntw): Likewise.
10487 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
10489         PR rtl-optimization/57468
10490         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
10491         spilled pseudos.
10493 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
10495         PR rtl-optimization/57459
10496         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
10497         type when setting live regs.
10499 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
10501         * config/s390/s390.opt (mlra): New option.
10502         * config/s390/s390.c (s390_decompose_address): Check displacement
10503         for all registers for LRA.
10504         (s390_secondary_reload): Don't used secondary reloads for LRA.
10505         (s390_lra_p): New function.
10506         (TARGET_LRA_P): Define.
10507         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
10508         of attribute cpu_facility to zarch for the last alternative.
10509         (*cmpmem_short): Ditto.
10511 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
10513         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
10514         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
10515         (arm_expand_prologue): Likewise.
10517 2013-06-06  Teresa Johnson  <tejohnson@google.com>
10519         PR c++/53743
10520         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
10521         as this is now done by redirect_edge_and_branch_force.
10522         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
10523         barriers, and fix interaction with splitting.
10524         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
10525         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
10526         reflect changes made in the routine.
10527         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
10528         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
10529         since this is called in cfglayout mode, and replace partition fixup
10530         with assert as that is now done by force_nonfallthru_and_redirect.
10531         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
10532         already be marked with region crossing note.
10533         (insert_section_boundary_note): Make non-static, gate on flag
10534         has_bb_partition, rewrite to also check for multiple partitions.
10535         (rest_of_handle_reorder_blocks): Remove call to
10536         insert_section_boundary_note, now done later during free_cfg.
10537         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
10538         * bb-reorder.h (insert_section_boundary_note): Declare.
10539         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
10540         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
10541         invoke insert_section_boundary_note.
10542         (try_redirect_by_replacing_jump): Remove unnecessary
10543         check for region crossing note.
10544         (fixup_partition_crossing): New function.
10545         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
10546         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
10547         in non-cfglayout mode.
10548         (force_nonfallthru_and_redirect): Fixup partition boundaries,
10549         remove old code that tried to do this. Emit barrier correctly
10550         when we are in cfglayout mode.
10551         (last_bb_in_partition): New function.
10552         (rtl_split_edge): Correctly fixup partition boundaries.
10553         (commit_one_edge_insertion): Remove old code that tried to
10554         fixup region crossing edge since this is now handled in
10555         split_block, and set up insertion point correctly since
10556         block may now end in a jump.
10557         (verify_hot_cold_block_grouping): Guard against checking when not in
10558         linearized RTL mode.
10559         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
10560         notes.
10561         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
10562         rtl_verify_flow_info, so not called in cfglayout mode.
10563         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
10564         (fixup_reorder_chain): Remove old code that attempted to fixup region
10565         crossing note as this is now handled in force_nonfallthru_and_redirect.
10566         (duplicate_insn_chain): Don't duplicate switch section notes.
10567         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
10568         note.
10569         * basic-block.h (emit_barrier_after_bb): Declare.
10571 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10573         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
10574         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
10575         arm_usatsihi): Adjust alternatives for arm_restrict_it.
10577 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10579         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
10580         where appropriate.
10581         * config/arm/ldmstm.md: Regenerate.
10583 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10585         * config/arm/sync.md (atomic_loaddi_1):
10586         Disable predication for arm_restrict_it.
10587         (arm_load_exclusive<mode>): Likewise.
10588         (arm_load_exclusivesi): Likewise.
10589         (arm_load_exclusivedi): Likewise.
10590         (arm_load_acquire_exclusive<mode>): Likewise.
10591         (arm_load_acquire_exclusivesi): Likewise.
10592         (arm_load_acquire_exclusivedi): Likewise.
10593         (arm_store_exclusive<mode>): Likewise.
10594         (arm_store_exclusive<mode>): Likewise.
10595         (arm_store_release_exclusivedi): Likewise.
10596         (arm_store_release_exclusive<mode>): Likewise.
10598 2013-06-06  Richard Biener  <rguenther@suse.de>
10600         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
10601         after LTO_null.
10602         (lto_tag_is_tree_code_p): Adjust.
10603         (lto_tag_is_gimple_code_p): Likewise.
10604         (lto_gimple_code_to_tag): Likewise.
10605         (lto_tag_to_gimple_code): Likewise.
10606         (lto_tree_code_to_tag): Likewise.
10607         (lto_tag_to_tree_code): Likewise.
10608         * data-streamer.h (streamer_write_hwi_in_range): Use
10609         uhwi streaming to stream the normalized range.
10610         (streamer_read_hwi_in_range): Likewise.
10612 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10614         * config/arm/arm.md (enabled_for_depr_it): New attribute.
10615         (predicable_short_it): Likewise.
10616         (predicated): Likewise.
10617         (enabled): Handle above.
10618         (define_cond_exec): Set predicated attribute to yes.
10620 2013-06-05  Mike Stump  <mikestump@comcast.net>
10622         * gdbinit.in (__FUNCTION__): Add.
10624 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
10626         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
10627         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
10629 2013-06-05  Jan Hubicka  <jh@suse.cz>
10631         * varasm.c (mark_decl_referenced): Revert the removal until targets
10632         are fixed.
10634 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
10636         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
10637         instead of mark_decl_referenced.
10639 2013-06-05  Jan Hubicka  <jh@suse.cz>
10641         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
10642         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
10643         and symtab_used_from_object_file_p.
10644         (cgraph_make_node_local_1): Clear forced_by_abi.
10645         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
10646         * cgraph.h (symtab_node_base): Add forced_by_abi.
10647         (decide_is_variable_needed): Remove.
10648         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
10649         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
10650         (decide_is_symbol_needed): ... this one; handle symbols in general;
10651         always analyze virtuals; honnor forced_by_abi.
10652         (cgraph_finalize_function): Update.
10653         (varpool_finalize_decl): Update.
10654         (symbol_defined_and_needed): Remove.
10655         (analyze_functions): Update.
10656         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
10657         output_refs, input_overwrite_node): Handle forced_by_abi.
10658         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
10659         (address_taken_from_non_vtable_p): ... this one.
10660         (comdat_can_be_unshared_p_1): New function.
10661         (cgraph_comdat_can_be_unshared_p): Rename to ...
10662         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
10663         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
10664         (function_and_variable_visibility): Clear forced_by_abi as needed.
10665         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
10666         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
10667         * symtab.c (dump_symtab_base): Dump forced_by_abi.
10668         * varpool.c (decide_is_variable_needed): Remove.
10670 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10672         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
10673         (arm_option_override): Override arm_restrict_it where appropriate.
10674         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
10675         * config/arm/arm.opt (mrestrict-it): New command-line option.
10676         * doc/invoke.texi: Document -mrestrict-it.
10678 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
10680         * tsan.c (tsan_atomic_table): Make const.
10682 2013-06-05  Richard Biener  <rguenther@suse.de>
10684         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
10685         index associated with the tree we are supposed to replace.
10686         * tree-streamer-out.c (pack_ts_base_value_fields): Output
10687         TREE_ASM_WRITTEN as zero for everything but SSA names.
10689 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
10691         * tree-ssa-structalias.c (call_stmt_vars): Make static.
10693 2013-06-04  Jan Hubicka  <jh@suse.cz>
10695         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
10696         (input_node, input_varpool_node): Handle correctly external same
10697         body aliases.
10698         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
10699         nodes at ltrans stage.
10701 2013-06-04  Jan Hubicka  <jh@suse.cz>
10703         * ipa-inline.c (update_caller_keys): Fix availability test.
10704         (update_callee_keys): Likewise.
10705         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
10706         to follow ELF standard.
10708 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
10710         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
10711         (mips64r5900el-*-elf*): New configurations.
10712         * config/mips/mips-cpus.def (r5900): New processor.
10713         * config/mips/mips-tables.opt: Regenerate.
10714         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
10715         (mips_issue_rate): Handle PROCESSOR_R5900.
10716         (mips_reorg_process_insns): Force reorder mode for the R5900.
10717         * config/mips/mips.h (TARGET_MIPS5900): Define.
10718         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
10719         TARGET_MIPS5900.
10720         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
10721         TARGET_MIPS5900.
10722         * config/mips/mips.md (processor): Add r5900.
10723         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
10725 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
10727         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
10728         into function to generate MOVI instruction.
10729         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
10730         (aarch64_preferred_simd_mode): Turn into wrapper.
10731         (aarch64_output_scalar_simd_mov_immediate): New function.
10732         * config/aarch64/aarch64-protos.h: Add prototype for above.
10734 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
10736         * config/aarch64/aarch64.c (simd_immediate_info): Remove
10737         element_char member.
10738         (sizetochar): Return signed char.
10739         (aarch64_simd_valid_immediate): Remove elchar and other
10740         unnecessary variables.
10741         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
10742         Calculate element_char as required.
10743         * config/aarch64/aarch64-protos.h: Update and move prototype
10744         for aarch64_output_simd_mov_immediate.
10745         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
10746         Update arguments.
10748 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
10750         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
10751         information completed by aarch64_simd_valid_immediate.
10752         (aarch64_legitimate_constant_p): Update arguments.
10753         (aarch64_simd_valid_immediate): Work with struct rather than many
10754         pointers.
10755         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
10756         (aarch64_simd_make_constant): Update arguments.
10757         (aarch64_output_simd_mov_immediate): Work with struct rather than
10758         many pointers.  Output immediate directly rather than as operand.
10759         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
10760         Update prototype.
10761         * config/aarch64/constraints.md (Dn): Update arguments.
10763 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
10765         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
10766         longer static.
10767         (aarch64_simd_immediate_valid_for_move): Remove.
10768         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
10769         (aarch64_simd_make_constant): Update call.
10770         (aarch64_output_simd_mov_immediate): Update call.
10771         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
10772         Add prototype.
10773         * config/aarch64/constraints.md (Dn): Update call.
10775 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
10777         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
10778         return type to bool for prototype.
10779         (aarch64_legitimate_constant_p): Check for true instead of not -1.
10780         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
10781         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
10783 2013-06-04  Catherine Moore  <clm@codesourcery.com>
10785         * config/mips/mips.opt (meva): New.
10786         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
10787         (ASM_SPEC): Handle -meva.
10788         * doc/invoke.texi (meva):  Document.
10790 2013-06-04  Alan Modra  <amodra@gmail.com>
10792         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
10793         constant output.
10795 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10797         * rtl.def: Add extra fourth optional field to define_cond_exec.
10798         * gensupport.c (process_one_cond_exec): Process attributes from
10799         define_cond_exec.
10800         * doc/md.texi: Document fourth field in define_cond_exec.
10802 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
10804         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
10805         out the processing order as in store_bit_field_1.
10807 2013-06-04  Jan Hubicka  <jh@suse.cz>
10809         PR middle-end/57500
10810         * cgraphunit.c (cgraph_process_same_body_aliases): Create
10811         non-VAR_DECL node if it does not exist yet.
10813 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
10815         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
10816         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
10817         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
10818         target_cpu_default setting.
10820 2013-06-03  Teresa Johnson  <tejohnson@google.com>
10822         * dumpfile.c (opt_info_switch_p): Change -fopt-info
10823         default to -fopt-info=optimized instead of all.
10824         * doc/invoke.texi: Ditto.
10825         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
10826         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
10827         (execute_vect_slp): Emit BB vectorization success under
10828         MSG_OPTIMIZED_LOCATIONS.
10829         * tree-vect-slp.c (vect_slp_transform_bb): Change
10830         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
10831         * tree-vect-loop.c (vect_transform_loop): Ditto.
10833 2013-06-03  Jason Merrill  <jason@redhat.com>
10835         PR c++/57415
10836         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10837         Use TARGET_EXPR for C++.
10839 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
10841         PR rtl-optimization/57268
10842         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
10843         if DEBUG_INSN_P (insn).
10845         Reapply
10846         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
10848         PR rtl-optimization/57268
10849         * sched-deps.c (sched_analyze_2): Flush dependence lists if
10850         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
10852 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
10854         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
10855         (ix86_avoid_lea_for_addr): Likewise.
10856         (exact_dependency_1): Likewise.
10857         (ix86_adjust_cost): Likewise.
10858         (swap_top_of_ready_list): Fix formatting and !reload_completed check
10859         removed.
10860         (do_reorder_for_imul): Fix typo, formatting and
10861         !reload_completed check removed.
10862         (ix86_sched_reorder): Fix typo and formatting.
10863         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
10864         list.
10866 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
10868         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
10870 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
10872         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
10873         <STRING_CST>: Likewise.
10874         <VECTOR_CST>: Likewise.
10876 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
10877             Mikael Morin  <mikael@gcc.gnu.org>
10879         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
10880         * config.in: Regenerated.
10881         * configure: Regenerated.
10883 2013-06-01  Jan Hubicka  <jh@suse.cz>
10885         PR middle-end/57366
10886         * cgraphunit.c (compile): When weakref is not supported,
10887         set up transparent aliases before final output pass.
10888         * varasm.c (assemble_alias): Do not try to do it here.
10890 2013-06-01  Jan Hubicka  <jh@suse.cz>
10892         PR middle-end/57467
10893         * passes.c (for_per_function): Skip unanalyzed functions.
10895 2013-06-01  Jan Hubicka  <jh@suse.cz>
10897         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
10898         (lto_symtab_merge_symbols_1): ... this one.
10899         (lto_symtab_merge_cgraph_nodes): Rename to ...
10900         (lto_symtab_merge_symbols): ... this one; simplify.
10901         * cgraph.c (same_body_aliases_done): Rename to ...
10902         (cpp_implicit_aliases_done): ... this one.
10903         (cgraph_create_function_alias): Update.
10904         (cgraph_same_body_alias): Update.
10905         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
10906         (verify_edge_corresponds_to_fndecl): Simplify.
10907         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
10908         (cgraph_node): Remove same_body_alias.
10909         (varpool_node): Remove alias_of and extra_name_alias.
10910         (same_body_aliases_done): Rename to ..
10911         (cpp_implicit_aliases_done): ... this one.
10912         (symtab_alias_ultimate_target): Add default parameter.
10913         (symtab_resolve_alias): New function.
10914         (fixup_same_cpp_alias_visibility): Declare.
10915         (cgraph_function_node): Add default parameter.
10916         (cgraph_node_asm_name): Likewise.
10917         (cgraph_function_or_thunk_node): Add default parameter; do
10918         not ICE when it is NULL.
10919         (varpool_variable_node): Likewise.
10920         * tree-emutls.c (create_emultls_var): Update.
10921         (ipa_lower_emutls): Update.
10922         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
10923         (cgraph_reset_node): Reset alias info.
10924         (cgraph_finalize_function): Update.
10925         (fixup_same_cpp_alias_visibility): Move to symtab.c.
10926         (analyze_function): Simplify.
10927         (cgraph_process_same_body_aliases): Simplify.
10928         (analyze_functions): Fixup same body aliases.
10929         (handle_alias_pairs): Simplify.
10930         (assemble_thunk): Update.
10931         (assemble_thunks_and_aliases): Update.
10932         (output_weakrefs): Rewrite.
10933         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
10934         (lto_output_varpool_node): Likewise.
10935         (compute_ltrans_boundary): Remve assert.
10936         (get_alias_symbol): New functoin.
10937         (input_node): Rewrite alias handling.
10938         (input_varpool_node): Likewise.
10939         * ipa-pure-const.c (propagate_pure_const): Fix formating.
10940         * ipa.c (process_references): Handle weakrefs correctly.
10941         (symtab_remove_unreachable_nodes): Likewise.
10942         * trans-mem.c (get_cg_data): Update.
10943         (ipa_tm_create_version_alias): Update.
10944         (ipa_tm_execute): Update.
10945         * symtab.c (dump_symtab_base): Dump aliases.
10946         (verify_symtab_base): Verify aliases.
10947         (symtab_node_availability): New function.
10948         (symtab_alias_ultimate_target): Simplify.
10949         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
10950         handle all the fixup cases.
10951         (symtab_resolve_alias): New function.
10952         * passes.c (ipa_write_summaries): Handle weakrefs.
10953         * varpool.c (varpool_analyze_node): Simplify.
10954         (assemble_aliases): Update.
10955         (varpool_create_variable_alias): Simplify.
10956         (varpool_extra_name_alias): Simplify.
10957         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
10958         (lto_symtab_merge_symbols): ... this one.
10960 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
10962         Revert
10963         PR rtl-optimization/57268
10964         * sched-deps.c (sched_analyze_2): Flush dependence lists if
10965         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
10967 2013-06-01  Tobias Burnus  <burnus@net-b.de>
10969         Partially reverted:
10970         2013-05-31  Tobias Burnus  <burnus@net-b.de>
10972         PR middle-end/57073
10973         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
10974         further up.
10976 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
10978         PR rtl-optimization/57268
10979         * sched-deps.c (sched_analyze_2): Flush dependence lists if
10980         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
10982 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
10984         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
10985         unordered comparison operators when -fno-trapping-math is in effect
10986         on the e500.
10987         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
10988         and implement unordered comparison operators properly on the e500.
10990 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
10992         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
10993         for constant scalar integers.
10994         (simplify_relational_operation_1): Likewise.
10996 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
10998         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
10999         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
11000         Fix comment.
11002 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
11003             Igor Zamyatin  <igor.zamyatin@intel.com>
11005         Silvermont (SLM) architecture performance tuning.
11006         * config/i386/i386.h (enum ix86_tune_indices): Add
11007         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
11008         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
11010         * config/i386/i386.c (initial_ix86_tune_features)
11011         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
11012         (ix86_lea_outperforms): Handle Silvermont tuning.
11013         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
11014         call.
11015         (ix86_use_lea_for_mov): Likewise.
11016         (ix86_avoid_lea_for_addr): Likewise.
11017         (ix86_lea_for_add_ok): Likewise.
11018         (exact_dependency_1): New function.
11019         (exact_store_load_dependency): Likewise.
11020         (ix86_adjust_cost): Handle Silvermont tuning.
11021         (do_reoder_for_imul): Likewise.
11022         (swap_top_of_ready_list): New function.
11023         (ix86_sched_reorder): Changed to handle Silvermont tuning.
11025         * config/i386/i386.md (peepholes that split memory operand in fp
11026         converts): New.
11028 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
11030         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11031         Remove un-necessary braces.
11033 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
11035         * config/aarch64/aarch64.c (aarch64_classify_symbol):
11036         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
11038 2013-05-31  Tobias Burnus  <burnus@net-b.de>
11040         PR middle-end/57073
11041         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
11043 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11045         PR target/56315
11046         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
11047         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
11048         * config/arm/neon.md (iordi3_neon): Remove.
11049         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
11050         * config/arm/predicates.md (imm_for_neon_logic_operand):
11051         Move to earlier in the file.
11052         (neon_logic_op2): Likewise.
11053         (arm_iordi_operand_neon): New predicate.
11055 2013-05-31  Richard Biener  <rguenther@suse.de>
11057         PR tree-optimization/57478
11058         PR tree-optimization/57453
11059         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
11060         are life as well.
11062 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
11064         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
11065         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
11067 2013-05-30  Tobias Burnus  <burnus@net-b.de>
11068             Thomas Koenig  <tkoenig@gcc.gnu.org>
11070         PR middle-end/57073
11071         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
11072         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
11074 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
11076         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
11078 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
11080         * target.def (register_usage_leveling_p): New hook.
11081         * targhooks.c (default_register_usage_leveling_p): New.
11082         * targhooks.h (default_register_usage_leveling_p): New prototype.
11083         * lra-assigns.c (register_usage_leveling_p): Use the hook.
11084         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
11085         * doc/tm.texi: Update.
11086         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
11088 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
11090         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
11091         (*insv_reg<mode>): New define_insn.
11093 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
11095         PR rtl-optimization/57439
11096         * postreload.c (move2add_valid_value_p): Check that we have
11097         a zero subreg_regno_offset when accessing the register in
11098         the requested mode.
11100 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
11101             Igor Zamyatin  <igor.zamyatin@intel.com>
11103         Silvermont (SLM) architecture pipeline model, tuning and
11104         insn selection.
11105         * config.gcc: Add slm config options and target.
11107         * config/i386/slm.md: New.
11109         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
11111         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
11112         PROCESSOR_SLM.
11113         (ix86_target_macros_internal): Likewise.
11115         * config/i386/i386.c (slm_cost): New cost.
11116         (m_SLM): New macro flag.
11117         (initial_ix86_tune_features): Set m_SLM.
11118         (x86_accumulate_outgoing_args): Likewise.
11119         (x86_arch_always_fancy_math_387): Likewise.
11120         (processor_target_table): Add slm cost.
11121         (cpu_names): Add slm cpu name.
11122         (x86_option_override_internal): Set SLM ISA.
11123         (ix86_issue_rate): New case PROCESSOR_SLM.
11124         (ia32_multipass_dfa_lookahead): Likewise.
11125         (fold_builtin_cpu): Add slm.
11127         * config/i386/i386.h (TARGET_SLM): New target macro.
11128         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
11129         (processor_type): Add PROCESSOR_SLM.
11131         * config/i386/i386.md (cpu): Add new value "slm".
11132         (slm.md): Include slm.md.
11134 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
11135             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
11137         * config/arm/arm-protos.h: Add and update function protos.
11138         * config/arm/arm.c (use_simple_return_p): New added.
11139         (thumb2_expand_return): Check simple_return flag.
11140         * config/arm/arm.md: Add simple_return and conditional simple_return.
11141         * config/arm/iterators.md: Add iterator for return and simple_return.
11143 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
11145         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
11146         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
11147         (arm_emit_vfp_multi_reg_pop): Likewise.
11148         (thumb2_emit_ldrd_pop): Likewise.
11149         (arm_expand_epilogue): Add misc REG_CFA notes.
11150         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
11152 2013-05-29  Lawrence Crowl  <crowl@google.com>
11154         * config/arm/t-arm: Update for below.
11156         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
11157         Change type to hash_table.  Update dependent calls and types.
11159         * config/i386/t-cygming: Update for below.
11161         * config/i386/t-interix: Update for below.
11163         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
11164         Change type to hash_table.  Update dependent calls and types.
11165         (i386_find_on_wrapper_list::wrappers): Likewise.
11167         * config/ia64/t-ia64: Update for below.
11169         * config/ia64/ia64.c (bundle_state_table):
11170         Change type to hash_table.  Update dependent calls and types.
11172         * config/mips/mips.c (mips_reorg_process_insns::htab):
11173         Change type to hash_table.  Update dependent calls and types.
11175         * config/sol2.c (solaris_comdat_htab):
11176         Change type to hash_table.  Update dependent calls and types.
11178         * config/t-sol2: Update for above.
11180 2013-05-29  Teresa Johnson  <tejohnson@google.com>
11182         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
11183         functions are not yet marked as defined.
11185 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
11186             Pat Haugen  <pthaugen@us.ibm.com>
11187             Peter Bergner  <bergner@vnet.ibm.com>
11189         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
11190         instructions.
11191         (VEC_A): Likewise.
11192         (VEC_C): Likewise.
11193         (vrotl<mode>3): Likewise.
11194         (vashl<mode>3): Likewise.
11195         (vlshr<mode>3): Likewise.
11196         (vashr<mode>3): Likewise.
11198         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11199         support for power8 V2DI builtins.
11201         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
11202         power8 V2DI builtins.
11203         (vupkhsw): Likewise.
11204         (vupklsw): Likewise.
11205         (vaddudm): Likewise.
11206         (vminsd): Likewise.
11207         (vmaxsd): Likewise.
11208         (vminud): Likewise.
11209         (vmaxud): Likewise.
11210         (vpkudum): Likewise.
11211         (vpksdss): Likewise.
11212         (vpkudus): Likewise.
11213         (vpksdus): Likewise.
11214         (vrld): Likewise.
11215         (vsld): Likewise.
11216         (vsrd): Likewise.
11217         (vsrad): Likewise.
11218         (vsubudm): Likewise.
11219         (vcmpequd): Likewise.
11220         (vcmpgtsd): Likewise.
11221         (vcmpgtud): Likewise.
11222         (vcmpequd_p): Likewise.
11223         (vcmpgtsd_p): Likewise.
11224         (vcmpgtud_p): Likewise.
11225         (vupkhsw): Likewise.
11226         (vupklsw): Likewise.
11227         (vaddudm): Likewise.
11228         (vmaxsd): Likewise.
11229         (vmaxud): Likewise.
11230         (vminsd): Likewise.
11231         (vminud): Likewise.
11232         (vpksdss): Likewise.
11233         (vpksdus): Likewise.
11234         (vpkudum): Likewise.
11235         (vpkudus): Likewise.
11236         (vrld): Likewise.
11237         (vsld): Likewise.
11238         (vsrad): Likewise.
11239         (vsrd): Likewise.
11240         (vsubudm): Likewise.
11242         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
11243         support for power8 V2DI instructions.
11245         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
11246         power8 V2DI instructions.  Combine pack and unpack insns to use an
11247         iterator for each mode.  Check whether a particular mode supports
11248         Altivec instructions instead of just checking TARGET_ALTIVEC.
11249         (UNSPEC_VPKUWUM): Likewise.
11250         (UNSPEC_VPKSHSS): Likewise.
11251         (UNSPEC_VPKSWSS): Likewise.
11252         (UNSPEC_VPKUHUS): Likewise.
11253         (UNSPEC_VPKSHUS): Likewise.
11254         (UNSPEC_VPKUWUS): Likewise.
11255         (UNSPEC_VPKSWUS): Likewise.
11256         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
11257         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
11258         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
11259         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
11260         (UNSPEC_VUPKHSB): Likewise.
11261         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
11262         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
11263         (UNSPEC_VUPKHSH): Likewise.
11264         (UNSPEC_VUPKLSB): Likewise.
11265         (UNSPEC_VUPKLSH): Likewise.
11266         (VI2): Likewise.
11267         (VI_char): Likewise.
11268         (VI_scalar): Likewise.
11269         (VI_unit): Likewise.
11270         (VP): Likewise.
11271         (VP_small): Likewise.
11272         (VP_small_lc): Likewise.
11273         (VU_char): Likewise.
11274         (add<mode>3): Likewise.
11275         (altivec_vaddcuw): Likewise.
11276         (altivec_vaddu<VI_char>s): Likewise.
11277         (altivec_vadds<VI_char>s): Likewise.
11278         (sub<mode>3): Likewise.
11279         (altivec_vsubcuw): Likewise.
11280         (altivec_vsubu<VI_char>s): Likewise.
11281         (altivec_vsubs<VI_char>s): Likewise.
11282         (altivec_vavgs<VI_char>): Likewise.
11283         (altivec_vcmpbfp): Likewise.
11284         (altivec_eq<mode>): Likewise.
11285         (altivec_gt<mode>): Likewise.
11286         (altivec_gtu<mode>): Likewise.
11287         (umax<mode>3): Likewise.
11288         (smax<mode>3): Likewise.
11289         (umin<mode>3): Likewise.
11290         (smin<mode>3): Likewise.
11291         (altivec_vpkuhum): Likewise.
11292         (altivec_vpkuwum): Likewise.
11293         (altivec_vpkshss): Likewise.
11294         (altivec_vpkswss): Likewise.
11295         (altivec_vpkuhus): Likewise.
11296         (altivec_vpkshus): Likewise.
11297         (altivec_vpkuwus): Likewise.
11298         (altivec_vpkswus): Likewise.
11299         (altivec_vpks<VI_char>ss): Likewise.
11300         (altivec_vpks<VI_char>us): Likewise.
11301         (altivec_vpku<VI_char>us): Likewise.
11302         (altivec_vpku<VI_char>um): Likewise.
11303         (altivec_vrl<VI_char>): Likewise.
11304         (altivec_vsl<VI_char>): Likewise.
11305         (altivec_vsr<VI_char>): Likewise.
11306         (altivec_vsra<VI_char>): Likewise.
11307         (altivec_vsldoi_<mode>): Likewise.
11308         (altivec_vupkhsb): Likewise.
11309         (altivec_vupkhs<VU_char>): Likewise.
11310         (altivec_vupkls<VU_char>): Likewise.
11311         (altivec_vupkhsh): Likewise.
11312         (altivec_vupklsb): Likewise.
11313         (altivec_vupklsh): Likewise.
11314         (altivec_vcmpequ<VI_char>_p): Likewise.
11315         (altivec_vcmpgts<VI_char>_p): Likewise.
11316         (altivec_vcmpgtu<VI_char>_p): Likewise.
11317         (abs<mode>2): Likewise.
11318         (vec_unpacks_hi_v16qi): Likewise.
11319         (vec_unpacks_hi_v8hi): Likewise.
11320         (vec_unpacks_lo_v16qi): Likewise.
11321         (vec_unpacks_hi_<VP_small_lc>): Likewise.
11322         (vec_unpacks_lo_v8hi): Likewise.
11323         (vec_unpacks_lo_<VP_small_lc>): Likewise.
11324         (vec_pack_trunc_v8h): Likewise.
11325         (vec_pack_trunc_v4si): Likewise.
11326         (vec_pack_trunc_<mode>): Likewise.
11328         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
11329         V2DI builtins.
11330         (vec_vmaxsd): Likewise.
11331         (vec_vmaxud): Likewise.
11332         (vec_vminsd): Likewise.
11333         (vec_vminud): Likewise.
11334         (vec_vpksdss): Likewise.
11335         (vec_vpksdus): Likewise.
11336         (vec_vpkudum): Likewise.
11337         (vec_vpkudus): Likewise.
11338         (vec_vrld): Likewise.
11339         (vec_vsld): Likewise.
11340         (vec_vsrad): Likewise.
11341         (vec_vsrd): Likewise.
11342         (vec_vsubudm): Likewise.
11343         (vec_vupkhsw): Likewise.
11344         (vec_vupklsw): Likewise.
11346 2013-05-29  Jan Hubicka  <jh@suse.cz>
11348         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
11349         flags; reorder rest of fields in more consistent way.
11350         (varpool_node): Remove analyzed, finalized and alias.
11351         (cgraph_ndoe): Likewise.
11352         (symtab_alias_ultimate_target): New function.
11353         (cgraph_function_node): Move offline.
11354         (cgraph_reset_node): Declare.
11355         (cgraph_comdat_can_be_unshared_p): Remove.
11356         (varpool_remove_initializer): Declare.
11357         (varpool_first_defined_variable, varpool_next_defined_variable
11358         cgraph_first_defined_function, cgraph_next_defined_function): Update.
11359         (cgraph_function_with_gimple_body_p): Update.
11360         (varpool_all_refs_explicit_p): Update.
11361         (symtab_alias_target): New function.
11362         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
11363         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
11364         (cgraph_function_or_thunk_node): Simplify using
11365         symtab_alias_ultimate_target.
11366         (varpool_variable_node): Likewise.
11367         * cgraph.c (cgraph_create_function_alias): Update.
11368         (cgraph_add_thunk): Update.
11369         (cgraph_remove_node): Update.
11370         (dump_cgraph_node): Do not dump removed flags.
11371         (cgraph_function_body_availability): Update.
11372         (cgraph_propagate_frequency): Update.
11373         (verify_cgraph_node): Check sanity of local flag.
11374         (cgraph_function_node): Move here from cgraph.h; revamp for
11375         cgraph_function_or_thunk_node.
11376         * lto-symtab.c (lto_varpool_replace_node): Update.
11377         (lto_symtab_resolve_can_prevail_p): Update.
11378         (lto_symtab_merge_cgraph_nodes): Update.
11379         * ipa-cp.c (determine_versionability, initialize_node_lattices,
11380         propagate_constants_accross_call, devirtualization_time_bonus,
11381         ipcp_propagate_stage): Update.
11382         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
11383         * ipa-inline-transform.c (clone_inlined_nodes,
11384         preserve_function_body_p): Update.
11385         * ipa-reference.c (propagate): Update.
11386         (write_node_summary_p): Update.
11387         * toplev.c (wrapup_global_declaration_2): Update.
11388         * cgraphunit.c (cgraph_analyze_function): Rename to ...
11389         (analyze_function) ... this one.
11390         (cgraph_process_new_functions): Update.
11391         (cgraph_reset_node): Export.
11392         (cgraph_finalize_function): Update.
11393         (cgraph_add_new_function): Update.
11394         (process_function_and_variable_attributes): Update.
11395         (varpool_finalize_decl): Update.
11396         (symbol_finalized): Remove.
11397         (symbol_finalized_and_needed): Rename to ...
11398         (symbol_defined_and_needed): ... update.
11399         (cgraph_analyze_functions): Update.
11400         (handle_alias_pairs): Update.
11401         (mark_functions_to_output): Update.
11402         (assemble_thunk): Update.
11403         (output_in_order): Update.
11404         (output_weakrefs): Update.
11405         (finalize_compilation_unit): Update.
11406         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
11407         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
11408         input_node, input_varpool_node): Update.
11409         * dbxout.c (dbxout_expand_expr): Update.
11410         * cgraphclones.c (cgraph_clone_node): Update.
11411         (cgraph_copy_node_for_versioning): Update.
11412         (cgraph_materialize_clone): Update.
11413         (cgraph_materialize_all_clones): Update.
11414         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
11415         propagate_pure_const, propagate_nothrow): Update.
11416         * lto-streamer-out.c (lto_output, write_symbol): Update.
11417         * ipa-utils.c (ipa_reverse_postorder): Update.
11418         * ipa-inline.c (can_inline_edge_p): Update.
11419         (update_caller_keys, ipa_inline): Update.
11420         * dwarf2out.c (reference_to_unused,
11421         premark_types_used_by_global_vars_helper): Update.
11422         * tree-eh.c (tree_could_trap_p): Update.
11423         * ipa-split.c (consider_split, execute_split_functions): Update.
11424         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
11425         has_addr_references_p): Update; move ahead in file for better
11426         readability.
11427         (process_references): Simplify.
11428         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
11429         bodies are removed.
11430         (cgraph_comdat_can_be_unshared_p): Make static.
11431         (cgraph_externally_visible_p): Update.
11432         (varpool_externally_visible_p): Update.
11433         (function_and_variable_visibility): Update.
11434         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
11435         ipa_tm_mark_force_output_node): Update.
11436         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
11437         estimate_edge_devirt_benefit, inline_generate_summary,
11438         inline_write_summary): Update.
11439         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
11440         * ipa-prop.c (ipa_compute_jump_functions): Update.
11441         (ipa_print_node_params, ipa_prop_read_section,
11442         ipa_update_after_lto_read, read_replacements_section): Update.
11443         * varasm.c (mark_decl_referenced): Update.
11444         (assemble_alias, dump_tm_clone_pairs): Update.
11445         * tree-inline.c (copy_bb): Update.
11446         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
11447         Update.
11448         * symtab.c (dump_symtab_base): Print new flags.
11449         (verify_symtab_base): Verify new flags.
11450         (symtab_alias_ultimate_target): New function.
11451         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
11452         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
11453         Update.
11454         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
11455         Update.
11456         * i386.c (ix86_get_function_versions_dispatcher,
11457         ix86_generate_version_dispatcher_body): Update.
11458         (fold_builtin_cpu): Use varpool_add_new_variable.
11459         * varpool.c (varpool_remove_initializer): Break out from ...
11460         (varpool_remove_node): ... this one.
11461         (dump_varpool_node, varpool_node_for_asm,
11462         cgraph_variable_initializer_availability, varpool_analyze_node,
11463         varpool_assemble_decl, varpool_remove_unreferenced_decls,
11464         varpool_finalize_named_section_flags, varpool_create_variable_alias):
11465         Update.
11467 2013-05-29  Jan Hubicka  <jh@suse.cz>
11469         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
11471 2013-05-29  Easwaran Raman  <eraman@google.com>
11473         PR tree-optimization/57442
11474         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
11475         when control exits the main loop.
11477 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
11479         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
11480         and RX600.
11481         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
11482         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
11483         * rx/t-rx: Add rx100 under multi library matches option for nofpu
11484         option.
11486 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11488         PR tree-optimization/57441
11489         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
11490         Don't limit size of incr_vec to number of candidates.
11492 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
11494         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
11495         and mips16 directories.
11496         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
11497         (MULTILIB_DIRNAMES): Ditto.
11498         (MULTILIB_EXCEPTIONS): Add new exceptions.
11499         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
11500         (MULTILIB_DIRNAMES): Ditto.
11501         (MULTILIB_EXCEPTIONS): Add new exceptions.
11503 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
11504             Marcus Shawcroft  <marcus.shawcroft@arm.com>
11506         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
11507         SYMBOL_TINY_ABSOLUTE.
11508         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
11509         SYMBOL_TINY_ABSOLUTE.
11510         (aarch64_expand_mov_immediate): Likewise.
11511         (aarch64_classify_symbol): Likewise.
11512         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
11513         Permit SYMBOL_TINY_ABSOLUTE.
11514         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
11516 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
11517             Marcus Shawcroft  <marcus.shawcroft@arm.com>
11519         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
11520         Refactor if/switch.  Replace gcc_assert with if.
11522 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
11524         * config/i386/i386.c (initial_ix86_tune_features): Enable
11525         FP Reassociation for AMD bdver1 and bdver2.
11527 2013-05-29  Martin Jambor  <mjambor@suse.cz>
11529         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
11530         and IMAGPART_EXPR do not occur within other handled_components.
11532 2013-05-29  Richard Biener  <rguenther@suse.de>
11534         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
11535         access on whether the use is in the BB we currently try to
11536         vectorize.
11537         (vect_bb_vectorization_profitable_p): Pass the BB we currently
11538         vectorize to vect_bb_slp_scalar_cost.
11540 2013-05-29  Richard Biener  <rguenther@suse.de>
11542         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
11543         computing scalar cost offsetted by stmts that are kept live
11544         by scalar uses.
11545         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
11546         for computation of scalar cost.
11548 2013-05-28  Steve Ellcey  <sellcey@mips.com>
11550         * config/mips/mips-cpus.def (mips32r2): Change processor type.
11552 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
11554         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
11555         array notation built-in reduction functions.
11556         * doc/passes.texi (Passes): Added documentation about changes done
11557         for Cilk Plus.
11558         * doc/invoke.texi (C Dialect Options): Added documentation about
11559         the -fcilkplus flag.
11560         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
11561         (BUILTINS_DEF): Depend on cilkplus.def.
11562         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
11563         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
11564         * cilkplus.def: New file.
11566 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
11568         PR rtl-optimization/57439
11569         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
11571 2013-05-28  Easwaran Raman  <eraman@google.com>
11573         PR tree-optimization/57337
11574         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
11575         (find_insert_point): Correctly identify the insertion point
11576         when two statements with the same UID is compared.
11578 2013-05-28  Richard Biener  <rguenther@suse.de>
11580         PR tree-optimization/56787
11581         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
11582         from the list of data references.
11583         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
11584         clobbers.
11585         (vect_analyze_loop_operations): Likewise.
11586         (vect_transform_loop): Remove clobbers.
11588 2013-05-28  Martin Jambor  <mjambor@suse.cz>
11590         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
11591         and REALPART_EXPRs have scalar type.
11593 2013-05-28  Richard Biener  <rguenther@suse.de>
11595         PR tree-optimization/57411
11596         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
11597         virtual operands.
11598         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
11599         virtual operand propagation.
11601 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
11603         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
11604         destination register for bmasksi_vis.
11605         (vector_init_bshuffle): Likewise.
11606         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
11608 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
11610         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
11611         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
11612         mode if the instruction isn't available in the original mode.
11613         * config/sparc/sparc.opt (mfix-ut699): New option.
11614         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
11615         (divdf3): Turn into expander.
11616         (divdf3_nofix): New insn.
11617         (divdf3_fix): Likewise.
11618         (divsf3): Disable if -mfix-ut699.
11619         (sqrtdf2): Turn into expander.
11620         (sqrtdf2_nofix): New insn.
11621         (sqrtdf2_fix): Likewise.
11622         (sqrtsf2): Disable if -mfix-ut699.
11624 2013-05-27  Richard Biener  <rguenther@suse.de>
11626         PR middle-end/57412
11627         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
11628         block for the new loop.
11630 2013-05-27  Richard Biener  <rguenther@suse.de>
11632         PR tree-optimization/57343
11633         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
11634         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
11635         (number_of_iterations_cond): Do not build the folded tree.
11637 2013-05-27  Richard Biener  <rguenther@suse.de>
11639         Revert
11640         PR middle-end/57381
11641         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
11642         OEP_CONSTANT_ADDRESS_OF retained.
11644         PR tree-optimization/57417
11645         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
11646         for unchanged base.
11647         (set_ssa_val_to): Compare addresses using
11648         get_addr_base_and_unit_offset.
11650 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
11652         PR rtl-optimization/56833
11653         * postreload.c (move2add_record_mode): New function.
11654         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
11655         (move2add_use_add2_insn): Use move2add_record_sym_value.
11656         (move2add_use_add3_insn): Likewise.
11657         (reload_cse_move2add): Use move2add_valid_value_p and
11658         move2add_record_mode.  Invalidate call-clobbered and REG_INC
11659         affected regs by setting reg_mode to VOIDmode.
11660         (move2add_note_store): Don't pretend the inside of a SUBREG is
11661         the actual destination.  Invalidate single/leading registers by
11662         setting reg_mode to VOIDmode.
11663         Use move2add_record_sym_value, move2add_valid_value_p and
11664         move2add_record_mode.
11666 2013-05-27  Richard Biener  <rguenther@suse.de>
11668         PR tree-optimization/57396
11669         * tree-affine.c (double_int_constant_multiple_p): Properly
11670         return false for val == 0 and div != 0.
11672 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
11674         * config/mips/mips.h: Use #elif in preprocessor conditions.
11676 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
11678         PR target/53916
11679         * config/mips/constraints.md (kl): New constraint.
11680         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
11681         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
11682         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
11683         from LO for MIPS16.
11684         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
11686 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
11688         PR target/55777
11689         * config/mips/mips.c (mips_can_inline_p): New function.
11690         (TARGET_CAN_INLINE_P): Define.
11692 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
11694         * sched-int.h (ds_t, dw_t): Make unsigned int.
11695         Fix documentation that describes how all the ds_t bits are used.
11696         Reserve the last bit for delayed-branch scheduling.
11697         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
11698         (BITS_PER_DEP_WEAK): Fix definition and documentation.
11699         (gen_dep_weak_1): Remove prototype.
11700         * sched-deps.c (get_dep_weak_1): Make static.
11701         * target.def (speculate_insn, needs_block_p, gen_spec_check,
11702         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
11703         * doc/tm.texi: Regenerate.
11704         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
11706 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
11708         PR debug/56950
11709         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
11711 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
11712             Sandra Loosemore  <sandra@codesourcery.com>
11714         * config.gcc (powerpc-*): Allow native for with-cpu.
11716 2013-05-24  Jeff Law  <law@redhat.com>
11718         PR tree-optimization/57124
11719         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
11720         conversion feeding a condition if the range has an overflow
11721         if -fstrict-overflow.  Add warnings for when we do make the
11722         transformation.
11724 2013-05-24  Dehao Chen  <dehao@google.com>
11726         * tree-cfg.c (locus_discrim_map): Fix the typo.
11727         (locus_discrim_hasher): Likewise.
11728         (locus_discrim_hasher::hash): Likewise.
11729         (locus_discrim_hasher::equal): Likewise.
11731 2013-05-24  Martin Jambor  <mjambor@suse.cz>
11733         PR tree-optimization/57294
11734         * cgraph.h (ipa_record_stmt_references): Declare.
11735         * cgraphbuild.c (ipa_record_stmt_references): New function.
11736         (build_cgraph_edges): Use ipa_record_stmt_references.
11737         (rebuild_cgraph_edges): Likewise.
11738         (cgraph_rebuild_references): Likewise.
11739         * ipa-prop.c (ipa_modify_call_arguments): Discard references
11740         associated with the old statement and build references from the
11741         newly built statements.
11742         * ipa-ref.c (ipa_remove_stmt_references): New function.
11743         * ipa-ref.h (ipa_remove_stmt_references): Declare.
11745 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
11747         * lra-constraints.c (emit_spill_move): Use smaller mode for
11748         mem-mem moves.
11749         (check_and_process_move): Consider mem-reg moves for secondary
11750         too.
11751         (curr_insn_transform): Don't lose insns emitted before for
11752         secondary memory moves.
11753         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
11754         reg set up in the current insn.
11756 2013-05-24  Dehao Chen  <dehao@google.com>
11758         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
11759         hash function.
11760         (locus_descrim_hasher::equal): Likewise.
11761         (build_gimple_cfg): New discriminator assignment algorithm.
11762         (make_edges): Likewise.
11763         (next_discriminator_for_locus): Likewise.
11764         (same_line_p): Likewise.
11765         (assign_discriminators): Likewise.
11766         (make_cond_expr_edges): Likewise.
11767         (make_gimple_switch_edges): Likewise.
11768         (make_goto_expr_edges): Likewise.
11769         (make_gimple_asm_edges): Likewise.
11771 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
11773         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
11774         X format specifier to only display bottom 16 bits.
11775         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
11776         immediate to match for operand 2, since it will be masked.
11778 2013-05-24  Richard Biener  <rguenther@suse.de>
11780         PR tree-optimization/57287
11781         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
11782         all SSA names that occur in abnormal PHIs.
11784 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
11786         PR tree-ssa/57385
11787         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
11788         that index is not negative.
11790 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
11792         PR rtl-optimization/55177
11793         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
11794         (simplify_byte_swapping_operation): New.
11795         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
11796         (simplify_relational_operation_1): Deal with BSWAP.
11798 2013-05-23  Richard Henderson  <rth@redhat.com>
11800         PR target/56742
11801         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
11802         (ix86_reorg): Call it.
11804 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
11806         PR target/57379
11807         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
11808         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
11809         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
11811 2013-05-23  Christian Bruel  <christian.bruel@st.com>
11813         PR debug/57351
11814         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
11816 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
11817             Marcus Shawcroft  <marcus.shawcroft@arm.com>
11819         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
11820         * config/aarch64/constraints.md (Usa): Remove.
11821         * doc/md.texi (AArch64 Usa): Remove.
11823 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
11824             Marcus Shawcroft  <marcus.shawcroft@arm.com>
11826         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
11827         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
11828         * config/aarch64/predicates.md (aarch64_const_address): Remove.
11829         (aarch64_mov_operand): Use aarch64_mov_operand_p.
11831 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
11833         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
11834         instruction (AdvSIMD).
11835         * config/aarch64/aarch64-builtins.c
11836         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
11837         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
11839 2013-05-23  Martin Jambor  <mjambor@suse.cz>
11841         PR middle-end/57347
11842         * tree.h (contains_bitfld_component_ref_p): Declare.
11843         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
11844         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
11845         caller.
11846         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
11847         not access a bit-field.  Assert all final offsets are byte-aligned.
11849 2013-05-23  Richard Biener  <rguenther@suse.de>
11851         PR tree-optimization/57380
11852         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
11853         least one invariant or re-used load.
11854         * passes.c (init_optimization_passes): Move pass_phiprop before
11855         pass_forwprop.
11857 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
11859         * config/aarch64/aarch64-simd.md
11860         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
11862 2013-05-23  Richard Biener  <rguenther@suse.de>
11864         PR middle-end/57381
11865         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
11866         OEP_CONSTANT_ADDRESS_OF retained.
11868 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
11870         PR middle-end/57344
11871         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
11872         don't lower unit.  Handle unit not being always BITS_PER_WORD.
11874 2013-05-23  Richard Biener  <rguenther@suse.de>
11876         PR rtl-optimization/57341
11877         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
11878         instead of true_dependence.
11880 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
11882         * bb-reorder.c (branch_threshold): Make const.
11883         (exec_threshold): Ditto.
11885 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
11886             Pat Haugen  <pthaugen@us.ibm.com>
11887             Peter Bergner  <bergner@vnet.ibm.com>
11889         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
11890         documentation for the power8 crypto builtins.
11892         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
11894         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
11895         macros for defining power8 builtin functions.
11896         (BU_P8V_AV_2): Likewise.
11897         (BU_P8V_AV_P): Likewise.
11898         (BU_P8V_VSX_1): Likewise.
11899         (BU_P8V_OVERLOAD_1): Likewise.
11900         (BU_P8V_OVERLOAD_2): Likewise.
11901         (BU_CRYPTO_1): Likewise.
11902         (BU_CRYPTO_2): Likewise.
11903         (BU_CRYPTO_3): Likewise.
11904         (BU_CRYPTO_OVERLOAD_1): Likewise.
11905         (BU_CRYPTO_OVERLOAD_2): Likewise.
11906         (XSCVSPDP): Fix typo, point to the correct instruction.
11907         (VCIPHER): Add power8 crypto builtins.
11908         (VCIPHERLAST): Likewise.
11909         (VNCIPHER): Likewise.
11910         (VNCIPHERLAST): Likewise.
11911         (VPMSUMB): Likewise.
11912         (VPMSUMH): Likewise.
11913         (VPMSUMW): Likewise.
11914         (VPERMXOR_V2DI): Likewise.
11915         (VPERMXOR_V4SI: Likewise.
11916         (VPERMXOR_V8HI: Likewise.
11917         (VPERMXOR_V16QI: Likewise.
11918         (VSHASIGMAW): Likewise.
11919         (VSHASIGMAD): Likewise.
11920         (VPMSUM): Likewise.
11921         (VPERMXOR): Likewise.
11922         (VSHASIGMA): Likewise.
11924         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
11925         __CRYPTO__ if the crypto instructions are available.
11926         (altivec_overloaded_builtins): Add support for overloaded power8
11927         builtins.
11929         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
11930         support for power8 crypto builtins.
11931         (builtin_function_type): Likewise.
11932         (altivec_init_builtins): Add support for builtins that take vector
11933         long long (V2DI) arguments.
11935         * config/rs6000/crypto.md: New file, define power8 crypto
11936         instructions.
11938 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
11939             Pat Haugen  <pthaugen@us.ibm.com>
11940             Peter Bergner  <bergner@vnet.ibm.com>
11942         * doc/invoke.texi (Option Summary): Add power8 options.
11943         (RS/6000 and PowerPC Options): Likewise.
11945         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
11946         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
11947         wm, wn, wr documentation.
11949         * config/rs6000/constraints.md (wm): New constraint for VSX
11950         registers if direct move instructions are enabled.
11951         (wn): New constraint for no registers.
11952         (wq): New constraint for quad word even GPR registers.
11953         (wr): New constraint if 64-bit instructions are enabled.
11954         (wv): New constraint if power8 vector instructions are enabled.
11955         (wQ): New constraint for quad word memory locations.
11957         * config/rs6000/predicates.md (const_0_to_15_operand): New
11958         constraint for 0..15 for crypto instructions.
11959         (gpc_reg_operand): If VSX allow registers in VSX registers as well
11960         as GPR and floating point registers.
11961         (int_reg_operand): New predicate to match only GPR registers.
11962         (base_reg_operand): New predicate to match base registers.
11963         (quad_int_reg_operand): New predicate to match even GPR registers
11964         for quad memory operations.
11965         (vsx_reg_or_cint_operand): New predicate to allow vector logical
11966         operations in both GPR and VSX registers.
11967         (quad_memory_operand): New predicate for quad memory operations.
11968         (reg_or_indexed_operand): New predicate for direct move support.
11970         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
11971         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
11972         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
11973         (POWERPC_MASKS): Add power8 options.
11974         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
11975         various options.
11977         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
11978         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
11980         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
11981         (-mpower8-fusion): New power8 options.
11982         (-mpower8-fusion-sign): Likewise.
11983         (-mpower8-vector): Likewise.
11984         (-mcrypto): Likewise.
11985         (-mdirect-move): Likewise.
11986         (-mquad-memory): Likewise.
11988         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
11989         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
11990         registers.
11991         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
11992         (rs6000_debug_vector_unit): Add p8_vector.
11993         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
11994         definitions.  Also print fusion state.
11995         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
11996         (rs6000_builtin_mask_calculate): Add power8 builtin support.
11997         (rs6000_option_override_internal): Add support for power8.
11998         (rs6000_common_init_builtins): Add debugging for skipped builtins
11999         if -mdebug=builtin.
12000         (rs6000_adjust_cost): Add power8 support.
12001         (rs6000_issue_rate): Likewise.
12002         (insn_must_be_first_in_group): Likewise.
12003         (insn_must_be_last_in_group): Likewise.
12004         (force_new_group): Likewise.
12005         (rs6000_register_move_cost): Likewise.
12006         (rs6000_opt_masks): Likewise.
12008         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
12009         power8 capable assembler, default to power7 options.
12010         (TARGET_DIRECT_MOVE): Likewise.
12011         (TARGET_CRYPTO): Likewise.
12012         (TARGET_P8_VECTOR): Likewise.
12013         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
12014         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
12015         (VECTOR_MEM_P8_VECTOR_P): Likewise.
12016         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
12017         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
12018         (TARGET_XSCVDPSPN): Likewise.
12019         (TARGET_XSCVSPDPN): Likewsie.
12020         (TARGET_SYNC_HI_QI): Likewise.
12021         (TARGET_SYNC_TI): Likewise.
12022         (MASK_CRYPTO): Likewise.
12023         (MASK_DIRECT_MOVE): Likewise.
12024         (MASK_P8_FUSION): Likewise.
12025         (MASK_P8_VECTOR): Likewise.
12026         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
12027         temporary used by some of the direct move instructions to get two FP
12028         temporary registers does not force creation of a stack frame.
12029         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
12030         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
12031         that any VSX registers are tieable, even if they are also an
12032         Altivec vector mode.
12033         (r6000_reg_class_enum): Add wm, wr, wv constraints.
12034         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
12035         (RS6000_BTM_CRYPTO): Likewise.
12036         (RS6000_BTM_COMMON): Likewise.
12038         * config/rs6000/rs6000.md (cpu attribute): Add power8.
12039         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
12040         (enum rs6000_vector): Add power8 vector support.
12042 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12044         PR target/19599
12045         PR target/57340
12046         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
12047         (any_sibcall_could_use_r3): this and handle indirect calls.
12048         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
12050 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12052         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
12054 2013-05-22  Richard Biener  <rguenther@suse.de>
12056         PR middle-end/57349
12057         * profile.c (branch_prob): Do not split blocks that are
12058         abnormally receiving from ECF_RETURNS_TWICE functions.
12060 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
12062         * recog.c (offsettable_address_addr_space_p): Fix calculation of
12063         address mode.  Move pointer mode initialization to the same place.
12065 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
12067         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
12068         while it has any effect.
12070 2013-05-21  Easwaran Raman  <eraman@google.com>
12072         PR tree-optimization/57322
12073         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
12074         UID of the statement added to the BB to be 1.
12076 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
12078         PR tree-optimization/57331
12079         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
12080         of conversion from pointer type to integral type with integer.
12082 2013-05-21  Martin Jambor  <mjambor@suse.cz>
12084         PR lto/57289
12085         * ipa-prop.c (ipa_read_node_info): Process param_used and
12086         controlled_uses in the same order as when writing.
12088 2013-05-21  Magnus Granberg  <baldrick@free.fr>
12090         PR plugins/56754
12091         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
12093 2013-05-21  Richard Biener  <rguenther@suse.de>
12095         PR tree-optimization/57318
12096         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
12097         estimate stmts with side-effects as likely eliminated.
12099 2013-05-21  Richard Biener  <rguenther@suse.de>
12101         PR tree-optimization/57330
12102         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
12103         preserve the call stmts fntype.
12105 2013-05-21  Richard Biener  <rguenther@suse.de>
12107         PR tree-optimization/57303
12108         * tree-ssa-sink.c (statement_sink_location): Improve killing
12109         stmt detection and properly handle self-assignments.
12111 2013-05-21  Christian Bruel  <christian.bruel@st.com>
12113         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
12114         spanning registers. LEAF_REG_REMAP is supported only for contiguous
12115         registers. Set register size out of the PARALLEL loop.
12117 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
12119         PR target/56547
12120         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
12121         (*fmasf4, *fmasf4_media): New insns.
12123 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
12125         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
12126         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
12127         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
12128         (mips_idiv_insns): Update the comments to say that the returned
12129         instruction counts are in units of BASE_INSN_LENGTH.
12130         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
12131         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
12132         using 2 rather than 4 as the length of indirect MIPS16 and
12133         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
12134         length of a NOP.  Don't divide MIPS16 lengths by 2.
12135         (mips16_split_long_branches): Assume a branch is long if the
12136         length is greater than 4 rather than 8.
12137         * config/mips/mips.md (length): Give MIPS16 lengths directly,
12138         rather than multiplying them by 2.  Multiply instruction counts
12139         by BASE_INSN_LENGTH rather than 4.
12140         (*jump_mips16, tls_get_tp_mips16_<mode>)
12141         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
12143 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
12145         * config/mips/mips.md (extended_mips16): Remove branch case.
12146         (length): Remove duplicated extended_mips16 test.
12148 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
12150         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
12152 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
12154         * recog.h (Recog_data): Rename to...
12155         (recog_data_d): ...this.
12156         (recog_data): Update accordingly.
12157         * recog.c (recog_data): Likewise.
12158         * reload.c (save_recog_data): Likewise.
12159         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
12160         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
12162 2013-05-17  Julian Brown  <julian@codesourcery.com>
12164         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
12165         found in a REG_EQUAL note, invalidate it.
12167 2013-05-17  Easwaran Raman  <eraman@google.com>
12169         * tree-ssa-reassoc.c (find_insert_point): New function.
12170         (insert_stmt_after): Likewise.
12171         (get_def_stmt): Likewise.
12172         (ensure_ops_are_available): Likewise.
12173         (not_dominated_by): Likewise.
12174         (rewrite_expr_tree): Do not move statements beyond what is
12175         necessary. Remove call to swap_ops_for_binary_stmt...
12176         (reassociate_bb): ... and move it here.
12177         (build_and_add_sum): Assign UIDs for new statements.
12178         (linearize_expr): Likewise.
12179         (do_reassoc): Renumber gimple statement UIDs.
12181 2013-05-17  Jan Hubicka  <jh@suse.cz>
12183         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
12184         weakrefs.
12185         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
12186         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
12187         weakrefs.
12188         (output_weakrefs): Update.
12190 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
12191             Martin Jambor  <mjambor@suse.cz>
12193         PR middle-end/57276
12194         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
12195         value that corresponds to the given aggval is found in values vector.
12197 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
12199         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
12200         sse, sse2, sse3, ssse3 and sse4a flags to options.
12202 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
12204         * gengtype-state.c: (s_expr_writer): New class, to handle
12205         prettifying of output layout of s-expressions.
12206         (state_writer): New class, to write out gtype.state.
12207         (state_written_type_count): Move this variable into member data of
12208         state_writer.
12209         (s_expr_writer::s_expr_writer): New code: constructor for new class
12210         (state_writer::state_writer(): ditto
12211         (s_expr_writer::write_new_line): New function
12212         (s_expr_writer::write_any_indent): ditto
12213         (s_expr_writer::begin_s_expr): ditto
12214         (s_expr_writer::end_s_expr): ditto
12215         (write_state_fileloc): convert to method of state_writer...
12216         (state_writer:: write_state_fileloc): ...and use methods of
12217         s_expr_writer to write indentation into the gtype.state output file
12218         to visually represent the hierarchical structure of the list
12219         structures
12220         (write_state_fields): ditto, renaming to...
12221         (state_writer::write_state_fields)
12222         (write_state_a_string): ditto, renaming to...
12223         (state_writer::write_state_a_string)
12224         (write_state_string_option): ditto, renaming to...
12225         (state_writer::write_state_string_option)
12226         (write_state_type_option): ditto, renaming to...
12227         (state_writer::write_state_type_option)
12228         (write_state_nested_option): ditto, renaming to...
12229         (state_writer::write_state_nested_option)
12230         (write_state_option): ditto, renaming to...
12231         (state_writer::write_state_option)
12232         (write_state_options): ditto, renaming to...
12233         (state_writer::write_state_options)
12234         (write_state_lang_bitmap): ditto, renaming to...
12235         (state_writer::write_state_lang_bitmap)
12236         (write_state_version): ditto, renaming to...
12237         (state_writer::write_state_version)
12238         (write_state_scalar_type): ditto, renaming to...
12239         (state_writer::write_state_scalar_type)
12240         (write_state_string_type): ditto, renaming to...
12241         (state_writer::write_state_string_type)
12242         (write_state_undefined_type): ditto, renaming to...
12243         (state_writer::write_state_undefined_type)
12244         (write_state_struct_union_type): ditto, renaming to...
12245         (state_writer::write_state_struct_union_type)
12246         (write_state_struct_type): ditto, renaming to...
12247         (state_writer::write_state_struct_type)
12248         (write_state_user_struct_type): ditto, renaming to...
12249         (state_writer::write_state_user_struct_type)
12250         (write_state_lang_struct_type): ditto, renaming to...
12251         (state_writer::write_state_lang_struct_type)
12252         (write_state_param_struct_type): ditto, renaming to...
12253         (state_writer::write_state_param_struct_type)
12254         (write_state_pointer_type): ditto, renaming to...
12255         (state_writer::write_state_pointer_type)
12256         (write_state_array_type): ditto, renaming to...
12257         (state_writer::write_state_array_type)
12258         (write_state_gc_used): ditto, renaming to...
12259         (state_writer::write_state_gc_used)
12260         (write_state_common_type_content): ditto, renaming to...
12261         (state_writer::write_state_common_type_content)
12262         (write_state_type): ditto, renaming to...
12263         (state_writer::write_state_type)
12264         (write_state_pair_list): ditto, renaming to...
12265         (state_writer::write_state_pair_list)
12266         (write_state_pair): ditto, renaming to...
12267         (state_writer::write_state_pair)
12268         (write_state_typedefs): ditto, renaming to...
12269         (state_writer::write_state_typedefs)
12270         (write_state_structures): ditto, renaming to...
12271         (state_writer::write_state_structures)
12272         (write_state_param_structs): ditto, renaming to...
12273         (state_writer::write_state_param_structs)
12274         (write_state_variables): ditto, renaming to...
12275         (state_writer::write_state_variables)
12276         (write_state_srcdir): ditto, renaming to...
12277         (state_writer::write_state_srcdir)
12278         (write_state_files_list): ditto, renaming to...
12279         (state_writer::write_state_files_list)
12280         (write_state_languages): ditto, renaming to...
12281         (state_writer::write_state_languages)
12282         (write_state): create a state_writer instance and use it when
12283         writing out the state file
12285 2013-05-17  Mike Stump  <mikestump@comcast.net>
12287         PR rtl-optimization/57304
12288         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
12289         accessing DF_REF_REAL_LOC.
12291 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
12293         PR rtl-optimization/57281
12294         PR rtl-optimization/57300
12295         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
12296         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
12297         what the other splitter did if the registers are dead.
12299 2013-05-17  Richard Biener  <rguenther@suse.de>
12301         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
12302         MEM_REF offsets.
12304 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
12306         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
12307         linking.
12309 2013-05-17  Marek Polacek  <polacek@redhat.com>
12311         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
12312         length when doing non-zero store of storing '\0' to '\0'.
12314 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
12316         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
12317         vect_external_def oprnd1 with loop_vinfo, try to emit
12318         optional cast, negation and and stmts on the loop preheader
12319         edge instead of into the pattern def seq.
12321         PR tree-optimization/57051
12322         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
12323         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
12325 2013-05-16  Nick Clifton  <nickc@redhat.com>
12327         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
12328         (rl78_is_naked_func): New function.
12329         (rl78_expand_prologue): Skip prologue generation for naked functions.
12330         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
12331         * doc/extend.texi (naked): Add RL78 to the list of processors
12332         that supports this attribute.
12334 2013-05-16  Jeff Law  <law@redhat.com>
12336         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
12338 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
12340         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
12341         cache parameters using detect_caches_amd also for CYRIX,
12342         NSC and TM2 signatures.
12344 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
12345             Dzianis Kahanovich  <mahatma@eu.by>
12347         PR target/45359
12348         PR target/46396
12349         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
12350         VIA/Centaur processors and determine their cache parameters
12351         using detect_caches_amd.
12353 2013-05-16  Teresa Johnson  <tejohnson@google.com>
12355         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
12356         (rtl_verify_edges): New function.
12357         (rtl_verify_bb_insns): Ditto.
12358         (rtl_verify_bb_pointers): Ditto.
12359         (rtl_verify_bb_insn_chain): Ditto.
12360         (rtl_verify_fallthru): Ditto.
12361         (rtl_verify_bb_layout): Ditto.
12362         (rtl_verify_flow_info_1): Outline checks into new functions.
12363         (rtl_verify_flow_info): Ditto.
12365 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
12367         * cfghooks.c (copy_bbs): Add update_dominance argument.
12368         * cfghooks.h (copy_bbs): Update prototype.
12369         * tree-cfg.c (gimple_duplicate_sese_region):
12370         Add update_dominance argument.
12371         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
12372         * tree-ssa-loop-ch.c (copy_loop_headers): Update
12373         gimple_duplicate_sese_region call.
12374         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
12375         Update copy_bbs call.
12376         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
12377         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
12379 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
12381         * tree-vectorizer.h (NUM_PATTERNS): Increment.
12382         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
12383         vect_recog_rotate_pattern.
12384         (vect_recog_rotate_pattern): New function.
12386 2013-05-16  Jason Merrill  <jason@redhat.com>
12388         * Makefile.in (LLINKER): New variable.
12389         (mostlyclean): Remove link mutex.
12390         * configure.ac: Handle --enable-link-mutex.
12391         * lock-and-run.sh: New script.
12393 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12395         PR target/19599
12396         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
12397         for NULL decl.
12399 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
12401         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
12403 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
12405         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
12406         * config/arm/arm.c (next_consecutive_mem): New function.
12407         (gen_movmem_ldrd_strd): Likewise.
12408         * config/arm/arm.md (movmemqi): Update condition and code.
12409         (unaligned_loaddi, unaligned_storedi): New patterns.
12411 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
12413         * config.gcc: Obsolete *-*-solaris2.9*.
12414         * doc/install.texi (Specific, *-*-solaris2*): Document it.
12416 2013-05-16  Richard Biener  <rguenther@suse.de>
12418         * passes.c (init_optimization_passes): Move pass_parallelize_loops
12419         earlier, after GRAPHITE transforms and IV canonicalization.
12421 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
12423         * omp-low.c (extract_omp_for_data): For collapsed loops,
12424         if at least one of the loops is known at compile time to
12425         iterate zero times, set count to 0.
12426         (expand_omp_regimplify_p): New function.
12427         (expand_omp_for_generic): For collapsed loops, if at least
12428         one of the loops isn't known to iterate at least once,
12429         add runtime check with setting count to 0.
12430         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
12431         For unsigned types if it isn't known at compile time that
12432         the loop will iterate at least once, add runtime check to bypass
12433         the whole loop if initial condition isn't true.
12435 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
12437         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
12439 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
12441         PR middle-end/57286
12442         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
12443         transformations to avoid an infinite loop.
12445 2013-05-16  Marek Polacek  <polacek@redhat.com>
12447         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
12449 2013-05-15  Leif Ekblad  <leif@rdos.net>
12451         * config/i386/i386.c (ix86_decompose_address): Use
12452         DEFAULT_TLS_SEG_REG to access TLS segment register.
12453         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
12454         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
12455         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
12457 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
12459         PR target/57260
12460         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
12461         sibling calls to functions that would normally be lazily bound,
12462         unless $gp is call-clobbered.
12464 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
12466         * config/i386/i386.c (ix86_option_override_internal): Update
12467         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
12468         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
12469         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
12470         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
12471         of TARGET_3DNOW.
12472         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
12474 2013-05-15  Andreas Schwab  <schwab@suse.de>
12476         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
12477         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
12478         third operand.
12480 2013-05-15  Teresa Johnson  <tejohnson@google.com>
12482         * loop-unroll.c (report_unroll_peel): Check decision before
12483         emitting unroll/peel message.
12485 2013-05-15  Teresa Johnson  <tejohnson@google.com>
12487         * function.h (has_bb_partition): New rtl_data flag.
12488         (bb_reorder_complete): Ditto.
12489         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
12490         instead of flag_reorder_blocks_and_partition.
12491         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
12492         with some enhancements.
12493         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
12494         * bb-reorder.c (connect_traces): Check for has_bb_partition
12495         instead of flag_reorder_blocks_and_partition.
12496         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
12497         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
12498         verify_hot_cold_block_grouping.
12499         (partition_hot_cold_basic_blocks): Set has_bb_partition.
12501 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12503         PR target/19599
12504         * config/arm/predicates.md (call_insn_operand): New predicate.
12505         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
12506         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
12507         if insn is not a tail call.
12508         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
12509         registers.
12510         * config/arm/arm.h (enum reg_class): New caller save register class.
12511         (REG_CLASS_NAMES): Likewise.
12512         (REG_CLASS_CONTENTS): Likewise.
12513         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
12514         without decls.
12516 2013-05-15  Richard Biener  <rguenther@suse.de>
12518         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
12519         of MSG_OPTIMIZED_LOCATIONS.
12520         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
12521         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
12522         message.
12523         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
12524         of MSG_OPTIMIZED_LOCATIONS.
12525         (execute_vect_slp): Likewise.
12526         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
12527         (vect_create_cond_for_alias_checks): Likewise.
12528         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
12529         (vect_recog_widen_mult_pattern): Likewise.
12530         (vect_recog_widen_sum_pattern): Likewise.
12531         (vect_recog_over_widening_pattern): Likewise.
12532         (vect_recog_widen_shift_pattern): Likewise.
12533         (vect_recog_vector_vector_shift_pattern): Likewise.
12534         (vect_recog_divmod_pattern): Likewise.
12535         (vect_recog_mixed_size_cond_pattern): Likewise.
12536         (vect_recog_bool_pattern): Likewise.
12537         (vect_pattern_recog_1): Likewise.
12539 2013-05-15  Martin Jambor  <mjambor@suse.cz>
12541         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
12542         non-functions to builtin_unreachable.
12543         * ipa-inline-transform.c (inline_call): Do not assert estimates were
12544         correct when new direct edges were discovered.
12546 2013-05-15  Martin Jambor  <mjambor@suse.cz>
12548         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
12549         header, print symbol order instead of node uid, print more information
12550         about indirect edge targets.
12551         (ipa_make_edge_direct_to_target): Print symbol order instead of node
12552         uids.
12553         (ipa_make_edge_direct_to_target): Likewise.
12554         (remove_described_reference): Likewise.
12555         (propagate_controlled_uses): Likewise.
12556         (ipa_print_node_params): Also print symbol order.
12557         (ipcp_transform_function): Print symbol order instead of node uids.
12558         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
12559         (cgraph_get_create_real_symbol_node): Likewise.
12560         * ipa-cp.c (print_lattice): Likewise.
12561         (print_all_lattices): Likewise.
12562         (determine_versionability): Likewise.
12563         (initialize_node_lattices): Likewise.
12564         (estimate_local_effects): Likewise.
12565         (update_profiling_info): Likewise.
12566         (create_specialized_node): Likewise.
12567         (perhaps_add_new_callers): Likewise.
12568         (decide_about_value): Likewise.
12569         (decide_whether_version_node): Likewise.
12570         (identify_dead_nodes): Likewise.
12571         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
12572         (dump_inline_summary): Likewise.
12573         (estimate_node_size_and_time): Likewise.
12574         (inline_analyze_function): Likewise.
12575         * ipa-inline.c (report_inline_failed_reason): Likewise.
12576         (want_early_inline_function_p): Likewise.
12577         (edge_badness): Likewise.
12578         (update_edge_key): Likewise.
12579         (inline_small_functions): Likewise.  Add dumping of order to two other
12580         dumps.
12581         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
12582         instead of node uids.
12583         (propagate_pure_const): Likewise.
12584         (propagate_pure_const): Likewise.
12585         * ipa-utils.c (dump_cgraph_node_set): Likewise.
12586         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
12587         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
12588         of node uids.
12589         * tree-pretty-print.c (dump_function_header): Likewise.
12590         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
12591         Print symbol order instead of node uids.
12593 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12595         * config/s390/s390.c (s390_register_move_cost): Don't impose the
12596         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
12598 2013-05-15  Richard Biener  <rguenther@suse.de>
12600         PR tree-optimization/57275
12601         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
12602         return value for fail to do runtime alias checks for gather loads.
12604 2013-05-15  Jan Hubicka  <jh@suse.cz>
12606         PR lto/57038
12607         PR lto/47375
12608         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
12609         weakrefs are not external.
12610         (lto_symtab_merge_decls): Fix thinko when dealing with
12611         non-lto_symtab decls.
12612         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
12613         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
12614         * varpool.c (dump_varpool_node): Dump more flags.
12616 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
12618         * config/i386/i386.c (processor_alias_table): Add instruction
12619         FSGSBASE for AMD bdver3 architecture.
12621 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
12623         * tree.c (warn_deprecated_use): Print file:line using locus color.
12624         * diagnostic.c (diagnostic_report_current_module): Print file:line
12625         and file:line:column using locus color.
12627 2013-05-14  Mike Stump  <mikestump@comcast.net>
12629         * gdbinit.in: Add __null.
12631 2013-05-14  Mike Stump  <mikestump@comcast.net>
12633         * recog.h: Rename struct recog_data to Recog_data.
12634         * recog.c: Likewise.
12635         * reload.c (can_reload_into): Likewise.
12636         * config/picochip/picochip.c: Likewise.
12638 2013-05-14  Mike Stump  <mikestump@comcast.net>
12640         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
12642 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
12644         * resource.h (struct resources): Remove unch_memory member.
12645         (CLEAR_RESOURCE): Don't clear unch_memory.
12646         * resource.c (mark_referenced_resources): Don't set it.
12647         (mark_set_resources): Likewise.
12648         (mark_target_live_regs): Don't clear it.
12649         (init_resource_info): Likewise.
12650         * reorg.c (resource_conflicts_p): Don't compare it.
12651         (redundant_insn): Don't set it.
12653         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
12654         Remove prototypes.
12655         * emit-rtl.c (next_label): Remove unused function.
12656         (skip_consecutive_labels, link_cc0_insns): Move to ...
12657         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
12658         only place where these functions are used, and make them static.
12660 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
12662         * fold-const.c (fold_negate_expr): Handle vectors.
12663         (fold_truth_not_expr): Make it static.
12664         (fold_invert_truthvalue): New static function.
12665         (invert_truthvalue_loc): Handle vectors. Do not call
12666         fold_truth_not_expr directly.
12667         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
12668         <TRUTH_NOT_EXPR>: Do not cast to boolean.
12669         (fold_comparison): Handle vector constants.
12670         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
12671         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
12672         * tree.h (fold_truth_not_expr): Remove declaration.
12674 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
12676         * config/aarch64/aarch64-simd.md
12677         (aarch64_vcond_internal<mode>): Rename to...
12678         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
12679         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
12680         float modes. Clarify all iterator modes.
12681         (vcond<mode><mode>): Use new name for vcond expanders.
12682         (vcond<v_cmp_result><mode>): Likewise.
12683         (vcondu<mode><mode>: Likewise.
12684         * config/aarch64/iterators.md (VDQF_COND): New.
12686 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
12688         PR bootstrap/57266
12689         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
12690         variable for the shift amount. Check that we shift by non-negative
12691         amounts.
12693 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
12695         PR target/42017
12696         * config/arm/arm.h (EPILOGUE_USES): Only return true
12697         for LR_REGNUM after epilogue_completed.
12699 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
12701         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
12702         is error_mark_node.
12704 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
12706         PR target/57261
12707         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
12708         and Solaris 11+/x86 with gld.
12709         * configure: Regenerate.
12711 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
12713         * expmed.c (expand_shift_1): Canonicalize rotates by
12714         constant bitsize / 2 to bitsize - 1.
12715         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
12716         case ROTATERT>: Likewise.
12718         Revert:
12719         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
12721         * config/i386/i386.md (rotateinv): New code attr.
12722         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
12723         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
12724         roll $31, %eax, etc.
12726 2013-05-14  Richard Biener  <rguenther@suse.de>
12728         PR middle-end/57235
12729         * tree-eh.c (sink_clobbers): Give up for successors with
12730         multiple predecessors and no virtual uses.
12732 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
12734         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
12735         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
12737 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
12739         PR middle-end/57251
12740         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
12741         the case when both op0 and op1 have VOIDmode.
12743 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
12745         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
12746         in multiply-accumulate mode.
12748 2013-05-13  Guozhi Wei  <carrot@google.com>
12750         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
12752 2013-05-13  Kai Tietz  <ktietz@redhat.com>
12754         PR target/56975
12755         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
12756         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
12757         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
12758         * config/i386/i386.c (ix86_option_override_internal): Likewise.
12759         (ix86_expand_prologue): Likewise.
12760         (ix86_expand_split_stack_prologue): Likewise.
12761         (legitimate_pic_address_disp_p): Likewise.
12762         (legitimize_pic_address): Likewise.
12763         (legitimize_tls_address): Likewise.
12764         (legitimize_pe_coff_symbol): Likewise.
12765         (output_pic_addr_const): Likewise.
12766         (construct_plt_address): Likewise.
12767         (ix86_expand_call): Likewise.
12768         (x86_output_mi_thunk): Likewise.
12769         (x86_function_profiler): Likewise.
12771 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
12773         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
12774         similar switch cases.
12775         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
12776         (aarch64_simd_mov_to_<mode>low): Delete.
12777         (aarch64_simd_mov_to_<mode>high): Delete.
12778         (move_lo_quad_<mode>): Add w<-r alternative.
12779         (aarch64_simd_move_hi_quad_<mode>): Likewise.
12780         (aarch64_simd_mov_from_*): Update type attribute.
12781         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
12782         statement.
12784 2013-05-13  Jan Hubicka  <jh@suse.cz>
12786         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
12787         * config/i386/i386.c (ix86_compute_frame_layout,
12788         ix86_expand_epilogue, emit_i387_cw_initialization,
12789         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
12790         ix86_local_alignment): Fix use of size/speed predicates.
12792 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
12794         PR tree-optimization/45216
12795         PR tree-optimization/57157
12796         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
12797         the (-Y) & (B - 1) variant if OP is |.
12798         * expmed.c (expand_shift_1): For rotations by const0_rtx just
12799         return shifted.  Use (-op1) & (prec - 1) as other_amount
12800         instead of prec - op1.
12802 2013-05-13  Martin Jambor  <mjambor@suse.cz>
12804         PR middle-end/42371
12805         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
12806         (ipa_constant_data): New type.
12807         (ipa_jump_func): Use ipa_constant_data to hold information about
12808         constant jump functions.
12809         (ipa_get_jf_constant): Adjust to jump function type changes.
12810         (ipa_get_jf_constant_rdesc): New function.
12811         (ipa_param_descriptor): New field controlled_uses.
12812         (ipa_get_controlled_uses): New function.
12813         (ipa_set_controlled_uses): Likewise.
12814         * ipa-ref.h (ipa_find_reference): Declare.
12815         * ipa-prop.c (ipa_cst_ref_desc): New type.
12816         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
12817         changes.
12818         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
12819         New parameter cs.  Adjust all callers.
12820         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
12821         (remove_described_reference): New function.
12822         (jfunc_rdesc_usable): Likewise.
12823         (try_make_edge_direct_simple_call): Decrement controlled use count,
12824         attempt to remove reference if it hits zero.
12825         (combine_controlled_uses_counters): New function.
12826         (propagate_controlled_uses): Likewise.
12827         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
12828         (ipa_edge_duplication_hook): Duplicate reference descriptions.
12829         (ipa_print_node_params): Print described use counter.
12830         (ipa_write_jump_function): Adjust to jump function type changes.
12831         (ipa_read_jump_function): New parameter CS, pass it to
12832         ipa_set_jf_constant.  Adjust caller.
12833         (ipa_write_node_info): Stream controlled use count
12834         (ipa_read_node_info): Likewise.
12835         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
12836         asserting.
12837         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
12838         count.  Remove cloning-added reference if it reaches zero.
12839         * ipa-ref.c (ipa_find_reference): New function.
12841 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
12843         * config/i386/i386.c (processor_target_table): Modified default
12844         alignment values for AMD BD and BT architectures.
12846 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
12848         * tree-vect-generic.c (uniform_vector_p): Move ...
12849         * tree.c (uniform_vector_p): ... here.
12850         * tree.h (uniform_vector_p): Declare it.
12851         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
12852         into a scalar.
12854 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
12856         PR tree-optimization/57230
12857         * tree-ssa-strlen.c (handle_char_store): Record length for
12858         array store from STRING_CST.
12860         PR tree-optimization/57230
12861         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
12862         check.
12864 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
12866         * config/epiphany/epiphany.c (epiphany_init): Check size of
12867         NUM_MODES_FOR_MODE_SWITCHING.
12868         (epiphany_expand_prologue):
12869         Remove CONFIG_REGNUM initial value handling code.
12870         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
12871         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
12872         (emit_set_fp_mode, epiphany_mode_after): Likewise.
12873         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
12874         Don't return 1 for FP_MODE_NONE.
12875         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
12876         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
12877         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
12878         * config/epiphany/epiphany.md (save_config): New pattern.
12880 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
12882         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
12884 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
12886         * config/i386/i386.md (memory): Handle sseishft1.
12887         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
12888         (*vec_extractv2di_1): Ditto.
12890 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
12892         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
12893         saved registers.
12895 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
12897         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
12898         Add mthumb/march=armv7-a multilib.
12899         Add mthumb/march=armv7-r multilib.
12900         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
12902 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
12904         * config/v850/t-rtems: Add more multilibs.
12906 2013-05-10  Richard Biener  <rguenther@suse.de>
12908         PR tree-optimization/57214
12909         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
12910         not propagate from SSA names that occur in abnormal PHI nodes.
12912 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
12914         * stor-layout.c (element_precision): New function.
12915         * machmode.h (element_precision): Declare it.
12916         * tree.c (build_minus_one_cst): New function.
12917         (element_precision): Likewise.
12918         * tree.h (build_minus_one_cst): Declare new function.
12919         (element_precision): Likewise.
12920         * fold-const.c (operand_equal_p): Use element_precision.
12921         (fold_binary_loc): Handle vector types.
12922         * convert.c (convert_to_integer): Use element_precision.
12923         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
12924         separately.
12926 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
12928         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
12929         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
12930         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
12931         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
12932         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
12933         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
12934         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
12935         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
12936         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
12937         (Uuw8): New constraints.
12938         (Usb4): Move into alphabetical order.
12939         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
12940         (sd8_operand, ub8_operand, uw8_operand): New predicates.
12941         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
12942         previously unnamed patterns.
12943         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
12944         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
12945         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
12946         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
12947         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
12948         of specific lengths.
12950 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
12952         * config/i386/i386.md (rotateinv): New code attr.
12953         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
12954         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
12955         roll $31, %eax, etc.
12957         PR tree-optimization/45216
12958         PR tree-optimization/57157
12959         * tree-ssa-forwprop.c (simplify_rotate): New function.
12960         (ssa_forward_propagate_and_combine): Call it.
12962 2013-05-10  Richard Biener  <rguenther@suse.de>
12964         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
12965         disable peeling when we version for aliasing.
12966         (vector_alignment_reachable_p): Honor explicit user alignment.
12967         (vect_supportable_dr_alignment): Likewise.
12968         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
12969         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
12970         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
12971         then peeling to arrange for the cost-model check to come first.
12973 2013-05-10  Alan Modra  <amodra@gmail.com>
12975         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
12976         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
12977         * configure: Regenerate.
12979 2013-05-10  Alan Modra  <amodra@gmail.com>
12981         PR target/55033
12982         * varasm.c (default_elf_select_section): Move !DECL_P check..
12983         (get_named_section): ..to here before calling get_section_name.
12984         Adjust assertion.
12985         (default_section_type_flags): Add DECL_P check.
12986         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
12987         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
12989 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
12991         * config/epiphany/epiphany.c (epiphany_expand_prologue):
12992         When using gen_stack_adjust_str with a register offset, add a
12993         REG_FRAME_RELATED_EXPR note.
12995 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
12997         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
12998         (*vec_extractv4si_zext_mem): Ditto.
12999         (*vec_extractv2di): Add 0->x and x->x alternatives.
13000         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
13001         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
13003 2013-05-09  Jason Merrill  <jason@redhat.com>
13005         N3639 C++1y VLA support
13006         * gimplify.c (gimplify_vla_decl): Don't touch an existing
13007         DECL_VALUE_EXPR.
13009         * tree.c (build_constructor_va): New.
13010         * tree.h: Declare it.
13012 2013-05-09  Martin Jambor  <mjambor@suse.cz>
13014         PR lto/57084
13015         * gimple-fold.c (canonicalize_constructor_val): Call
13016         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
13018 2013-05-09  Jan Hubicka  <jh@suse.cz>
13019             Richard Biener  <rguenther@suse.de>
13021         PR lto/54095
13022         * symtab.c (symtab_make_decl_local): Do not add private names.
13024 2013-05-09  Jan Hubicka  <jh@suse.cz>
13026         PR lto/54095
13027         * symtab.c (insert_to_assembler_name_hash): Handle clones.
13028         (unlink_from_assembler_name_hash): Likewise.
13029         (symtab_prevail_in_asm_name_hash, symtab_register_node,
13030         symtab_unregister_node, symtab_initialize_asm_name_hash,
13031         change_decl_assembler_name): Update.
13033 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
13035         * config/aarch64/aarch64.md: New movtf split.
13036         (*movtf_aarch64): Update.
13037         (aarch64_movdi_tilow): Handle TF modes and rename to
13038         aarch64_movdi_<mode>low.
13039         (aarch64_movdi_tihigh): Handle TF modes and rename to
13040         aarch64_movdi_<mode>high
13041         (aarch64_movtihigh_di): Handle TF modes and rename to
13042         aarch64_mov<mode>high_di
13043         (aarch64_movtilow_di): Handle TF modes and rename to
13044         aarch64_mov<mode>low_di
13045         (aarch64_movtilow_tilow): Remove spurious whitespace.
13046         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
13047         splits.
13048         (aarch64_print_operand): Update.
13050 2013-05-09  Alan Modra  <amodra@gmail.com>
13052         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
13053         powerpc64le.
13054         * configure: Regenerate.
13056 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
13058         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
13059         splitter preparation statements.
13060         * config/i386/sse.md (*vec_extract* splitters): Ditto.
13061         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
13062         adjust_address_nv.
13064 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13066         * gimple-ssa-strength-reduction.c (count_candidates): Change
13067         return value to int.
13068         (analyze_candidates_and_replace): Change type of length to int.
13070 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
13072         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
13073         (*vec_extract<mode>): Use VI12_128 mode iterator.
13074         (*vec_extract<mode>_mem): Ditto.
13075         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
13076         attribute.
13078 2013-05-08  Diego Novillo  <dnovillo@google.com>
13080         PR bootstrap/54659
13082         Revert:
13083         2012-08-17  Diego Novillo  <dnovillo@google.com>
13085         PR bootstrap/54281
13086         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
13087         * config.in: Regenerate.
13088         * configure: Regenerate.
13089         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
13091 2013-05-08  Jan Hubicka  <jh@suse.cz>
13093         PR lto/54095
13094         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
13095         * cgraph.h (symtab_node_base): Add unique_name.
13096         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
13097         input_overwrite_node, input_varpool_node): Stream unique_name.
13098         * cgraphclones.c (cgraph_create_virtual_clone,
13099         cgraph_function_versioning): Set unique_name.
13100         * ipa.c (function_and_variable_visibility): Set unique_name.
13102 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13104         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
13105         (alloc_cand_and_find_basis): Restrict conditional candidate
13106         processing to CAND_MULTs.
13108 2013-05-08  Jan Hubicka  <jh@suse.cz>
13110         PR lto/54095
13111         lto-symtab.c (lto_symtab_symbol_p): New function.
13112         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
13113         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
13114         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
13115         Skip static symbols.
13117 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
13119         PR tree-optimization/57200
13120         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
13121         Only call inform if the preceding warning_at returns true.
13123 2013-05-07  Han Shen  <shenhan@google.com>
13125         * cfgexpand.c (record_or_union_type_has_array_p): New function.
13126         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
13127         * common.opt (fstack-protector-strong): New option.
13128         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
13129         * doc/invoke.texi (Optimization Options): Document
13130         "-fstack-protector-strong".
13131         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
13133 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
13135         * config/mips/mips.c (mips_machine_reorg2): Return 0.
13137 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
13139         * ira.c (update_equiv_regs): Add insn having equiv memory even if
13140         it is not lhs of the insn.
13141         (setup_reg_equiv): Remove insn having equiv memory which it is not
13142         lhs of the insn.
13143         * lra-constraints.c (process_address): Try to improve generation
13144         code for address base + disp.
13145         (lra_constraints): Make correct the code for checking insn setting
13146         up backward equivalence.  Remove insn only if it is in the init
13147         insn list.
13148         * lra-eliminations.c (update_reg_eliminate): Change return value.
13149         (lra_eliminate): Use the result.
13151 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
13153         * config/i386/sse.md (ssescalarnummask): New mode attribute.
13154         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
13155         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
13156         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
13157         register target operands.
13158         (*vec_extractv8hi_sse2): New pattern.
13159         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
13160         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
13161         (*vec_extract<mode>_mem): New insn and split pattern.
13163 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
13165         * config/arm/arm.c (arm_asan_shadow_offset): New function.
13166         (TARGET_ASAN_SHADOW_OFFSET): Define.
13167         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
13168         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
13170 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13172         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
13173         (incr_vec_index): Return -1 if increment not found.
13174         (create_add_on_incoming_edge): Assert if increment not found.
13175         (record_increment): Limit number of increments recorded.
13176         (all_phi_incrs_profitable): Return false if an increment not found.
13177         (replace_profitable_candidates): Don't process increments that were
13178         not recorded.
13179         (analyze_candidates_and_replace): Limit size of incr_vec.
13181 2013-05-07  Richard Biener  <rguenther@suse.de>
13183         * calls.c (special_function_p): setjmp-like functions are leaf.
13184         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
13185         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
13187 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
13189         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
13190         (aarch64_simd_mov<mode>): New expander.
13191         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
13192         (aarch64_simd_mov_to_<mode>high): Likewise.
13193         (aarch64_simd_mov_from_<mode>low): Likewise.
13194         (aarch64_simd_mov_from_<mode>high): Likewise.
13195         (aarch64_dup_lane<mode>): Update.
13196         (aarch64_dup_lanedi): New instruction pattern.
13197         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
13198         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
13200 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13202         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
13203         (replace_mult_candidate): Remove unnecessary argument; remove
13204         unnecessary parameter from call to introduce_cast_before_cand.
13205         (replace_unconditional_candidate): Remove unnecessary parameter
13206         from call to replace_mult_candidate.
13207         (replace_conditional_candidate): Likewise.
13208         (insert_initializers): Use make_temp_ssa_name.
13209         (introduce_cast_before_cand): Remove unnecessary argument; use
13210         make_temp_ssa_name.
13211         (replace_one_candidate): Remove unnecessary argument; remove
13212         unnecessary parameter from calls to introduce_cast_before_cand.
13213         (replace_profitable_candidates): Remove unnecessary parameters
13214         from calls to replace_one_candidate.
13216 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13218         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13219         phi def as possibly hiding a basis for a CAND_ADD whose operands
13220         have been commuted in the analysis.
13221         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
13223 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13225         * config/aarch64/aarch64.md
13226         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
13227         shift value between 0-4.
13229 2013-05-07  Richard Biener  <rguenther@suse.de>
13231         * double-int.h (rshift): New overload.
13232         * double-int.c (rshift): New function.
13233         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
13234         (create_reference_ops_from_ref): Remove.
13235         (vn_reference_insert): Use shared ops for constructing the
13236         reference and copy it.
13238 2013-05-07  Richard Biener  <rguenther@suse.de>
13240         PR middle-end/57190
13241         * tree-eh.c (sink_clobbers): Properly propagate
13242         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
13244 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
13246         PR tree-optimization/57149
13247         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
13248         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
13249         collect_phi_def_edges, execute_late_warn_uninitialized): Use
13250         uninit_undefined_value_p instead of ssa_undefined_value_p.
13252         PR debug/57184
13253         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
13254         for modifier == EXPAND_INITIALIZER.
13256 2013-05-07  Anton Blanchard  <anton@samba.org>
13258         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
13259         for powerpc64 little endian.
13260         * configure: Regenerate.
13262 2013-05-06  Graham Stott  <grahams@btinternet.com>
13264         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
13265         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
13266         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
13267         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
13269 2013-05-06  Graham Stott  <grahams@btinternet.com>
13271         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
13272         codes which allow non-lvalues.
13274 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
13276         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
13277         components are all 1s.
13278         (integer_minus_onep): New function.
13279         * tree.h (integer_minus_onep): Declare it.
13280         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
13281         integer_minus_onep instead of integer_all_onesp.
13283 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
13285         PR target/52933
13286         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
13287         variations of these patterns.
13289 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
13291         * config/i386/i386.md (isa): Add x64_sse4 member.
13292         (enabled): Handle x64_sse4.
13293         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
13294         instruction for 64bit SSE4_1 targets.  Update insn attributes.
13295         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
13296         instruction for SSE4_1 targets.  Update insn attributes.
13297         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
13298         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
13299         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
13300         const_1 selector.
13301         (*vec_extractv4si): Rename from *sse4_1_pextrd.
13302         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
13303         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
13305 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
13307         PR target/57108
13308         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
13310 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
13312         * final.c (do_assembler_dialects): Don't handle curly braces and
13313         vertical bar escaped by % as dialect delimiters.
13314         (output_asm_insn): Print curly braces and vertical bar if escaped
13315         by % and ASSEMBLER_DIALECT defined.
13316         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
13317         * doc/tm.texi: Regenerated.
13319 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
13321         * config/mips/mips.c: Include tree-pass.h.
13322         (mips_reorg): Split in pre- and post-dbr_schedule parts.
13323         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
13324         (pass_mips_machine_reorg2): New machine specific pass.
13325         (insert_pass_mips_machine_reorg2): New pass plugin definition.
13326         (mips_option_override): Register the new pass.
13327         * rtl.h (cleanup_barriers): Remove prototype.
13328         (dbr_schedule): Likewise.
13329         * jump.c (cleanup_barriers): Make static.
13330         * reorg.c (dbr_schedule): Likewise.
13332 2013-05-06  Richard Biener  <rguenther@suse.de>
13334         PR tree-optimization/57185
13335         * tree-parloops.c (add_field_for_reduction): Handle anonymous
13336         SSA names properly.
13338 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
13340         PR target/57106
13341         * config/i386/i386.c (add_parameter_dependencies): Add dependence
13342         between "first_arg" and "insn", not "last" and "insn".
13344 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13346         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
13347         (find_candidates_in_block): Re-enable slsr_process_phi.
13348         (create_phi_basis): Fix double counting of candidate adjustment.
13350 2013-05-06  Richard Biener  <rguenther@suse.de>
13352         PR middle-end/57147
13353         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
13354         the edge is also fallthru, preserve it and just clear the
13355         abnormal flag.
13356         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
13357         also complex, preserve that and just clear the fallthru flag.
13358         * tree-inline.c (update_ssa_across_abnormal_edges): Also
13359         update virtual operands.
13361 2013-05-06  Alan Modra  <amodra@gmail.com>
13363         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
13364         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
13365         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
13366         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
13367         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
13368         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
13370 2013-05-06  Alan Modra  <amodra@gmail.com>
13372         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
13373         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
13374         (DEFAULT_ASM_ENDIAN): Define.
13375         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
13376         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
13377         Update -K PIC clause from sysv4.h.
13378         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
13379         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
13381 2013-05-06  Alan Modra  <amodra@gmail.com>
13383         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
13384         twice for little-endian.
13385         (ashrdi3_no_power, ashrdi3): Support little-endian.
13387 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
13389         PR target/55303
13390         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
13391         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
13392         related expanders.
13393         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
13394         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
13395         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
13396         New predicates.
13398 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
13399             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
13401         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
13402         * config/pa/pa.opt: Make mbig-switch a no-op.
13403         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
13404         (CASE_VECTOR_MODE): Always return SImode.
13405         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
13406         for the !TARGET_BIG_SWITCH case.
13407         * config/pa/pa-linux.h: Likewise.
13408         * config/pa/pa-openbsd.h: Likewise.
13409         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
13410         * config/pa/pa.md (short_jump): Remove define_insn.
13411         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
13412         (casesi0): Remove define_insn.
13413         (type): Remove btable_branch.
13414         (pa_combine_type): Likewise.
13415         (in_nullified_branch_delay): Likewise.
13416         (in_call_delay): Likewise.
13417         (define_delay): Likewise.
13418         (define_insn_reservation "Z3"): Likewise.
13419         (define_insn_reservation "Z4"): Likewise.
13420         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
13421         (pa_adjust_insn_length): Remove adjustment for btable branches.
13422         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
13423         and mno-big-switch
13425 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
13427         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
13428         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
13429         Add m->r,x alternatives.
13430         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
13431         splitters using SWI48x mode iterator.
13432         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
13433         TARGET_64BIT.  Add m->x alternative.
13434         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
13435         Add o->x alternative.  Enable for TARGET_SSE.
13436         (sse_storeq): Remove expander.
13437         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
13438         with memory input operand.
13439         (*vec_extractv2di_1 splitter): New.
13440         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
13441         * config/i386/i386.md (ssevecmodelower): New mode attribute.
13443 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
13445         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
13446         (INT_LOWPART): Delete.
13447         (extract_MB): Adjust.
13448         (extract_ME): Adjust.
13449         (print_operand): Adjust.
13451 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
13453         * config/rs6000/predicates.md (reg_or_add_cint_operand,
13454         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
13455         (reg_or_logical_cint_operand, easy_fp_constant,
13456         logical_const_operand): Delete "CONST_DOUBLE" case.
13457         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
13458         "HOST_BITS_PER_WIDE_INT == 64" test.
13459         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
13460         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
13461         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
13462         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
13463         test.
13464         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
13465         CONST_DOUBLE DImode/VOIDmode case.
13466         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
13467         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
13468         CONST_DOUBLE VOIDmode case.
13469         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
13470         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
13471         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
13472         Delete CONST_DOUBLE case.
13473         (splitters for mov FMOVE64 const_double): Delete
13474         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
13475         "HOST_BITS_PER_WIDE_INT >= 64" test.
13476         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
13477         case.
13478         (mov DI const_double): Delete.
13480 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
13482         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
13483         on op shows all bits zero in mode of a lowpart subreg, return zero.
13485 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
13487         PR target/57150
13488         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
13489         to save TFmode registers and DImode to save TImode registers for
13490         caller save operations.
13491         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
13492         mark being partially clobbered since they only use the first
13493         double word.
13495         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
13496         and TDmode only use the upper 64-bits of each VSX register.
13498 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13500         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
13501         (find_candidates_in_block): Disable slsr_process_phi.
13503 2013-05-03  Guozhi Wei  <carrot@google.com>
13505         * coverage.c (coverage_obj_init): Move the construction of gcov
13506         constructor to ...
13507         (build_init_ctor): ... here.
13509 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
13511         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
13512         (slsr_cand_d): Redefine def_phi.
13513         (stride_status, phi_adjust_status, count_phis_status): New enums.
13514         (find_phi_def): New.
13515         (find_basis_for_base_expr): New.
13516         (find_basis_for_candidate): Handle hidden bases.
13517         (alloc_cand_and_find_basis): Handle phi candidates.
13518         (slsr_process_phi): New.
13519         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
13520         (create_mul_imm_cand): Likewise.
13521         (create_add_ssa_cand): Exclude phi base candidates.
13522         (create_add_imm_cand): Likewise.
13523         (slsr_process_cast): Likewise.
13524         (slsr_process_copy): Likewise.
13525         (find_candidates_in_block): Handle phi candidates.
13526         (dump_candidate): Likewise.
13527         (unconditional_cands): Delete.
13528         (unconditional_cands_with_known_stride_p): Delete.
13529         (phi_dependent_cand_p): New.
13530         (cand_increment): Handle phi-dependent candidates.
13531         (replace_dependent): Delete.
13532         (replace_mult_candidate): New.
13533         (replace_unconditional_candidate): New.
13534         (incr_vec_index): Move to avoid forward reference.
13535         (create_add_on_incoming_edge): New.
13536         (create_phi_basis): New.
13537         (replace_dependents): Delete.
13538         (replace_conditional_candidate): New.
13539         (phi_add_costs): New.
13540         (replace_uncond_cands_and_profitable_phis): New.
13541         (record_increment): Handle phi adjustments.
13542         (record_phi_increments): New.
13543         (record_increments): Handle phi adjustments.
13544         (phi_incr_cost): New.
13545         (lowest_cost_path): Handle phis.
13546         (total_savings): Likewise.
13547         (analyze_increments): Likewise.
13548         (ncd_with_phi): New.
13549         (ncd_of_cand_and_phis): New.
13550         (nearest_common_dominator_for_cands): Handle phi increments.
13551         (all_phi_incrs_profitable): New.
13552         (replace_profitable_candidates): Handle phi-dependent candidates.
13553         (analyze_candidates_and_replace): Likewise.
13555 2013-05-03  Teresa Johnson  <tejohnson@google.com>
13557         PR bootstrap/57154
13558         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
13559         do not exceed REG_BR_PROB_BASE.
13561 2013-05-03  Jeff Law  <law@redhat.com>
13563         PR tree-optimization/57144
13564         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
13565         operand of the condition will bit into the new type when eliminating
13566         a cast feeding a condition.
13568 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
13570         PR rtl-optimization/57130
13571         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
13572         of COMPARE as in_code to the recursive call if needed.
13574 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
13576         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
13577         (enabled): Handle new members.
13578         * config/i386/sse.md (*vec_concatv2si): Merge from
13579         *vec_concatv2si_sse2 and vec_concatv2si_sse.
13580         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
13582 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
13584         PR tree-optimization/57027
13585         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
13586         for fnms opportunity, check we got the prerequisite kind
13587         of tree / gimple before using accessor functions.
13589 2013-05-03  Richard Biener  <rguenther@suse.de>
13591         * double-int.h (lshift): New overload without precision
13592         and arith argument.
13593         (operator *=, operator +=, operator -=): Move ...
13594         * double-int.c (operator *=, operator +=, operator -=): ... here
13595         and implement more efficiently.
13596         (mul_double_with_sign): Remove.
13597         (lshift_double): Adjust to take unsinged shift argument, push
13598         dispatching code to callers.
13599         (mul_double_wide_with_sign): Add early out for callers that
13600         are not interested in high parts or overflow.
13601         (lshift): New function.
13602         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
13603         dispatch code here.
13604         (lrotate, rrotate): Use logical shifts.
13605         * expr.c (get_inner_reference): Use lshift.
13606         * fixed-value.c (do_fixed_divide): Likewise.
13607         * tree-dfa.c (get_ref_base_and_extent): Likewise.
13608         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
13609         (indirect_refs_may_alias_p): Likewise.
13610         (stmt_kills_ref_p_1): Likewise.
13612 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
13614         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
13616 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
13618         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
13619         scalar form of FABD instruction.
13621 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
13623         * lra-constraints.c (process_alt_operands): Add checking alt
13624         number to choose the best alternative.
13626 2013-05-02  Richard Biener  <rguenther@suse.de>
13628         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
13629         bitmap and its handling.
13630         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
13632 2013-05-02  Richard Biener  <rguenther@suse.de>
13634         PR middle-end/57140
13635         * tree-inline.c (copy_loops): Properly handle removed loops.
13636         (copy_cfg_body): Mark destination loops for fixup if source
13637         loops needed fixup.
13639 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
13641         PR target/56732
13642         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
13643         generating simple_return for naked functions.
13645 2013-05-02  Martin Jambor  <mjambor@suse.cz>
13647         PR middle-end/56988
13648         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
13649         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
13650         flags match.
13651         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
13652         ipa_agg_replacement_value structures.
13653         (known_aggs_to_agg_replacement_list): Likewise.
13654         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
13655         (read_agg_replacement_chain): Likewise.
13656         (ipcp_transform_function): Also check that by_ref flags match.
13658 2013-05-02  Richard Biener  <rguenther@suse.de>
13660         * graphds.h (struct graph): Add obstack member.
13661         * graphds.c (new_graph): Initialize obstack and allocate
13662         vertices from it.
13663         (add_edge): Allocate edge from the obstack.
13664         (free_graph): Free the obstack instead of all edges and vertices.
13666 2013-05-02  Teresa Johnson  <tejohnson@google.com>
13668         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
13669         divides.
13670         * cfg.c (update_bb_profile_for_threading): Ditto.
13671         * tree-inline.c (copy_bb): Ditto.
13672         (copy_edges_for_bb): Ditto.
13673         (initialize_cfun): Ditto.
13674         (copy_cfg_body): Ditto.
13675         (expand_call_inline): Ditto.
13676         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
13677         (estimate_node_size_and_time): Ditto.
13678         (inline_merge_summary): Ditto.
13679         * cgraphclones.c (cgraph_clone_edge): Ditto.
13680         (cgraph_clone_node): Ditto.
13681         * sched-rgn.c (compute_dom_prob_ps): Ditto.
13682         (compute_trg_info): Ditto.
13684 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
13686         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
13687         S reg when fp attribute set.
13688         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
13690 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
13692         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
13693         New pattern.
13694         (*and_one_cmplsi3_compare0_uxtw): Likewise.
13695         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
13696         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
13698 2013-05-02  Richard Biener  <rguenther@suse.de>
13700         * tree-scalar-evolution.c (scev_info_hasher): Remove.
13701         (struct instantiate_cache_entry): New type.
13702         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
13703         (struct instantiate_cache_type): New type.
13704         (set_instantiated_value, get_instantiated_value): Remove.
13705         (get_instantiated_value_entry): New function.
13706         (instantiate_scev_name): Use the new cache and adjust.
13707         (instantiate_scev_poly): Adjust.
13708         (instantiate_scev_binary): Likewise.
13709         (instantiate_array_ref): Likewise.
13710         (instantiate_scev_convert): Likewise.
13711         (instantiate_scev_not): Likewise.
13712         (instantiate_scev_3): Likewise.
13713         (instantiate_scev_2): Likewise.
13714         (instantiate_scev_r): Likewise.
13715         (instantiate_scev): Likewise.
13716         (resolve_mixers): Likewise.
13718 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
13720         PR target/57091
13721         * lra-constraints.c (best_small_class_operands_num): Remove.
13722         (process_alt_operands): Remove small_class_operands_num.  Take
13723         small classes operands into losers and only if the operand is not
13724         matched.  Modify debugging output.
13725         (curr_insn_transform): Remove best_small_class_operands_num.
13726         Print insn name.
13728 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13730         * config/aarch64/aarch64-builtins.c
13731         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
13732         * config/aarch64/aarch64-simd-builtins.def
13733         (reduc_splus_): Add new modes.
13734         (reduc_uplus_): New.
13735         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
13736         (reduc_uplus_v4sf): Likewise.
13737         (reduc_splus_v4sf): Likewise.
13738         (aarch64_addv<mode>): Likewise.
13739         (reduc_uplus_<mode>): Likewise.
13740         (reduc_splus_<mode>): Likewise.
13741         (aarch64_addvv2di): Likewise.
13742         (reduc_uplus_v2di): Likewise.
13743         (reduc_splus_v2di): Likewise.
13744         (aarch64_addvv2si): Likewise.
13745         (reduc_uplus_v2si): Likewise.
13746         (reduc_splus_v2si): Likewise.
13747         (reduc_<sur>plus_<mode>): New.
13748         (reduc_<sur>plus_v2di): Likewise.
13749         (reduc_<sur>plus_v2si): Likewise.
13750         (reduc_<sur>plus_v4sf): Likewise.
13751         (aarch64_addpv4sf): Likewise.
13752         * config/aarch64/arm_neon.h
13753         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
13754         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
13755         add UNSPEC_SADDV, UNSPEC_UADDV.
13756         (SUADDV): New.
13757         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
13759 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13761         * config/aarch64/arm_neon.h
13762         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
13764 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13766         * config/aarch64/aarch64-builtins
13767         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
13769 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13771         * config/aarch64/aarch64-simd-builtins.def
13772         (reduc_smax_): New.
13773         (reduc_smin_): Likewise.
13774         (reduc_umax_): Likewise.
13775         (reduc_umin_): Likewise.
13776         (reduc_smax_nan_): Likewise.
13777         (reduc_smin_nan_): Likewise.
13778         (fmax): Remove.
13779         (fmin): Likewise.
13780         (smax): Update for V2SF, V4SF and V2DF modes.
13781         (smin): Likewise.
13782         (smax_nan): New.
13783         (smin_nan): Likewise.
13784         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
13785         (<su><maxmin><mode>3): ...This, refactor.
13786         (s<maxmin><mode>3): New.
13787         (<maxmin_uns><mode>3): Likewise.
13788         (reduc_<maxmin_uns>_<mode>): Refactor.
13789         (reduc_<maxmin_uns>_v4sf): Likewise.
13790         (reduc_<maxmin_uns>_v2si): Likewise.
13791         (aarch64_<fmaxmin><mode>: Remove.
13792         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
13793         new builtin names.
13794         (vmin<q>_f<32,64>): Likewise.
13795         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
13796         (FMAXMIN): New.
13797         (su): Add mappings for smax, smin, umax, umin.
13798         (maxmin): New.
13799         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
13800         (FMAXMIN): Rename as...
13801         (FMAXMIN_UNS): ...This.
13802         (maxminv): Remove.
13803         (fmaxminv): Likewise.
13804         (fmaxmin): Likewise.
13805         (maxmin_uns): New.
13806         (maxmin_uns_op): Likewise.
13808 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13810         * config/aarch64/arm_neon.h
13811         (vac<ge, gt><sd>_f<32, 64>): Rename to...
13812         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
13813         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
13815 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13817         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
13818         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
13820 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13822         * config/aarch64/aarch64-simd.md
13823         (vcond<mode>_internal): Handle special cases for constant masks.
13824         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
13825         (vcondu<mode><mode>): Likewise.
13826         (vcond<v_cmp_result><mode>): New.
13828 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13830         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
13831         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
13832         * config/aarch64/aarch64-simd-builtins.def
13833         (cmeq): Update to BUILTIN_VALLDI.
13834         (cmgt): Likewise.
13835         (cmge): Likewise.
13836         (cmle): Likewise.
13837         (cmlt): Likewise.
13838         * config/aarch64/arm_neon.h
13839         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
13840         to builtins or C as appropriate.
13842 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
13844         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
13845         (cmgeu): ...This.
13846         (cmhi): Rename to...
13847         (cmgtu): ...This.
13848         * config/aarch64/aarch64-simd.md
13849         (simd_mode): Add SF.
13850         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
13851         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
13852         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
13853         (cstore<mode>_neg): ...This.
13854         * config/aarch64/iterators.md
13855         (VALLF): new.
13856         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
13857         (COMPARISONS): New.
13858         (UCOMPARISONS): Likewise.
13859         (optab): Add missing comparisons.
13860         (n_optab): New.
13861         (cmp_1): Likewise.
13862         (cmp_2): Likewise.
13863         (CMP): Likewise.
13864         (cmp): Remove.
13865         (VCMP_S): Likewise.
13866         (VCMP_U): Likewise.
13867         (V_cmp_result): Add DF, SF modes.
13868         (v_cmp_result): Likewise.
13869         (v): Likewise.
13870         (vmtype): Likewise.
13871         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
13873 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
13875         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
13876         define_insn to define_insn_and_split.
13877         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
13878         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
13879         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
13880         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
13881         (thumb2_negscc): Likewise.
13883 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
13885         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
13887 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
13889         * config/arm/thumb2.md: Remove trailing whitespaces.
13891 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
13893         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
13894         Use gen_int_mode rather than GEN_INT.
13896 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
13898         * value-prof.c (stream_in_histogram_value): Remove the strayed
13899         debug_gimple_stmt.
13901 2013-04-30  Richard Biener  <rguenther@suse.de>
13903         PR middle-end/57122
13904         * cfghooks.c (split_edge): Properly check for the loop latch edge.
13906 2013-04-30  Richard Biener  <rguenther@suse.de>
13908         PR middle-end/57107
13909         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
13911 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
13913         PR rtl-optimization/56957
13914         PR rtl-optimization/57105
13915         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
13916         variable.  Use just INSN_UID for determining whether an insn
13917         should be only disconnected from the insn stream.
13918         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
13920 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
13922         PR tree-optimization/57104
13923         * tsan.c (instrument_expr): Don't instrument accesses to
13924         DECL_HARD_REGISTER VAR_DECLs.
13926 2013-04-30  Richard Biener  <rguenther@suse.de>
13928         * function.h (loops_for_fn): New inline function.
13929         (set_loops_for_fn): Likewise.
13930         * cfgloop.h (place_new_loop): Add struct function parameter.
13931         (get_loop): Likewise.
13932         (get_loops): Likewise.
13933         (number_of_loops): Likewise.
13934         (fel_next): Adjust.
13935         (fel_init): Likewise.
13936         * cfg.c (get_loop_copy): Adjust.
13937         * cfgloop.c (flow_loops_dump): Likewise.
13938         (record_loop_exits): Likewise.
13939         (verify_loop_structure): Likewise.
13940         * cfgloopanal.c (mark_irreducible_loops): Likewise.
13941         (estimate_reg_pressure_cost): Likewise.
13942         (mark_loop_exit_edges): Likewise.
13943         * cfgloopmanip.c (place_new_loop): Likewise.
13944         (add_loop): Likewise.
13945         (duplicate_loop): Likewise.
13946         * graph.c (draw_cfg_nodes): Likewise.
13947         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
13948         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
13949         (extract_affine_chrec): Likewise.
13950         (build_scop_iteration_domain): Likewise.
13951         * graphite.c (graphite_initialize): Likewise.
13952         * ira-build.c (create_loop_tree_nodes): Likewise.
13953         (more_one_region_p): Likewise.
13954         (rebuild_regno_allocno_maps): Likewise.
13955         (mark_loops_for_removal): Likewise.
13956         (mark_all_loops_for_removal): Likewise.
13957         (remove_unnecessary_regions): Likewise.
13958         (ira_build): Likewise.
13959         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
13960         * loop-init.c (fix_loop_structure): Likewise.
13961         (gate_rtl_move_loop_invariants): Likewise.
13962         (gate_rtl_unswitch): Likewise.
13963         (gate_rtl_unroll_and_peel_loops): Likewise.
13964         (rtl_doloop): Likewise.
13965         * lto-streamer-in.c (input_cfg): Likewise.
13966         * lto-streamer-out.c (output_cfg): Likewise.
13967         * modulo-sched.c (sms_schedule): Likewise.
13968         * predict.c (tree_estimate_probability): Likewise.
13969         (tree_estimate_probability_driver): Likewise.
13970         (estimate_loops): Likewise.
13971         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
13972         (move_sese_region_to_fn): Likewise.
13973         (debug_loop_num): Likewise.
13974         * tree-chrec.c (chrec_evaluate): Likewise.
13975         (hide_evolution_in_other_loops_than_loop): Likewise.
13976         (chrec_component_in_loop_num): Likewise.
13977         (reset_evolution_in_loop): Likewise.
13978         (evolution_function_is_invariant_rec_p): Likewise.
13979         * tree-if-conv.c (main_tree_if_conversion): Likewise.
13980         * tree-inline.c (copy_loops): Likewise.
13981         (copy_cfg_body): Likewise.
13982         (tree_function_versioning): Likewise.
13983         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
13984         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
13985         Likewise.
13986         (add_to_evolution_1): Likewise.
13987         (scev_const_prop): Likewise.
13988         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
13989         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
13990         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
13991         (tree_ssa_lim_initialize): Likewise.
13992         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
13993         (verify_loop_closed_ssa): Likewise.
13994         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
13995         (tree_ssa_loop_im): Likewise.
13996         (tree_ssa_loop_unswitch): Likewise.
13997         (tree_vectorize): Likewise.
13998         (check_data_deps): Likewise.
13999         (tree_ssa_loop_ivcanon): Likewise.
14000         (tree_ssa_loop_bounds): Likewise.
14001         (tree_complete_unroll): Likewise.
14002         (tree_complete_unroll_inner): Likewise.
14003         (tree_parallelize_loops): Likewise.
14004         (tree_ssa_loop_prefetch): Likewise.
14005         (tree_ssa_loop_ivopts): Likewise.
14006         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
14007         * tree-vectorizer.c (vectorize_loops): Likewise.
14009 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
14011         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
14012         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
14013         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
14014         with EABI_LINK_SPEC.
14016 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
14018         PR target/44578
14019         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
14020         alternative.
14022 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
14024         PR target/57097
14025         * lra-constraints.c (process_alt_operands): Discourage a bit more
14026         using memory for pseudos.  Print cost dump for alternatives.
14027         Modify cost values for conflicts with early clobbers.
14028         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
14030 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
14032         PR target/57098
14033         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
14035 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
14037         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
14038         from/to S register.
14039         (movdi_aarch64): Support LDR/STR from/to D register.
14041 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
14043         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
14044         or higher by default.
14046 2013-04-29  Richard Biener  <rguenther@suse.de>
14048         PR middle-end/57075
14049         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
14050         even if not adding abnormal edges for calls that can make
14051         abnormal gotos.
14053 2013-04-29  Richard Biener  <rguenther@suse.de>
14055         PR middle-end/57103
14056         * tree-cfg.c (move_stmt_op): Fix condition under which to update
14057         TREE_BLOCK.
14058         (move_stmt_r): Remove redundant checking.
14060 2013-04-29  Teresa Johnson  <tejohnson@google.com>
14062         PR bootstrap/57077
14063         * basic-block.h (apply_scale): New function.
14064         (apply_probability): Use apply_scale.
14065         * gimple-streamer-in.c (input_bb): Ditto.
14066         * lto-streamer-in.c (input_cfg): Ditto.
14067         * lto-cgraph.c (merge_profile_summaries): Ditto.
14068         * tree-optimize.c (execute_fixup_cfg): Ditto.
14069         * tree-inline.c (copy_bb): Update comment to use apply_scale.
14070         (copy_edges_for_bb): Ditto.
14071         (copy_cfg_body): Ditto.
14073 2013-04-29  Tom de Vries  <tom@codesourcery.com>
14075         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
14076         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
14077         (tail_merge_optimize): Handle current_loops == NULL.
14079 2013-04-26  Jeff Law  <law@redhat.com>
14081         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
14082         (simplify_cond_using_ranges): Generalize code to simplify
14083         COND_EXPRs where one argument is a constant and the other
14084         is an SSA_NAME created by an integral type conversion.
14086 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14088         * config/arm/arm.md (store_minmaxsi): Use only when
14089         optimize_insn_for_size_p.
14091 2013-04-29  Christian Bruel  <christian.bruel@st.com>
14093         PR target/57108
14094         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
14096 2013-04-29  Richard Biener  <rguenther@suse.de>
14098         PR middle-end/57089
14099         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
14100         loop tree make sure to schedule a fixup for the child as well.
14101         (expand_omp_for_generic): Properly add loops.
14102         (expand_omp_for_static_nochunk): Likewise.
14103         (expand_omp_for_static_chunk): Likewise.
14104         (expand_omp_for): For the degenerate case fixup loops.
14105         (expand_omp_sections): Fix default bb placement in loops.
14106         (expand_omp_atomic_pipeline): Properly add loops.
14108 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14110         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
14112 2013-04-29  Tom de Vries  <tom@codesourcery.com>
14114         * tree-ssa-tail-merge.c: Update header comment.
14116 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14118         * config/aarch64/arm_neon.h
14119         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
14120         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
14121         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
14122         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
14123         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
14124         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
14125         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
14126         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
14128 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14130         * config/aarch64/aarch64-simd.md
14131         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
14132         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
14133         fix_trunc, fixuns_trunc.
14134         (ftrunc<VDQF:mode>2): New.
14135         * config/aarch64/iterators.md (optab): Add fix, fixuns.
14136         (fix_trunc_optab): New.
14138 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14140         * config/aarch64/aarch64-builtins.c
14141         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
14142         iceilf, lround, iroundf.
14144 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
14146         PR target/54349
14147         * config/i386/i386.h (enum ix86_tune_indices)
14148         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
14149         New, split from X86_TUNE_INTER_UNIT_MOVES.
14150         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
14151         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
14152         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
14153         (TARGET_INTER_UNIT_MOVES): Remove.
14154         * config/i386/i386.c (initial_ix86_tune_features): Update.
14155         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
14156         (ix86_expand_convert_uns_didf_sse): Use
14157         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
14158         (ix86_expand_vector_init_one_nonzero): Ditto.
14159         (ix86_expand_vector_init_interleave): Ditto.
14160         (inline_secondary_memory_needed): Return true for moves from SSE class
14161         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
14162         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
14163         * config/i386/constraints.md (Yi, Ym): Depend on
14164         TARGET_INTER_UNIT_MOVES_TO_VEC.
14165         (Yj, Yn): New constraints.
14166         * config/i386/i386.md (*movdi_internal): Change constraints of
14167         operand 1 from Yi to Yj and from Ym to Yn.
14168         (*movsi_internal): Ditto.
14169         (*movdf_internal): Ditto.
14170         (*movsf_internal): Ditto.
14171         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
14172         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
14173         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
14174         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
14175         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
14176         * config/i386/sse.md (movdi_to_sse): Ditto.
14177         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
14178         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
14179         TARGET_INTER_UNIT_MOVES.
14180         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
14181         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
14182         instead of TARGET_INTER_UNIT_MOVES.
14183         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
14184         operand 1 from Yi to Yj and from Ym to Yn.
14186 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14188         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
14189         (float_truncate_hi_): Likewise.
14190         (float_extend_lo_): Likewise.
14191         (float_truncate_lo_): Likewise.
14192         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
14193         (aarch64_float_extend_lo_v2df): Likewise.
14194         (vec_unpacks_hi_v4sf): Likewise.
14195         (aarch64_float_truncate_lo_v2sf): Likewise.
14196         (aarch64_float_truncate_hi_v4sf): Likewise.
14197         (vec_pack_trunc_v2df): Likewise.
14198         (vec_pack_trunc_df): Likewise.
14200 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14202         * config/aarch64/aarch64-builtins.c
14203         (aarch64_fold_builtin): Fold float conversions.
14204         * config/aarch64/aarch64-simd-builtins.def
14205         (floatv2si, floatv4si, floatv2di): New.
14206         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
14207         * config/aarch64/aarch64-simd.md
14208         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
14209         * config/aarch64/iterators.md (FLOATUORS): New.
14210         (optab): Add float, floatuns.
14211         (su_optab): Likewise.
14213 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14215         * config/aarch64/aarch64-builtins.c
14216         (aarch64_builtin_vectorized_function): Use new names for
14217         fcvt builtins.
14218         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
14219         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
14220         (fcvtzu): Split as...
14221         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
14222         (fcvtas): Split as...
14223         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
14224         (fcvtau): Split as...
14225         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
14226         (fcvtps): Split as...
14227         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
14228         (fcvtpu): Split as...
14229         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
14230         (fcvtms): Split as...
14231         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
14232         (fcvtmu): Split as...
14233         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
14234         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
14235         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
14236         (lfrintnusf, lfrintnudf): Likewise.
14237         * config/aarch64/aarch64-simd.md
14238         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
14239         define_insn.
14240         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
14241         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
14242         (fcvt_pattern): Likewise.
14244 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14246         * config/aarch64/aarch64-simd.md
14247         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
14248         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
14250 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14252         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
14253         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
14254         (vrnd<a,m,n,p>_f32): Implement using builtins.
14255         (vrnd<i,x><q>_f<32, 64>): New.
14257 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
14259         * config/aarch64/aarch64-builtins.c
14260         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
14261         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
14262         (frintz): Rename to...
14263         (btrunc): ...this.
14264         (frintp): Rename to...
14265         (ceil): ...this.
14266         (frintm): Rename to...
14267         (floor): ...this.
14268         (frinti): Rename to...
14269         (nearbyint): ...this.
14270         (frintx): Rename to...
14271         (rint): ...this.
14272         (frinta): Rename to...
14273         (round): ...this.
14274         * config/aarch64/aarch64-simd.md
14275         (aarch64_frint<frint_suffix><mode>): Delete.
14276         (<frint_pattern><mode>2): Convert to insn.
14277         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
14278         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
14279         (frint_pattern): Likewise.
14280         (frint_suffix): Likewise.
14282 2013-04-29  Richard Biener  <rguenther@suse.de>
14284         PR tree-optimization/57081
14285         * loop-init.c: Include tree-flow.h.
14286         (loop_optimizer_finalize): Free number of iteration estimates.
14287         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
14289 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
14291         PR tree-optimization/57083
14292         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
14293         non-singleton shift count range, zero extend low_bound for uns case.
14295         * config/i386/predicates.md (general_vector_operand): New predicate.
14296         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
14297         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
14298         if they aren't nonimmediate operands.  If their original values
14299         satisfy const_vector_equal_evenodd_p, don't shift them.
14300         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
14301         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
14302         if not nonimmediate_operand.
14303         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
14304         instead of register_operand.
14305         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
14307 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
14309         * stor-layout.c (finalize_size_functions): Allocate a structure and
14310         reset cfun before dumping the functions.
14312 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
14314         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
14316         PR target/56866
14317         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
14318         use xop_pmacsdqh if uns_p.
14319         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
14320         the immediate rotate count.
14322 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
14324         * rtl.h (struct rtx_def): Add comment for field jump.
14325         (LRA_SUBREG_P): New macro.
14326         * recog.c (register_operand): Check LRA_SUBREG_P.
14327         * lra.c (lra): Add note at the end of RTL code. Align non-empty
14328         stack frame.
14329         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
14330         (lra_final_code_change): Skip subreg change for operators.
14331         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
14332         if there are no operand changes.
14333         * lra-constraints.c (curr_insn_set): New.
14334         (match_reload): Set LRA_SUBREG_P.
14335         (emit_spill_move): Ditto.
14336         (check_and_process_move): Use curr_insn_set. Process only single
14337         set insns.  Don't initialize sec_mem_p and change_p.
14338         (simplify_operand_subreg): Use LRA_SUBREG_P.
14339         (reg_in_class_p): New function.
14340         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
14341         of #ifdef.  Add code to remove cycling.
14342         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
14343         non-null disp.  Reload inner instead of disp when base and index
14344         are null.  Try to put lo_sum into register.
14345         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
14346         (check_and_process_move): Move code for move cost check to
14347         simple_move_p.  Remove equiv_substitution.
14348         (simple_move_p): New function.
14349         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
14350         curr_insn_set.  Call check_and_process_move only for single set
14351         insns.  Use the new function.  Move call of check_and_process_move
14352         after operand equiv substitution and address process.
14354 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
14356         PR go/57045
14357         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
14358         with nonlocal goto receivers or returns twice calls, ignore
14359         unininitialized values from abnormal edges to nl goto receiver
14360         or returns twice call.
14362 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
14364         PR tree-optimization/57051
14365         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
14366         and VEC_RSHIFT_EXPR if shift count is a multiple of element
14367         bitsize.
14369 2013-04-26  Richard Biener  <rguenther@suse.de>
14371         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
14372         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
14373         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
14374         (fixup_loop_arrays_after_move): New function.
14375         (move_sese_region_to_fn): Properly outline the loop tree parts
14376         of the SESE region.
14378 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
14380         * config/i386/i386.md (type, unit): Fix long lines.
14382 2013-04-26  Richard Biener  <rguenther@suse.de>
14384         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
14385         (lto-streamer-out.o): Likewise.
14386         * cfgloop.c (init_loops_structure): Export, add struct function
14387         argument and adjust.
14388         (flow_loops_find): Adjust.
14389         * cfgloop.h (enum loop_estimation): Add EST_LAST.
14390         (init_loops_structure): Declare.
14391         * lto-streamer-in.c: Include cfgloop.h.
14392         (input_cfg): Input the loop tree.
14393         * lto-streamer-out.c: Include cfgloop.h.
14394         (output_cfg): Output the loop tree.
14395         (output_struct_function_base): Do not drop PROP_loops.
14397 2013-03-26  Richard Biener  <rguenther@suse.de>
14399         * tree-cfg.c (execute_build_cfg): Build the loop tree.
14400         (pass_build_cfg): Provide PROP_loops.
14401         (move_sese_region_to_fn): Remove loops that are outlined into fn
14402         for now.
14403         * tree-inline.c: Include cfgloop.h.
14404         (initialize_cfun): Do not drop PROP_loops.
14405         (copy_loops): New function.
14406         (copy_cfg_body): Copy loop structure.
14407         (tree_function_versioning): Initialize destination loop tree.
14408         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
14409         (pass_parallelize_loops): Do IL verification.
14410         * loop-init.c (loop_optimizer_init): Fixup loops if required.
14411         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
14412         the CFG make sure we fixup loops as well.
14413         * tree-ssa-tail-merge.c: Include cfgloop.h.
14414         (replace_block_by): When merging loop latches mark loops for fixup.
14415         * lto-streamer-out.c (output_struct_function_base): Drop
14416         PROP_loops for now.
14417         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
14418         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
14419         * ipa-split.c: Include cfgloop.h.
14420         (split_function): Add the new return block to the loop tree root.
14421         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
14422         whether we have removed the forwarder block.
14423         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
14424         * cfgloop.h (place_new_loop): Declare.
14425         * cfgloopmanip.c (place_new_loop): Export.
14426         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
14427         (tree-switch-conversion.o): Likewise.
14428         (tree-complex.o): Likewise.
14429         (tree-inline.o): Likewise.
14430         (tree-ssa-tailmerge.o): Likewise.
14431         (ipa-split.o): Likewise.
14432         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
14433         (tree-ssa-copy.o): Likewise.
14434         * tree-switch-conversion.c: Include cfgloop.h
14435         (process_switch): If we emit a bit-test cascade, schedule loops
14436         for fixup.
14437         * tree-complex.c: Include cfgloop.h.
14438         (expand_complex_div_wide): Properly add new basic-blocks to loops.
14439         * asan.c: Include cfgloop.h.
14440         (create_cond_insert_point): Properly add new basic-blocks to
14441         loops, schedule loop fixup.
14442         * cfgloop.c (verify_loop_structure): Check that looks are not
14443         marked for fixup.
14444         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
14445         to loops.
14446         (expand_omp_for_generic): Likewise.
14447         (expand_omp_sections): Likewise.
14448         (expand_omp_atomic_pipeline): Schedule loops for fixup.
14449         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
14450         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
14451         is initialized, not when loops are present.
14452         * tree-parloops.c (parallelize_loops): Remove checking here.
14453         * passes.c (init_optimization_passes): Schedule a copy-propagation
14454         pass before complete unrolling of inner loops.
14456 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
14458         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
14459         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
14460         (colorize_init): Add argument to _WIN32 version.
14461         * toplev.c: Include diagnostic-color.h.
14462         (process_options): Default to -fdiagnostics-color=auto if
14463         GCC_COLORS env var is in the environment.
14464         * common.opt (fdiagnostics-color=): Add Var and Init.
14465         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
14466         env var is in the environment, the default is auto rather than never.
14468         * diagnostic.h (file_name_as_prefix): Add context argument.
14469         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
14470         the string as locus.
14471         * langhooks.c (lhd_print_error_function): Adjust caller.
14473 2013-04-25  Lawrence Crowl  <crowl@google.com>
14475         * var-tracking.c (shared_hash_def::htab):
14476         Change type to hash_table.  Update dependent calls and types.
14478 2013-04-25  Lawrence Crowl  <crowl@google.com>
14480         * Makefile.in: Update as needed below.
14482         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
14483         Move declaration to after the type's method definitons.
14485         * attribs.c (htab_t scoped_attributes::attribute_hash):
14486         Change type to hash_table.  Update dependent calls and types.
14488         * bitmap.c (htab_t bitmap_desc_hash):
14489         Change type to hash_table.  Update dependent calls and types.
14491         * cselib.c (htab_t cselib_hash_table):
14492         Change type to hash_table.  Update dependent calls and types.
14494         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
14495         (hash_string_slot_node): Move implementation into lto-streamer.h
14496         struct string_slot_hasher.
14497         (eq_string_slot_node): Likewise.
14499         * data-streamer-out.c: Update output_block::string_hash_table
14500         dependent calls and types.
14502         * dwarf2cfi.c (htab_t trace_index):
14503         Change type to hash_table.  Update dependent calls and types.
14505         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
14506         Change type to hash_table.  Update dependent calls and types.
14507         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
14508         (htab_t optimize_external_refs::map): Likewise.
14509         (htab_t output_comp_unit::extern_map): Likewise.
14510         (htab_t output_comdat_type_unit::extern_map): Likewise.
14511         (htab_t output_macinfo::macinfo_htab): Likewise.
14512         (htab_t optimize_location_lists::htab): Likewise.
14513         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
14515         * except.c (htab_t ehspec_hash_type):
14516         Change type to hash_table.  Update dependent calls and types.
14517         (assign_filter_values::ttypes): Likewise.
14518         (assign_filter_values::ehspec): Likewise.
14519         (sjlj_assign_call_site_values::ar_hash): Likewise.
14520         (convert_to_eh_region_ranges::ar_hash): Likewise.
14522         * gcse.c (htab_t pre_ldst_table):
14523         Change type to hash_table.  Update dependent calls and types.
14525         * ggc-common.c (htab_t saving_htab):
14526         Change type to hash_table.  Update dependent calls and types.
14527         (htab_t loc_hash): Likewise.
14528         (htab_t ptr_hash): Likewise.
14529         (call_count): Rename ggc_call_count.
14530         (call_alloc): Rename ggc_call_alloc.
14531         (loc_descriptor): Rename make_loc_descriptor.
14532         (add_statistics): Rename ggc_add_statistics.
14534         * ggc-common.c (saving_htab):
14535         Change type to hash_table.  Update dependent calls and types.
14537         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
14538         (push_gimplify_context): Likewise.
14539         (pop_gimplify_context): Likewise.
14540         (struct gimple_temp_hash_elt): Added.
14541         (struct gimplify_hasher): Likewise.
14542         (struct gimplify_ctx.temp_htab):
14543         Change type to hash_table.  Update dependent calls and types.
14545         * gimple-fold.c: Include gimplify-ctx.h.
14547         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
14548         Change type to hash_table.  Update dependent calls and types.
14549         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
14550         avoid potential global name collision.
14552         * gimplify.c: Include gimplify-ctx.h.
14553         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
14554         (htab_t gimplify_ctx::temp_htab):
14555         Update dependent calls and types for new type hash_table.
14556         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
14557         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
14559         * gimplify-ctx.h: New.
14560         (struct gimple_temp_hash_elt): Move from gimplify.c.
14561         (class gimplify_hasher): New.
14562         (struct gimplify_ctx): Move from gimple.h.
14563         (htab_t gimplify_ctx::temp_htab):
14564         Change type to hash_table.  Update dependent calls and types.
14566         * graphite-clast-to-gimple.c: Include graphite-htab.h.
14567         (htab_t ivs_params::newivs_index):
14568         Change type to hash_table.  Update dependent calls and types.
14569         (htab_t ivs_params::params_index): Likewise.
14570         (htab_t print_generated_program::params_index): Likewise.
14571         (htab_t gloog::newivs_index): Likewise.
14572         (htab_t gloog::params_index): Likewise.
14574         * graphite.c: Include graphite-htab.h.
14575         4htab_t graphite_transform_loops::bb_pbb_mapping):
14576         Change type to hash_table.  Update dependent calls and types.
14578         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
14579         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
14580         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
14582         * graphite-dependences.c: Include graphite-htab.h.
14583         (loop_is_parallel_p): Change hash table type of parameter.
14585         * graphite-htab.h: New.
14586         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
14587         (extern find_pbb_via_hash): Move from graphite-poly.h.
14588         (extern loop_is_parallel_p): Move from graphite-poly.h.
14589         (extern get_loop_body_pbbs): Move from graphite-poly.h.
14591         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
14592         (extern loop_is_parallel_p): Move to graphite-htab.h.
14593         (extern get_loop_body_pbbs): Move to graphite-htab.h.
14595         * haifa-sched.c (htab_t delay_htab):
14596         Change type to hash_table.  Update dependent calls and types.
14597         (htab_t delay_htab_i2): Likewise.
14599         * ira-color.c (htab_t allocno_hard_regs_htab):
14600         Change type to hash_table.  Update dependent calls and types.
14602         * ira-costs.c (htab_t cost_classes_htab):
14603         Change type to hash_table.  Update dependent calls and types.
14605         * loop-invariant.c (htab_t merge_identical_invariants::eq):
14606         Change type to hash_table.  Update dependent calls and types.
14608         * loop-iv.c (htab_t bivs):
14609         Change type to hash_table.  Update dependent calls and types.
14611         * loop-unroll.c (htab_t opt_info::insns_to_split):
14612         Change type to hash_table.  Update dependent calls and types.
14613         (htab_t opt_info::insns_with_var_to_expand): Likewise.
14615         * lto-streamer.h (struct string_slot): Move from data-streamer.h
14616         (struct string_slot_hasher): New.
14617         (htab_t output_block::string_hash_table):
14618         Change type to hash_table.  Update dependent calls and types.
14620         * lto-streamer-in.c (freeing_string_slot_hasher): New.
14621         (htab_t file_name_hash_table):
14622         Change type to hash_table.  Update dependent calls and types.
14624         * lto-streamer-out.c: Update output_block::string_hash_table dependent
14625         calls and types.
14627         * lto-streamer.c (htab_t tree_htab):
14628         Change type to hash_table.  Update dependent calls and types.
14630         * omp-low.c: Include gimplify-ctx.h.
14632         * passes.c (htab_t name_to_pass_map):
14633         Change type to hash_table.  Update dependent calls and types.
14634         (pass_traverse): Rename to passes_pass_traverse.
14636         * plugin.c (htab_t event_tab):
14637         Change type to hash_table.  Update dependent calls and types.
14639         * postreload-gcse.c (htab_t expr_table):
14640         Change type to hash_table.  Update dependent calls and types.
14641         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
14643         * sese.c (debug_rename_map_1): Make extern.
14644         (htab_t copy_bb_and_scalar_dependences::rename_map):
14645         Change type to hash_table.  Update dependent calls and types.
14647         * sese.h (extern debug_rename_map): Move to .c file.
14649         * store-motion.c (htab_t store_motion_mems_table):
14650         Change type to hash_table.  Update dependent calls and types.
14652         * trans-mem.c (htab_t tm_new_mem_hash):
14653         Change type to hash_table.  Update dependent calls and types.
14655         * tree-browser.c (htab_t TB_up_ht):
14656         Change type to hash_table.  Update dependent calls and types.
14658         * tree-cfg.c (htab_t discriminator_per_locus):
14659         Change type to hash_table.  Update dependent calls and types.
14661         * tree-complex.c: Include tree-hasher.h
14662         (htab_t complex_variable_components):
14663         Change type to hash_table.  Update dependent calls and types.
14665         * tree-eh.c (htab_t finally_tree):
14666         Change type to hash_table.  Update dependent calls and types.
14668         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
14669         struct int_tree_hasher.
14670         (extern int_tree_map_eq): Likewise.
14671         (uid_decl_map_hash): Removed.
14672         (extern decl_tree_map_eq): Likewise.
14674         * tree-hasher.h: New.
14675         (struct int_tree_hasher): New.
14676         (typedef int_tree_htab_type): New.
14678         * tree-inline.c: Include gimplify-ctx.h.
14680         * tree-mudflap.c: Include gimplify-ctx.h.
14682         * tree-parloops.c: Include tree-hasher.h.
14683         (htab_t eliminate_local_variables_stmt::decl_address):
14684         Change type to hash_table.  Update dependent calls and types.
14685         (htab_t separate_decls_in_region::decl_copies): Likewise.
14687         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
14688         Change type to hash_table.  Update dependent calls and types.
14690         * tree-sra.c (candidates):
14691         Change type to hash_table.  Update dependent calls and types.
14693         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
14694         in tree-flow.h.
14695         (int_tree_map_hash): Likewise.
14697         * tree-ssa-dom.c (htab_t avail_exprs):
14698         Change type to hash_table.  Update dependent calls and types.
14700         * tree-ssa-live.c (var_map_base_init::tree_to_index):
14701         Change type to hash_table.  Update dependent calls and types.
14703         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
14704         Change type to hash_table.  Update dependent calls and types.
14706         * tree-ssa-phiopt.c (seen_ssa_names):
14707         Change type to hash_table.  Update dependent calls and types.
14709         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
14710         Change type to hash_table.  Update dependent calls and types.
14712         * tree-ssa-uncprop.c (equiv):
14713         Change type to hash_table.  Update dependent calls and types.
14715 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
14717         PR rtl-optimization/57003
14718         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
14719         call note_stores with kill_clobbered_value callback again after
14720         killing regs_invalidated_by_call.
14722 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
14724         * config/aarch64/aarch64-simd.md
14725         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
14726         (aarch64_simd_bsl<mode>): Likewise.
14727         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
14729 2013-04-25  Marek Polacek  <polacek@redhat.com>
14731         PR tree-optimization/57066
14732         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
14734 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
14736         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
14738 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
14740         * config/aarch64/aarch64-builtins.c
14741         (aarch64_fold_builtin): New.
14742         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
14743         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
14744         * config/aarch64/aarch64-simd-builtins.def (abs): New.
14745         * config/aarch64/arm_neon.h
14746         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
14748 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
14749             Tejas Belagod  <tejas.belagod@arm.com>
14751         * config/aarch64/aarch64-builtins.c
14752         (aarch64_gimple_fold_builtin): New.
14753         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
14754         * config/aarch64/aarch64-simd-builtins.def (addv): New.
14755         * config/aarch64/aarch64-simd.md (addpv4sf): New.
14756         (addvv4sf): Update.
14757         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
14759 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14761         * config/aarch64/aarch64.md
14762         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
14764 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14766         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
14767         (*ngcsi_uxtw): New pattern.
14769 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14770             Julian Brown  <julian@codesourcery.com>
14772         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
14773         (TB_DREG): Add T_V4HF.
14774         (v4hf_UP): New macro.
14775         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
14776         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
14777         Handle initialisation of V4HF. Adjust initialisation of reinterpret
14778         built-ins.
14779         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
14780         (arm_vector_mode_supported_p): Handle V4HF.
14781         (arm_mangle_map): Handle V4HFmode.
14782         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
14783         * config/arm/arm_neon_builtins.def: Add entries for
14784         vcvtv4hfv4sf, vcvtv4sfv4hf.
14785         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
14786         (neon_vcvtv4hfv4sf): Likewise.
14787         * config/arm/neon-gen.ml: Handle half-precision floating point
14788         features.
14789         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
14790         * config/arm/arm_neon.h: Regenerate.
14791         * config/arm/neon.ml (type elts): Add F16.
14792         (type vectype): Add T_float16x4, T_floatHF.
14793         (type vecmode): Add V4HF.
14794         (type features): Add Requires_FP_bit feature.
14795         (elt_width): Handle F16.
14796         (elt_class): Likewise.
14797         (elt_of_class_width): Likewise.
14798         (mode_of_elt): Refactor.
14799         (type_for_elt): Handle F16, fix error messages.
14800         (vectype_size): Handle T_float16x4.
14801         (vcvt_sh): New function.
14802         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
14803         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
14804         (string_of_mode): Handle V4HF.
14805         * doc/arm-neon-intrinsics.texi: Regenerate.
14807 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
14809         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
14810         format specifier in 'X' case.
14812 2013-04-25  Alan Modra  <amodra@gmail.com>
14814         PR target/57052
14815         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
14816         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
14817         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
14818         Repeat for many other rotate/shift and mask patterns using subregs.
14819         Name lshiftrt insns.
14820         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
14821         on WORDS_BIG_ENDIAN.
14823 2013-04-25  Alan Modra  <amodra@gmail.com>
14825         * config.gcc: Support little-endian powerpc-linux targets.
14826         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
14827         (LINK_OS_LINUX_SPEC): Define.
14828         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
14829         Preserve MASK_LITTLE_ENDIAN.
14830         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
14831         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
14832         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
14833         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
14834         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
14835         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
14836         Correct fp word order for little-endian.  Don't shift toc entries
14837         smaller than a word for little-endian.
14838         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
14839         (bswapdi2 splits): Correct low-part subreg for little-endian.
14840         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
14841         low/high where such is correct only for be.
14842         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
14843         little-endian for -mcall-aixdesc.
14845 2013-04-25  Alan Modra  <amodra@gmail.com>
14847         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
14848         replace_equiv_address_nv.
14850 2013-04-25  Alan Modra  <amodra@gmail.com>
14852         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
14854 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
14856         Revert:
14857         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
14858         * rtl.h (struct rtx_def): ...
14860 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
14862         PR rtl-optimizations/57046
14863         * lra-constraints (split_reg): Set up lra_risky_transformations_p
14864         for multi-reg splits.
14866 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
14868         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
14870 2013-04-24  Sterling Augustine  <saugustine@google.com>
14872         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
14873         (comp_dir_string, debug_str_dwo_section): New.
14874         (DEBUG_STR_DWO_SECTION): Rename to ...
14875         (DEBUG_DWO_STR_SECTION): ... this.
14876         (DEBUG_NORM_STR_SECTION): Delete.
14877         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
14878         (DEBUG_STR_DWO_SECTION_FLAGS): New.
14879         (find_AT_string): Move most logic to ...
14880         (find_AT_string_in_table): ... here.  New.
14881         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
14882         add_skeleton_AT_string.  Delete logic.
14883         (output_skeleton_debug_sections): Remove call to
14884         add_top_level_skeleton_die_attrs.
14885         (add_comp_dir_attribute): Move logic to comp_dir_string.
14886         (dwarf2out_init): Initialize debug_str_dwo_section.
14887         (output_indirect_string): Call find_string_form.
14888         (output_indirect_strings): Rewrite.
14889         (prune_unused_types): Empty skeleton_debug_str_hash.
14890         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
14891         (dwarf2out_finish):  Call output_indirect_strings.
14893 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
14895         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
14897 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
14899         * rtl.h (struct rtx_def): Add comment for field jump.
14900         (LRA_SUBREG_P): New macro.
14901         * recog.c (register_operand): Check LRA_SUBREG_P.
14902         * lra.c (lra): Add note at the end of RTL code. Align non-empty
14903         stack frame.
14904         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
14905         (lra_final_code_change): Skip subreg change for operators.
14906         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
14907         if there are no operand changes.
14908         * lra-constraints.c (curr_insn_set): New.
14909         (match_reload): Set LRA_SUBREG_P.
14910         (emit_spill_move): Ditto.
14911         (check_and_process_move): Use curr_insn_set. Process only single
14912         set insns.  Don't initialize sec_mem_p and change_p.
14913         (simplify_operand_subreg): Use LRA_SUBREG_P.
14914         (reg_in_class_p): New function.
14915         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
14916         of #ifdef.  Add code to remove cycling.
14917         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
14918         non-null disp.  Reload inner instead of disp when base and index
14919         are null.  Try to put lo_sum into register.
14920         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
14921         (check_and_process_move): Move code for move cost check to
14922         simple_move_p.  Remove equiv_substitution.
14923         (simple_move_p): New function.
14924         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
14925         curr_insn_set.  Call check_and_process_move only for single set
14926         insns.  Use the new function.  Move call of check_and_process_move
14927         after operand equiv substitution and address process.
14929 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
14931         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
14932         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
14933         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
14935 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
14937         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
14939 2013-04-24  Marek Polacek  <polacek@redhat.com>
14941         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
14942         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
14943         (select_loops_exit_conditions): Likewise.
14944         (number_of_iterations_for_all_loops): Likewise.
14945         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
14946         (scev_analysis): Likewise.
14948 2013-04-02  Catherine Moore  <clm@codesourcery.com>
14949             Chao-ying Fu  <fu@mips.com>
14951         * config/mips/micromips.md (jraddiusp): New pattern.
14952         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
14953         instruction if possible.
14955 2013-04-24  Alan Modra  <amodra@gmail.com>
14957         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
14959 2013-04-24  Julian Brown  <julian@codesourcery.com>
14960             Chung-Lin Tang  <cltang@codesourcery.com>
14962         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
14963         dependency behavior in enumeration type DIE generation. Add TODO note
14964         to comments about future DW_FORM_sdata/udata re-work of related code.
14966 2013-04-23  Lawrence Crowl  <crowl@google.com>
14968         * Makefile.in: Update as needed below.
14970         * hash-table.h (class hash_table):
14971         Correct many methods with parameter types compare_type to the correct
14972         value_type.  (Correct code was unlikely to notice the change.)
14973         (hash_table::elements_with_deleted) New.
14974         (class hashtable::iterator): New.
14975         (hashtable::begin()): New.
14976         (hashtable::end()): New.
14977         (FOR_EACH_HASH_TABLE_ELEMENT): New.
14979         * statistics.c (statistics_hashes):
14980         Change type to hash_table.  Update dependent calls and types.
14982         * tree-into-ssa.c (var_infos):
14983         Change type to hash_table.  Update dependent calls and types.
14985         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
14986         Change type to hash_table.  Update dependent calls and types.
14988         * tree-ssa-loop-im.c (struct mem_ref.refs):
14989         Change type to hash_table.  Update dependent calls and types.
14991         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
14992         Change type to hash_table.  Update dependent calls and types.
14994         * tree-ssa-sccvn.c (vn_tables_s::nary):
14995         Change type to hash_table.  Update dependent calls and types.
14996         (vn_tables_s::phis): Likewise.
14997         (vn_tables_s::references): Likewise.
14999         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
15000         (vn_reference_eq): Update parameter and return types.
15002         * tree-ssa-structalias.c (pointer_equiv_class_table):
15003         Change type to hash_table.  Update dependent calls and types.
15004         (location_equiv_class_table): Likewise.
15006         * tree-vect-data-refs.c: Consequential changes for making
15007         peeling a hash_table.
15009         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
15010         (destroy_loop_vec_info): Dependent hash_table update.
15012         * tree-vectorizer.h (peeling_htab):
15013         Change type to hash_table.  Update dependent calls and types.
15015 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
15017         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
15018         to check the register content is equal or not.
15019         * lra-constraints.c (match_reload): Use lra_assign_reg_val
15020         to assign register content record.
15021         * lra-eliminations.c (update_reg_eliminate): Use
15022         lra_update_reg_val_offset to update register content offset.
15023         * lra-int.h (struct lra_reg): Add offset member.
15024         (lra_reg_val_equal_p): New static inline function.
15025         (lra_update_reg_val_offset): New static inline function.
15026         (lra_assign_reg_val): New static inline function.
15027         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
15028         to assign register content record.
15029         (initialize_lra_reg_info_element): Initial offset to zero.
15031 2013-04-23  Catherine Moore  <clm@codesourcery.com>
15033         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
15034         operands.  Record compression.
15036 2013-04-23  Xinliang David Li  <davidxl@google.com>
15038         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
15040 2013-04-23  Richard Biener  <rguenther@suse.de>
15042         PR middle-end/57036
15043         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
15044         parameter, only add abnormal goto edges from the copied body
15045         if the call could perform abnormal gotos.
15046         (copy_cfg_body): Adjust.
15048 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
15050         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
15052 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
15054         * coretypes.h (gimple_stmt_iterator): Add struct to make
15055         compatible with C.
15057 2013-04-23  Richard Biener  <rguenther@suse.de>
15059         PR tree-optimization/57026
15060         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
15061         from SSA names occuring in abnormal PHI nodes.
15063 2013-04-22  Andi Kleen  <ak@linux.intel.com>
15065         * lto/lto.c (print_lto_report_1): Fix LTO report names.
15067 2013-04-22  Andi Kleen  <ak@linux.intel.com>
15069         * lto/lto.c (print_lto_report_1): Declare early.
15070         (read_cgraph_and_symbols): Call print_lto_report_1 early.
15072 2013-04-22  Andi Kleen  <ak@linux.intel.com>
15074         * common.opt (-flto-report-wpa): Add.
15075         * doc/invoke.texi (-flto-report-wpa): Add.
15076         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
15077         (lto_main): dito.
15079 2013-04-22  Xinliang David Li  <davidxl@google.com>
15081         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
15082         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
15083         * Makefile.in: New dependency
15085             David Daney <ddaney.cavm@gmail.com>
15087         * configure.ac (gcc_cv_as_micromips_support):  Use the
15088         --fatal-warnings option.
15089         * configure: Regenerate.
15091 2013-04-22  Marek Polacek  <polacek@redhat.com>
15093         PR sanitizer/56990
15094         * tsan.c (instrument_expr): Don't instrument expression
15095         in case its size is zero.
15097 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
15099         PR target/57032
15100         Revert:
15101         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
15103         * config/alpha/alpha.c (TARGET_LRA_P): New define.
15105 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
15107         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
15108         (gimple_stmt_iterator): New typedef.
15109         * gimple.h (gimple_stmt_iterator): Rename to...
15110         (gimple_stmt_iterator_d): ... This.
15111         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
15112         trees be valid for GIMPLE and GENERIC.
15113         (TARGET_GIMPLE_FOLD_BUILTIN): New.
15114         * gimple-fold.c (gimple_fold_call): Call target hook
15115         gimple_fold_builtin.
15116         * hooks.c (hook_bool_gsiptr_false): New.
15117         * hooks.h (hook_bool_gsiptr_false): New.
15118         * target.def (fold_stmt): New.
15119         * doc/tm.texi: Regenerate.
15121 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
15123         PR target/57018
15124         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
15125         a set sp if no stack realignment.
15127 2013-04-22  Nick Clifton  <nickc@redhat.com>
15129         * config.gcc (tilegx-linux): Extend extra_objs rather than
15130         overwriting it.
15131         (tilepro-linux): Likewise.
15133 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
15135         * config/aarch64/aarch64-builtins.c
15136         (CF): Remove.
15137         (CF0, CF1, CF2, CF3, CF4, CF10): New.
15138         (VAR<1-12>): Add MAP parameter.
15139         (BUILTIN_*): Likewise.
15140         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
15141         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
15142         (aarch64_ushl_n<mode>): Likewise.
15143         (aarch64_sshr_n<mode>): Likewise.
15144         (aarch64_ushr_n<mode>): Likewise.
15145         (aarch64_<maxmin><mode>): Likewise.
15146         (aarch64_sqrt<mode>): Likewise.
15147         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
15148         (vshr<q>_n_*): Likewise.
15150 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
15152         * config/aarch64/aarch64-builtins.c
15153         (aarch64_simd_builtin_type_mode): Handle SF types.
15154         (sf_UP): Define.
15155         (BUILTIN_GPF): Define.
15156         (aarch64_init_simd_builtins): Handle SF types.
15157         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
15158         (frecps): Likewise.
15159         (frecpx): Likewise.
15160         * config/aarch64/aarch64-simd.md
15161         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
15162         (aarch64_frecpe<mode>): New.
15163         (aarch64_frecps<mode>): Likewise.
15164         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
15165         (v8type): Add frecp<esx>.
15166         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
15167         (aarch64_frecps<mode>): Likewise.
15168         * config/aarch64/iterators.md (FRECP): New.
15169         (frecp_suffix): Likewise.
15170         * config/aarch64/arm_neon.h
15171         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
15173 2013-04-22  Christian Bruel  <christian.bruel@st.com>
15175         PR target/56995
15176         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
15177         (REG_CLASS_NAMES): Idem.
15178         (REG_CLASS_CONTENTS): Idem.
15179         (REGCLASS_HAS_FP_REG): Idem.
15180         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
15181         (sh_conditional_register_usage): Idem.
15183 2013-04-21  Jeff Law  <law@redhat.com>
15185         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
15186         (ssa_forward_propagate_and_combine): Use it.
15188 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
15190         * lra.c: Update the flow chart diagram.
15192 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
15194         PR rtl-optimization/56847
15195         * lra-constraints.c (process_alt_operands): Discourage alternative
15196         with non-matche doffsettable memory constraint fro memory with
15197         known offset.
15199 2013-04-19  Richard Biener  <rguenther@suse.de>
15201         PR tree-optimization/56982
15202         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
15203         function.
15204         * gimplify.c (gimplify_call_expr): Notice special calls.
15205         (gimplify_modify_expr): Likewise.
15206         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
15207         abnormal control flow receivers.
15208         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
15209         in the same way as cfun->has_nonlocal_labels.
15210         (gimple_purge_dead_abnormal_call_edges): Likewise.
15211         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
15212         receivers start a basic-block.
15214 2013-04-19  Richard Biener  <rguenther@suse.de>
15216         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
15217         member ...
15218         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
15219         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
15220         (SLP_TREE_LOAD_PERMUTATION): Add.
15221         (vect_transform_slp_perm_load): Adjust prototype.
15222         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
15223         (vect_free_slp_instance): Likewise.
15224         (vect_create_new_slp_node): Likewise.
15225         (vect_supported_slp_permutation_p): Remove.
15226         (vect_slp_rearrange_stmts): Adjust.
15227         (vect_supported_load_permutation_p): Likewise.  Inline
15228         vect_supported_slp_permutation_p here.
15229         (vect_analyze_slp_instance): Compute load permutations per
15230         slp node instead of per instance.
15231         (vect_get_slp_defs): Adjust.
15232         (vect_transform_slp_perm_load): Likewise.
15233         (vect_schedule_slp_instance): Remove redundant code.
15234         (vect_schedule_slp): Remove hack for PR56270, add it ...
15235         * tree-vect-stmts.c (vectorizable_load): ... here, do not
15236         CSE loads for SLP.  Adjust.
15238 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
15240         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
15241         spelling in two comments.
15243 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
15245         PR target/56797
15246         * config/arm/arm.c (load_multiple_sequence): Require SP
15247         as base register for loads if SP is in the register list.
15249 2013-04-19  Martin Jambor  <mjambor@suse.cz>
15251         PR tree-optimization/56718
15252         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
15253         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
15254         and made public.  Adjusted all callers.
15255         (ipa_intraprocedural_devirtualization): New function.
15256         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
15257         (ipa_intraprocedural_devirtualization): Likewise.
15258         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
15260 2013-04-19  Richard Biener  <rguenther@suse.de>
15262         PR tree-optimization/57000
15263         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
15265 2013-04-19  Terry Guo  <terry.guo@arm.com>
15267         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
15268         Replace with ...
15269         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
15270         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
15271         (cortex_m4_fmacs): Use new reservations.
15272         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
15274 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
15276         PR rtl-optimization/56999
15277         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
15278         related code.
15279         (lra_coalesce): Remove split_origin_bitmap and related code.
15280         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
15281         ranges if necessary.
15283 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
15285         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
15286         New array.
15287         (ix86_expand_call): Remove clobbered_registers array and use
15288         x86_64_ms_sysv_extra_clobbered_registers instead.
15289         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
15290         Declare here.
15291         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
15292         predicate.
15293         * config/i386/i386.md (*call_rex64_ms_sysv): Use
15294         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
15295         (*call_value_rex64_ms_sysv): Ditto.
15297 2013-04-18  Cary Coutant  <ccoutant@google.com>
15299         * dwarf2out.c (output_pubnames): Check die_perennial_p of
15300         parent instead of die_mark.
15302 2013-04-18  Diego Novillo  <dnovillo@google.com>
15304         * gimple.c (create_gimple_tmp): New.
15305         (get_expr_type): New.
15306         (build_assign): New.
15307         (build_type_cast): New.
15308         * gimple.h (enum ssa_mode): Define.
15309         (gimple_seq_set_location): New.
15310         * asan.c (build_check_stmt): Change some gimple_build_* calls
15311         to use build_assign and build_type_cast.
15313 2013-04-18  Richard Biener  <rguenther@suse.de>
15315         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
15316         handle negative step.  Remove redundant checks.
15317         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
15318         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
15319         for negative step and grouped loads fail to vectorize.
15321 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
15323         * emit-rtl.c (reset_insn_used_flags): New function.
15324         (reset_all_used_flags): Use it.
15325         (verify_insn_sharing): New function.
15326         (verify_rtl_sharing): Fix verification for SEQUENCEs.
15328 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
15330         PR tree-optimization/56984
15331         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
15332         and (x >> M) >= N don't register any assertion if N << M is the
15333         minimum value.
15335 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
15337         * lower-subreg.c (resolve_simple_move): If called self-recursive,
15338         do not delete_insn insns that have not yet been emitted, only
15339         unlink them with remove_insn.
15340         * df-scan.c (df_insn_delete): Revert r197492.
15342 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
15344         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
15345         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
15347 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
15349         * config/arm/arm.md (movsicc_insn): Convert define_insn into
15350         define_insn_and_split.
15351         (and_scc,ior_scc,negscc): Likewise.
15352         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
15354 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
15356         * config/arm/arm.c (use_return_insn): Return 0 for targets that
15357         can benefit from using a sequence of LDRD instructions in epilogue
15358         instead of a single LDM instruction.
15360 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
15362         PR 45688
15363         * doc/extend.texi: Fix typo.
15365 2013-04-17  Richard Biener  <rguenther@suse.de>
15367         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
15368         (vect_build_slp_tree): ... here.
15369         (vect_build_slp_tree_1): Compute which stmts of the SLP group
15370         match.  Remove special-casing of mismatched complex loads.
15371         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
15372         re-try the match with swapped commutative operands.
15373         (vect_supported_load_permutation_p): Remove special-casing of
15374         mismatched complex loads.
15375         (vect_analyze_slp_instance): Adjust.
15377 2013-04-17  Richard Biener  <rguenther@suse.de>
15379         PR rtl-optimization/56921
15380         * cfgloop.h (struct loop): Add simple_loop_desc member.
15381         (struct niter_desc): Mark with GTY(()).
15382         (simple_loop_desc): Do not use aux field but simple_loop_desc.
15383         * loop-iv.c (get_simple_loop_desc): Likewise.
15384         (free_simple_loop_desc): Likewise.
15386         Revert
15387         2013-04-16  Richard Biener  <rguenther@suse.de>
15389         PR rtl-optimization/56921
15390         * loop-init.c (pass_rtl_move_loop_invariants): Add
15391         TODO_do_not_ggc_collect to todo_flags_finish.
15392         (pass_rtl_unswitch): Same.
15393         (pass_rtl_unroll_and_peel_loops): Same.
15394         (pass_rtl_doloop): Same.
15396 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
15398         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
15399         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
15400         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
15401         references.
15402         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
15403         * tree-streamer.c (record_common_node): Adjust reference in comment.
15405 2013-04-17  Terry Guo  <terry.guo@arm.com>
15407         * config/arm/cortex-m4.md: Add a new bypass.
15409 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
15411         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
15412         New pattern.
15413         (*subs_<optab><mode>_multp2): New pattern.
15414         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
15415         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
15417 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
15419         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
15420         (*subs_mul_imm_<mode>): New pattern.
15422 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
15424         PR target/56948
15425         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
15426         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
15427         (vsx_movti_32bit): Change j->wa to O->wa.
15429 2013-04-16  Richard Biener  <rguenther@suse.de>
15431         PR rtl-optimization/56921
15432         * loop-init.c (pass_rtl_move_loop_invariants): Add
15433         TODO_do_not_ggc_collect to todo_flags_finish.
15434         (pass_rtl_unswitch): Same.
15435         (pass_rtl_unroll_and_peel_loops): Same.
15436         (pass_rtl_doloop): Same.
15438 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
15440         * config/arm/arm.c (emit_multi_reg_push): New declaration
15441         for an existing function.
15442         (arm_emit_strd_push): New function.
15443         (arm_expand_prologue): Used here.
15444         (arm_emit_ldrd_pop): New function.
15445         (arm_expand_epilogue): Used here.
15446         (arm_get_frame_offsets): Update condition.
15447         (arm_emit_multi_reg_pop): Add a special case for load of a single
15448         register with writeback.
15450 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
15452         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
15453         description.
15455 2013-04-16  Richard Biener  <rguenther@suse.de>
15457         PR tree-optimization/56756
15458         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
15459         (first_mem_ref_loc): New.
15460         (execute_sm): Place the load temporarily before a previous
15461         access instead of in the latch edge to ensure its SSA dependencies
15462         are defined at points dominating the load.
15464 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
15466         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
15467         correct fix by moving header and footer insn to the footer of
15468         the merged basic block.  Clear BB_END of the merged-away block.
15470         PR middle-end/43631
15471         * emit-rtl.c (make_note_raw): New function.
15472         (link_insn_into_chain): New static inline function.
15473         (add_insn): Use it.
15474         (add_insn_before, add_insn_after): Factor insn chain linking code...
15475         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
15476         using link_insn_into_chain.
15477         (note_outside_basic_block_p): New helper function for emit_note_after
15478         and emit_note_before.
15479         (emit_note_after): Use nobb variant of add_insn_after if the note
15480         should not be contained in a basic block.
15481         (emit_note_before): Use nobb variant of add_insn_before if the note
15482         should not be contained in a basic block.
15483         (emit_note_copy): Use make_note_raw.
15484         (emit_note): Likewise.
15485         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
15486         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
15487         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
15488         the moved barrier the tail of the basic block it follows.
15489         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
15491 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
15493         PR tree-optimization/56962
15494         * gimple-ssa-strength-reduction.c (record_increment): Only set
15495         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
15496         either rhs1 or rhs2 is equal to c->base_expr.
15498 2013-04-15  Richard Biener  <rguenther@suse.de>
15500         PR tree-optimization/56933
15501         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
15502         member.
15503         (GROUP_READ_WRITE_DEPENDENCE): Remove.
15504         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
15505         * tree-vect-data-refs.c (vect_analyze_group_access): Move
15506         dependence check ...
15507         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
15508         ... here.
15509         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
15510         GROUP_READ_WRITE_DEPENDENCE.
15512 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15514         * emit-rtl.c (reset_all_used_flags): New function.
15515         (verify_rtl_sharing): Call reset_all_used_flags before and after
15516         performing the checks.
15518 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15520         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
15521         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
15522         * config/arm/constraints.md (De): New constraint.
15523         * config/arm/neon.md (anddi3_neon): Delete.
15524         (neon_vand<mode>): Expand to standard anddi3 pattern.
15525         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
15526         Move earlier in the file.
15527         (neon_inv_logic_op2): Likewise.
15528         (arm_anddi_operand_neon): New predicate.
15530 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
15532         * configure.ac (gcc_cv_ld_as_needed): Set
15533         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
15534         Use -z ignore, -z record on *-*-solaris2*.
15535         (HAVE_LD_AS_NEEDED): Update comment.
15536         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
15537         * configure: Regenerate.
15538         * config.in: Regenerate.
15539         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
15540         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
15541         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
15542         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
15543         equivalents.  Fix markup.
15544         * doc/tm.texi: Regenerate.
15546 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
15548         * config/i386/i386.opt: New option mstack-protector-guard=.
15549         * config/i386/i386-opts.h: Add enum stack_protector_guard.
15550         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
15551         TARGET_SSP_TLS_GUARD.
15552         * config/i386/i386.c (ix86_option_override_internal): Set
15553         ix86_stack_protector_guard.
15554         * config/i386/i386.md (stack_protect_set): Enable for
15555         TARGET_SSP_TLS_GUARD only.
15556         (stack_protect_set_<mode>): Ditto.
15557         (stack_protect_test): Ditto.
15558         (stack_protect_test_<mode>): Ditto.
15559         * doc/invoke.texi (i386 Option): Document.
15561 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
15563         PR target/56890
15564         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
15565         (S_MODES): Set H_MODE bit.
15566         (SF_MODES): Set only S_MODE and SF_MODE bits.
15567         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
15568         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
15569         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
15570         <MODE_FLOAT>: Likewise.
15572 2013-04-15  Joey Ye  <joey.ye@arm.com>
15574         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
15576 2013-04-15  Joey Ye  <joey.ye@arm.com>
15578         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
15579         for real far jump.
15580         (thumb_far_jump_used_p): Count instruction size and set
15581         far_jump_used.
15583 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
15585         * reorg.c (fill_simple_delay_slots): Reindent block of code.
15586         * resource.c (mark_target_live_regs): Reformat conditional block.
15588 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
15590         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
15591         notes, they are emitted only just before final.
15592         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
15594 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
15596         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
15597         * cfgrtl.c (delete_insn): Call it here instead.
15598         * lra-spills.c (lra_final_code_change): Use delete_insn.
15599         * haifa-sched.c (sched_remove_insn): Likewise.
15600         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
15601         returning to the nop pool.
15602         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
15603         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
15605 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
15607         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
15608         * doc/tm.texi: Regenerated.
15610 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
15612         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
15613         QImode checks.
15615 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
15617         * df-core.c (df_find_def): Compare register numbers.
15618         (df_find_use): Likewise.
15620 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
15622         PR target/56903
15623         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
15624         lra_in_progress for return.
15626 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
15628         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
15629         define_insn into define_insn_and_split and emit movsicc patterns.
15631 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
15633         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
15635 2013-04-12  Richard Biener  <rguenther@suse.de>
15637         * tree-pass.h (TODO_do_not_ggc_collect): New.
15638         * passes.c (execute_one_ipa_transform_pass): Honor
15639         TODO_do_not_ggc_collect.
15640         (execute_one_pass): Likewise.
15642         Revert
15643         2013-04-10  Richard Biener  <rguenther@suse.de>
15645         * passes.c (init_optimization_passes): Remove reload pass.
15646         * ira.c (do_reload): Merge into ...
15647         (ira): ... this.
15648         (rest_of_handle_reload): Remove.
15649         (pass_reload): Likewise.
15650         * config/i386/i386.c (ix86_option_override): Refer to ira instead
15651         of reload for vzeroupper pass placement.
15653 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
15655         PR tree-optimization/56918
15656         PR tree-optimization/56920
15657         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
15658         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
15659         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
15660         use wide_mul_with_sign method.
15662 2013-04-12  Richard Biener  <rguenther@suse.de>
15664         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
15665         not be considered a gimple constant.
15667 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
15669         * fold-const.c (const_binop): Handle vector shifts by a scalar.
15670         (fold_binary_loc): Call const_binop also for mixed vector-scalar
15671         operations.
15673 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
15674             Jakub Jelinek  <jakub@redhat.com>
15676         * opts.c: Include diagnostic-color.h.
15677         (common_handle_option): Handle OPT_fdiagnostics_color_.
15678         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
15679         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
15680         (diagnostic-color.o): New.
15681         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
15682         (diagnostic_color_rule): New enum.
15683         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
15684         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
15685         the location string.
15686         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
15687         either NULL, or color kind.
15688         * diagnostic-color.c: New file.
15689         * diagnostic-color.h: New file.
15690         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
15691         arguments.
15692         * doc/invoke.texi (-fdiagnostics-color): Document.
15693         * pretty-print.h (pp_show_color): Define.
15694         (struct pretty_print_info): Add show_color field.
15695         * diagnostic.c: Include diagnostic-color.h.
15696         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
15697         macros.  Colorize error:, warning: etc. strings and also the location
15698         string.
15699         (diagnostic_show_locus): Colorize the caret line.
15700         * pretty-print.c: Include diagnostic-color.h.
15701         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
15702         inside of %< %> quotes or quoted through q format modifier.
15704 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15706         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
15708 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
15710         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
15711         code in CC_NZ mode.
15712         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
15713         pattern.
15715 2013-04-11  Marek Polacek  <polacek@redhat.com>
15717         PR tree-optimization/48184
15718         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
15720 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
15722         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
15723         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
15724         (skip_simple_arithmetic): Tidy up.
15725         * tree.h (skip_simple_constant_arithmetic): Declare.
15727 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
15729         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
15731 2013-04-11  Richard Biener  <rguenther@suse.de>
15733         * tree-vect-loop.c (get_initial_def_for_induction): Properly
15734         generate vector constants.
15736 2013-04-11  Richard Biener  <rguenther@suse.de>
15738         PR tree-optimization/56878
15739         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
15740         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
15741         New function.
15742         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
15743         Prefer to align the DR with the most invariant base address.
15745 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
15747         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
15748         comment.
15750 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
15752         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
15753         floating-point vector comparisons against 0.
15755 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
15757         PR tree-optimization/56899
15758         * fold-const.c (extract_muldiv_1): Apply distributive law
15759         only if TYPE_OVERFLOW_WRAPS (ctype).
15761 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
15763         PR target/56124
15764         * ira-costs.c (scan_one_insn): Check whether the source rtx of
15765         loading has side effect.
15767 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
15769         * config/sparc/sparc.c: Include tree-pass.h.
15770         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
15771         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
15772         head of file.  Change return type.  Split off gate function.
15773         (sparc_gate_work_around_errata): New function.
15774         (pass_work_around_errata): New pass definition.
15775         (insert_pass_work_around_errata) New pass insert definition to
15776         insert pass_work_around_errata just after delayed-branch scheduling.
15777         (sparc_option_override): Insert the pass.
15778         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
15780 2013-04-10  David S. Miller  <davem@davemloft.net>
15782         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
15783         or -mcpu=hypersparc.
15785         * target.def (cstore_mode): New hook.
15786         * target.h: Include insn-codes.h
15787         * targhooks.c: Likewise.
15788         (default_cstore_mode): New function.
15789         * targhooks.h: Declare it.
15790         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
15791         * doc/tm.texi: Rebuild.
15792         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
15793         target hook, rather than inspecting the insn_data.
15794         * config/sparc/sparc.c (sparc_cstore_mode): New function.
15795         (TARGET_CSTORE_MODE): Redefine.
15796         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
15797         result patterns.
15798         * config/sparc/predicates.md (cstore_result_operand): New special
15799         predicate.
15800         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
15801         Use it for operand 0.
15802         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
15803         (*snesi_special): Likewise.
15804         (*snesi_zero): Likewise.
15805         (*seqsi_zero): Likewise.
15806         (*sltu_insn): Likewise.
15807         (*sgeu_insn): Likewise.
15808         (*seqdi_special): Make operand 0 and comparison operation be of
15809         DImode.
15810         (*snedi_special): Likewise.
15811         (*snedi_special_vis3): Likewise.
15812         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
15813         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
15814         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
15815         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
15816         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
15817         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
15818         (*sltu_extend_sp64): Likewise.
15819         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
15820         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
15821         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
15822         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
15823         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
15825 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
15827         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
15828         (aarch64_start_file): Use the new function.
15830 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
15831             Jason Merrill  <jason@redhat.com>
15833         * common.opt: Add -gdwarf.
15834         * opts.c (common_handle_option): Handle it.
15835         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
15837 2013-04-10  Richard Biener  <rguenther@suse.de>
15839         * passes.c (execute_todo): Do not call ggc_collect conditional here.
15840         (execute_one_ipa_transform_pass): But unconditionally here.
15841         (execute_one_pass): And here.
15842         (init_optimization_passes): Remove reload pass.
15843         * tree-pass.h (TODO_ggc_collect): Remove.
15844         (pass_reload): Likewise.
15845         * ira.c (do_reload): Merge into ...
15846         (ira): ... this.
15847         (rest_of_handle_reload): Remove.
15848         (pass_reload): Likewise.
15849         * config/i386/i386.c (ix86_option_override): Refer to ira instead
15850         of reload for vzeroupper pass placement.
15851         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
15852         and todo_flags_finish of all passes.
15854 2013-04-10  Richard Biener  <rguenther@suse.de>
15856         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
15857         first_const_oprnd field, rename first_def_type to first_op_type.
15858         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
15859         (vect_get_and_check_slp_defs): Always use the type of the
15860         operand.  Allow mixed vect_external_def, vect_constant_def types.
15861         (vect_get_constant_vectors): Handle mixed vect_external_def,
15862         vect_constant_def types.
15864 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
15866         PR tree-optimization/55524
15867         * tree-ssa-math-opts.c
15868         (convert_mult_to_fma): Don't use an fms construct
15869         when we don't have an fms operation, but fnma, and it looks
15870         likely that we'll be able to use the latter.
15872 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
15874         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
15875         function.
15876         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
15877         inline fail caused by overwritable functions.
15879 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
15881         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
15882         unnecessary bits in the constant power of two case.
15884 2013-04-10  Richard Biener  <rguenther@suse.de>
15886         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
15887         broken code swapping operands.
15888         (vect_build_slp_tree): Do not compute load permutations here.
15889         (vect_analyze_slp_instance): Compute load permutations here,
15890         after building the SLP tree.
15892 2013-04-09  Christian Bruel  <christian.bruel@st.com>
15894         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
15895         of next/prev_real_insn.
15897 2013-04-09  Jan Hubicka  <jh@suse.cz>
15899         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
15900         Drop aliased parameter.
15901         (function_and_variable_visibility): Do not handle alias pairs.
15902         * cgraph.c (varpool_externally_visible_p): Update prototype.
15903         * varpool.c (varpool_add_new_variable): Update.
15905 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15907         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
15909 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
15911         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
15913         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
15915 2013-04-09  Marek Polacek  <polacek@redhat.com>
15917         PR tree-optimization/48762
15918         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
15920 2013-04-09  Richard Biener  <rguenther@suse.de>
15922         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
15923         dealing with cost.
15924         (vect_build_slp_tree): Likewise.
15925         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
15926         calculating the cost of a SLP instance.
15927         (vect_analyze_slp_instance): Use it from here, after building
15928         the SLP tree.
15930 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
15932         PR middle-end/56883
15933         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
15934         expand_omp_for_static_chunk): Use simple_p = true in
15935         force_gimple_operand_gsi calls when assigning to addressable decls.
15937 2013-04-09  Jeff Law  <law@redhat.com>
15939         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
15940         when the boolean was created by converting a wider object which
15941         had a boolean range.
15943 2013-04-09  Richard Biener  <rguenther@suse.de>
15945         * tree-vectorizer.h (slp_void_p): Remove.
15946         (slp_tree): Typedef before _slp_tree declaration.
15947         (struct _slp_tree): Use a vector of slp_tree as children.
15948         (vect_get_place_in_interleaving_chain): Remove.
15949         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
15950         Move ...
15951         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
15952         and make static.
15953         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
15954         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
15955         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
15956         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
15957         Use slp_node instead of slp_void_p and adjust.
15959 2013-04-09  Richard Biener  <rguenther@suse.de>
15961         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
15962         work that is not necessary.
15964 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
15966         PR tree-optimization/56854
15967         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
15968         forward into clobber stmts if it would change MEM_REF lhs into
15969         non-MEM_REF.
15971 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
15973         * tree.c (type_hash_lookup, type_hash_add): Make static.
15974         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
15976 2013-04-09  Richard Biener  <rguenther@suse.de>
15978         * tree.h (unsave_expr_now): Remove.
15979         * tree-inline.c (mark_local_for_remap_r): Remove.
15980         (unsave_expr_1): Likewise.
15981         (unsave_r): Likewise.
15982         (unsave_expr_now): Likewise.
15983         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
15984         (propagate_tree_value): Likewise.
15986 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
15988         * doc/rtl.texi (sequence): Rewrite documentation to match the
15989         current use of SEQUENCE rtl objects.
15990         * rtl.def (SEQUENCE): Likewise.
15992         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
15993         Update documentation.
15994         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
15995         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
15997         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
15999 2013-04-08  Teresa Johnson  <tejohnson@google.com>
16001         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
16002         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
16003         methods.
16004         (estimate_edge_size_and_time): Add comment to suggest using rounding
16005         methods.
16006         (estimate_node_size_and_time): Ditto.
16007         (remap_edge_change_prob): Use helper rounding divide methods.
16008         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
16009         (gimple_mod_pow2_value_transform): Ditto.
16010         (gimple_mod_subtract_transform): Ditto.
16011         (gimple_ic_transform): Ditto.
16012         (gimple_stringops_transform): Ditto.
16013         * stmt.c (conditional_probability): Ditto.
16014         (emit_case_dispatch_table): Ditto.
16015         * lto-cgraph.c (merge_profile_summaries): Ditto.
16016         * tree-optimize.c (execute_fixup_cfg): Ditto.
16017         * cfgcleanup.c (try_forward_edges): Ditto.
16018         * cfgloopmanip.c (scale_loop_profile): Ditto.
16019         (loopify): Ditto.
16020         (duplicate_loop_to_header_edge): Ditto.
16021         (lv_adjust_loop_entry_edge): Ditto.
16022         * tree-vect-loop.c (vect_transform_loop): Ditto.
16023         * profile.c (compute_branch_probabilities): Ditto.
16024         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
16025         * lto-streamer-in.c (input_cfg): Ditto.
16026         * gimple-streamer-in.c (input_bb): Ditto.
16027         * ipa-cp.c (update_profiling_info): Ditto.
16028         (update_specialized_profile): Ditto.
16029         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
16030         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
16031         rounding methods.
16032         * sched-rgn.c (compute_dom_prob_ps): Ditto.
16033         (compute_trg_info): Ditto.
16034         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
16035         (purge_dead_edges): Ditto.
16036         * loop-unswitch.c (unswitch_loop): Ditto.
16037         * cgraphclones.c (cgraph_clone_edge): Ditto.
16038         (cgraph_clone_node): Ditto.
16039         * tree-inline.c (copy_bb): Ditto.
16040         (copy_edges_for_bb): Ditto.
16041         (initialize_cfun): Ditto.
16042         (copy_cfg_body): Ditto.
16043         (expand_call_inline): Ditto.
16045 2013-04-08  Kai Tietz  <ktietz@redhat.com>
16047         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
16048         TARGET_CYGWIN64 by TARGET_64BIT.
16050 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
16052         * config/epiphany/epiphany.md (GPR_1): New constant.
16053         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
16054         * config/epiphany/epiphany.c (gen_compare_reg):
16055         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
16056         is already in place.
16057         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
16058         Don't require being called during rtl expansion; If y operlaps r0,
16059         return 0.
16060         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
16061         (epiphany_expand_epilogue): Likewise.
16063         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
16064         Don't use CC_FPmode for ORDERED / UNORDERED.
16065         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
16067         * config/epiphany/constraints.md (CnL): New constraint.
16068         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
16069         * config/epiphany/predicates.md (add_operand): Allow 1024.
16071         * config/epiphany/epiphany.md (logical_op): New code iterator.
16072         (op_mnc): New code attribute.
16073         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
16074         (mov_f+1, mov_f+2): New peephole2 patterns.
16076         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
16077         (cstoresi4): Also allow re-use of zero result when doing a NE
16078         comparison to a non-zero operand.
16079         Use (clobber (scratch)) for first insn if the gpr output is not needed.
16081         * config/epiphany/epiphany.md (<insn_opname>v2si3):
16082         Use gen_addsi3_i / gen_subsi3_i.
16084 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
16086         PR c++/34949
16087         PR c++/50243
16088         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
16089         contain anything but clobbers, at most one __builtin_stack_restore,
16090         optionally debug stmts and final resx, and if it has at least one
16091         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
16092         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
16093         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
16094         which isn't defaut definition, remove them.
16095         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
16096         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
16097         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
16098         with MEM_REF LHS with SSA_NAME address.
16100 2013-04-08  Jeff Law  <law@redhat.com>
16102         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
16104 2013-04-08  Richard Biener  <rguenther@suse.de>
16106         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
16107         extra newline.
16108         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
16109         determined vector type.
16110         (vect_analyze_data_refs): Likewise.
16111         (vect_get_new_vect_var): Adjust.
16112         (vect_create_destination_var): Preserve SSA name versions.
16113         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
16114         not dump anything here.
16116 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
16118         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
16119         Add member lr_slot_known.
16120         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
16121         if necessary.
16122         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
16123         Remove code that sets lr_slot_offset according to what a previous
16124         version of epiphany_emit_save_restore used to do.
16125         (epiphany_emit_save_restore): When doing an lr save or restore,
16126         set/verify lr_slot_known and lr_slot_offset.
16128 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
16130         PR target/54338
16131         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
16132         in ALL_REGS.
16134 2013-04-08  Richard Biener  <rguenther@suse.de>
16136         * alias.c (find_base_term): Fix thinko in previous change.
16138 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
16140         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
16141         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
16142         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
16143         if possible to compute val.
16144         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
16145         For QImode integers don't require anything about precision.  Use
16146         const_with_all_bytes_same to find out if the constant doesn't have
16147         repeated bytes in it.
16149 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16151         * config/s390/s390.c (s390_expand_insv): Only accept insertions
16152         within mode size.
16154 2013-04-08  Marek Polacek  <polacek@redhat.com>
16156         PR rtl-optimization/48182
16157         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
16158         value to 1.
16160 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16162         PR target/55487
16163         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
16164         nuses, make sure we have a label.
16166 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16168         PR target/56843
16169         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
16170         (rs6000_emit_swdiv_low_precision): Remove.
16171         (rs6000_emit_swdiv): Rewrite to handle between one and four
16172         iterations of Newton-Raphson generally; modify required number of
16173         iterations for some cases.
16174         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
16176 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
16178         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
16179         set-but-unused variable.
16181         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
16182         basic blocks of released function bodies garbage-collectable.
16184         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
16185         (struct rtl_opt_pass): Add TODO_df_finish.
16187         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
16189 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16191         * config/arm/constraints.md (q): New constraint.
16192         * config/arm/ldrdstrd.md: New file.
16193         * config/arm/arm.md (ldrdstrd.md) New include.
16194         (arm_movdi): Use "q" instead of "r" constraint
16195         for double-word memory access.
16196         (movdf_soft_insn): Likewise.
16197         * config/arm/vfp.md (movdi_vfp): Likewise.
16198         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
16199         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
16200         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
16201         (mem_ok_for_ldrd_strd): Likewise.
16202         (output_move_double): Update assertion.
16204 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16206         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
16208 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16210         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
16211         define_insn_and_split.
16212         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
16214 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16216         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
16217         define_insn_and_split.
16218         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
16219         (shiftsi3_compare): New pattern.
16220         (rrx): New pattern.
16221         * config/arm/unspecs.md (UNSPEC_RRX): New.
16223 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16225         * config/arm/arm.md (negdi_extendsidi): New pattern.
16226         (negdi_zero_extendsidi): Likewise.
16228 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16230         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
16231         define_insn_and_split.
16232         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
16233         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
16235 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16237         * config/arm/arm.md (arm_subdi3): Convert define_insn into
16238         define_insn_and_split.
16239         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
16240         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
16242 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16244         * config/arm/arm.md (subsi3_carryin): New pattern.
16245         (subsi3_carryin_const): Likewise.
16246         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
16247         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
16249 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16251         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
16253 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
16255         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
16256         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
16258 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16260         * config/arm/arm.c (arm_expand_builtin): Change fcode
16261         type to unsigned int.
16263 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16265         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
16267 2013-04-04  Ian Lance Taylor  <iant@google.com>
16269         * doc/standards.texi (Standards): The Go frontend supports the Go 1
16270         language standard.
16272 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
16274         PR middle-end/56729
16275         * df-scan.c (df_insn_delete): Disable failing assert.
16277 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16279         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
16280         New function prototype.
16281         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
16282         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
16283         (arm_builtin_vectorized_function): New function.
16285 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16287         * config/arm/arm_neon_builtins.def: New file.
16288         * config/arm/arm.c (neon_builtin_data): Move contents to
16289         arm_neon_builtins.def.
16290         (enum arm_builtins): Include neon builtin definitions.
16291         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
16292         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
16294 2013-04-04  Marek Polacek  <polacek@redhat.com>
16296         PR tree-optimization/48186
16297         * predict.c (maybe_hot_frequency_p): Return false if
16298         HOT_BB_FREQUENCY_FRACTION is 0.
16299         (cgraph_maybe_hot_edge_p): Likewise.
16301 2013-04-04  Richard Biener  <rguenther@suse.de>
16303         PR tree-optimization/56826
16304         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
16305         more accurately.
16307 2013-04-04  Richard Biener  <rguenther@suse.de>
16309         PR tree-optimization/56213
16310         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
16311         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
16313 2013-04-04  Richard Biener  <rguenther@suse.de>
16315         PR tree-optimization/56837
16316         * tree-loop-distribution.c (classify_partition): For non-zero
16317         values require that the value has the same precision as its
16318         mode to be useful as memset value.
16320 2013-04-03  Nick Clifton  <nickc@redhat.com>
16322         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
16323         (fmssf4): Use fmsf.s on E3V5 architectures.
16324         (fnmasf4): Use fnmaf.s on E3V5 architectures.
16325         (fnmssf4): Use fnmsf.s on E3V5 architectures.
16327 2013-04-03  Jeff Law  <law@redhat.com>
16329         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
16330         (lra-eliminations.o): Likewise.
16332 2013-04-03  Teresa Johnson  <tejohnson@google.com>
16334         * gcov-io.c (compute_working_sets): Moved most of body of old
16335         compute_working_sets here from profile.c.
16336         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
16337         (gcov_working_set_t): Moved typedef here from basic-block.h
16338         (compute_working_set): Declare.
16339         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
16340         (get_working_sets): Renamed from compute_working_set,
16341         replace most of body with call to new compute_working_sets.
16342         (get_exec_counts): Replace call to compute_working_sets
16343         to get_working_sets.
16344         * profile.h (get_working_sets): Renamed from compute_working_set.
16345         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
16346         to get_working_sets.
16347         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
16348         * gcov-dump.c (dump_working_sets): New function.
16350 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
16352         * hwint.c (sext_hwi, zext_hwi): New functions.
16353         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
16354         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
16355         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
16356         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
16357         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
16358         (sext_hwi, zext_hwi): New functions.
16360 2013-04-03  Jeff Law  <law@redhat.com>
16362         PR tree-optimization/56799
16363         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
16364         back test for widening conversion erroneously dropped in prior change.
16366 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16368         PR target/56809
16369         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
16370         instead of next_real_insn.
16372 2013-04-03  Marek Polacek  <polacek@redhat.com>
16374         PR sanitizer/55702
16375         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
16377 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16379         PR target/56809
16380         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
16381         next_real_insn.
16382         (thumb1_output_casesi): Likewise.
16383         (thumb2_output_casesi): Likewise.
16385 2013-04-03  Richard Biener  <rguenther@suse.de>
16387         PR tree-optimization/56817
16388         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
16389         Split out ...
16390         (tree_unroll_loops_completely_1): ... new function to manually
16391         walk the loop tree, properly defering outer loops of unrolled
16392         loops to later iterations.
16394 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
16396         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
16397         (vectorizable_load): Likewise.
16398         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
16399         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
16401 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
16403         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
16404         BIT_FIELD_REF.
16406 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
16408         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
16410 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
16412         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
16414 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
16416         PR tree-optimization/56790
16417         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
16418         folding.
16420 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
16422         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
16423         Handle VEC_MERGE.
16424         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
16425         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
16426         equal arguments.
16428 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
16430         PR c/19449
16431         * tree.h (force_folding_builtin_constant_p): New decl.
16432         * builtins.c (force_folding_builtin_constant_p): New variable.
16433         (fold_builtin_constant_p): Fold immediately also if
16434         force_folding_builtin_constant_p.
16436 2013-04-03  Richard Biener  <rguenther@suse.de>
16438         PR tree-optimization/56812
16439         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
16440         DRs of the same interleaving chain are independent.
16442 2013-04-02  Jason Merrill  <jason@redhat.com>
16444         * gdbinit.in (pbb): Use debug fn.
16446 2013-04-02  Lawrence Crowl  <crowl@google.com>
16448         * sese.h (struct ivtype_map_elt_s): Remove unused.
16449         (extern debug_ivtype_map): Remove unused.
16450         (extern eq_ivtype_map_elts): Remove unused.
16451         * sese.c (debug_ivtype_map): Removed unused.
16452         (debug_ivtype_map_1): Removed unused.
16453         (debug_ivtype_elt): Remove unused.
16454         (eq_ivtype_map_elts): Remove unused.
16457 2013-04-02  Kai Tietz  <ktietz@redhat.com>
16459         PR target/52790
16460         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
16461         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
16462         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
16463         function.
16464         (legitimize_pe_coff_symbol): Likewise.
16465         (is_imported_p): New helper-function.
16466         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
16467         for Windows x64 targets.
16468         (ix86_expand_prologue): Optimize for pe-coff targets.
16469         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
16470         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
16471         medium/large code-model.
16472         (legitimize_pic_address): Likewise.
16473         (legitimize_tls_address): Likewise.
16474         (ix86_expand_call): Likewise.
16475         (x86_output_mi_thunk): Likewise.
16476         (get_dllimport_decl): Add new beimport argument.
16477         (construct_plt_address): Don't assert for x64 pe-coff targets.
16478         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
16479         targets.
16480         (SYMBOL_FLAG_STUBVAR): New macro.
16481         (SYMBOL_REF_STUBVAR_P): Likewise.
16482         * config/i386/winnt.c (stub_list): New structure.
16483         (stub_head): New local variable.
16484         (i386_pe_record_stub): New function.
16485         (i386_pe_file_end): Emit refptr-stubs.
16487 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
16489         PR rtl-optimization/56745
16490         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
16491         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
16493         PR c++/34949
16494         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
16495         and both of them are MEM_REFs, just compare first argument for
16496         equality and attempt to deal even with differing offsets.
16498         PR c++/34949
16499         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
16500         of gimple_clobber_p to be MEM_REF.
16501         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
16502         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
16503         after gimplification.
16504         * asan.c (get_mem_ref_of_assignment): Don't instrument
16505         gimple_clobber_p stmts.
16506         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
16507         gimple_clobber_p stmt if they have MEM_REF lhs and
16508         are dead because of another gimple_clobber_p stmt.
16509         * tree-ssa-live.c (clear_unused_block_pointer): Treat
16510         gimple_clobber_p stmts like debug stmts.
16511         (remove_unused_locals): Remove clobbers with MEM_REF lhs
16512         that refer to unused VAR_DECLs or uninitialized values.
16513         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
16514         gimple_clobber_p stmts if they refer to removed parameters.
16515         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
16516         formatting.
16518 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
16520         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
16521         using SWI48 mode attribute.
16523 2013-04-02  Wei Mi  <wmi@google.com>
16525         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
16526         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
16527         *<rotate_insn><mode>3_mask in i386.md.
16529 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
16531         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
16533 2013-04-02  Richard Biener  <rguenther@suse.de>
16535         PR tree-optimization/56778
16536         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
16537         Runtime alias tests are not supported for gather loads.
16538         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
16539         stmts referenced from SSA operands before updating SSA form.
16541 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
16542             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16544         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
16545         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
16546         * config/arm/cortex-a53.md: New file.
16547         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
16548         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
16549         * config/arm/arm.c (arm_issue_rate): Likewise.
16550         * config/arm/arm-tune.md: Regenerate
16551         * config/arm/arm-tables.opt: Regenerate.
16552         * config/arm/arm-cores.def: Add cortex-a53.
16554 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
16556         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
16557         non-static link.
16559 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
16561         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
16562         scalar load/store operations using B/H registers.
16563         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
16565 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
16567         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
16568         scalar move.
16569         * config/aarch64/aarch64.c
16570         (aarch64_simd_scalar_immediate_valid_for_move): New.
16571         * config/aarch64/aarch64-protos.h
16572         (aarch64_simd_scalar_immediate_valid_for_move): New.
16573         * config/aarch64/constraints.md (Dh, Dq): New.
16574         * config/aarch64/iterators.md (hq): New.
16576 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
16578         * reorg.c (get_branch_condition): Deal with conditional returns.
16579         (fill_simple_delay_slots): Remove dead code dealing with jumps.
16581 2013-04-01  Wei Mi  <wmi@google.com>
16583         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
16584         Truncate operand 2 using %b asm operand modifier.
16585         (*<shift_insn><mode>3_mask): Ditto.
16586         (*<rotate_insn><mode>3_mask): Ditto.
16588 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
16590         PR middle-end/56798
16591         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
16593 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
16595         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
16596         of next_real_insn.
16597         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
16599 2013-03-30  Lawrence Crowl  <crowl@google.com>
16601         * dse.c (clear_alias_sets): Remove never set.
16602         (disqualified_clear_alias_sets): Remove never set.
16603         (clear_alias_mode_pool): Remove never set.
16604         (dse_step0): Remove condition that is never true.
16605         (canon_address): Remove condition that is never true.
16606         (dse_step7): Remove condition that is never true.
16607         (rest_of_handle_dse): Remove condition that is never true.
16608         (rest_of_handle_dse::did_global): Remove never read from above.
16609         (dse_step2_spill): Remove never called from above.
16610         (dse_step5_spill): Remove never called from above.
16612 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
16614         * doc/md.texi (Standard Names) <casesi>: Update documentation for
16615         JUMP_TABLE_DATA changes.
16616         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
16617         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
16618         (Insns) <jump_table_data>: New entry.
16619         * doc/tm.texi: Regenerate.
16621         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
16623         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
16624         for table jump at the end of a basic block using tablejump_p.
16625         * targhooks.c (default_invalid_within_doloop): Likewise.
16626         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
16627         target hook implementation that is identical to the default hook.
16628         (rs6000_invalid_within_doloop): Remove.
16630         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
16631         unused variable from tablejump_p call.
16633         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
16634         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
16635         (INSN_DELETED_P): Likewise.
16636         (emit_jump_table_data): New prototype.
16637         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
16638         after 4th as unused.
16639         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
16640         * sched-vis.c (print_insn): Likewise.
16641         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
16642         insn for compatibility with back ends that use next_active_insn to
16643         identify jump table data.
16644         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
16645         (remove_insn): Likewise.
16646         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
16647         to be emitted.
16648         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
16649         (emit_jump_table_data): New function.
16651         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
16652         basic block, a JUMP_TABLE_DATA never is.
16653         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
16654         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
16655         off from code handling real insns.
16656         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
16657         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
16658         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
16659         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
16660         is not a NONDEBUG_INSN_P.
16661         * ira-costs.c (scan_one_insn): Likewise.
16662         * jump.c (mark_all_labels): Likewise.
16663         (mark_jump_label_1): Likewise.
16664         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
16665         * lra.c (get_insn_freq): Expect all insns reaching here to be in
16666         a basic block.
16667         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
16668         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
16669         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
16670         JUMP_TABLE_DATA_P insns.
16671         (calculate_elim_costs_all_insns): Likewise.
16672         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
16673         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
16674         (delete_output_reload): Code style fixups.
16675         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
16676         insn flags on this non-insn.
16677         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
16678         as scheduling barriers, for pre-change compatibility.
16679         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
16680         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
16682         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
16683         redundant JUMP_TABLE_DATA_P test.
16684         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
16685         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
16686         (frv_for_each_packet): Likewise.
16687         * config/i386/i386.c (min_insn_size): Likewise.
16688         (ix86_avoid_jump_mispredicts): Likewise.
16689         * config/m32r/m32r.c (m32r_is_insn): Likewise.
16690         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
16691         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
16692         (mips16_insn_length): Robustify.
16693         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
16694         (mips16_split_long_branches): Likewise.
16695         * config/pa/pa.c (pa_combine_instructions): Likewise.
16696         * config/rs6000/rs6000.c (get_next_active_insn): Treat
16697         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
16698         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
16699         as contributing to pool range lengths.
16700         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
16701         Remove redundant JUMP_TABLE_DATA_P test.
16702         (sh_loop_align): Likewise.
16703         (split_branches): Likewise.
16704         (sh_insn_length_adjustment): Likewise.
16705         * config/spu/spu.c (get_branch_target): Likewise.
16707 2013-03-29  Jan Hubicka  <jh@suse.cz>
16709         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
16710         gcov streaming; stream hot bb threshold to ltrans.
16711         * predict.c (get_hot_bb_threshold): Break out from ....
16712         (maybe_hot_count_p): ... here.
16713         (set_hot_bb_threshold): New function.
16714         * lto-section-in.c (lto_section_name): Add profile.
16715         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
16716         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
16717         and data-streamer.h
16718         (histogram_entry): New structure.
16719         (histogram, histogram_pool): New global vars.
16720         (histogram_hash): New structure.
16721         (histogram_hash::hash): New method.
16722         (histogram_hash::equal): Likewise.
16723         (account_time_size): New function.
16724         (cmp_counts): New function.
16725         (dump_histogram): New function.
16726         (ipa_profile_generate_summary): New function.
16727         (ipa_profile_write_summary): New function.
16728         (ipa_profile_read_summary): New function.
16729         (ipa_profile): Decide on threshold.
16730         (pass_ipa_profile): Add ipa_profile_write_summary and
16731         ipa_profile_read_summary.
16732         * Makefile.in (ipa.o): Update dependencies.
16733         * lto-streamer.h (LTO_section_ipa_profile): New section.
16735 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
16737         * tree.h (VAR_P): New.
16739 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
16741         PR lto/56777
16742         * doc/invoke.texi ([-fwhole-program]): Fix typo.
16744 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
16746         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
16747         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
16748         (control_flow_insn_p): Likewise.
16749         * cfgrtl.c (duplicate_insn_chain): Likewise.
16750         * final.c (get_attr_length_1): Likewise.
16751         (shorten_branches): Likewise.
16752         (final_scan_insn): Likewise.
16753         * function.c (instantiate_virtual_regs): Likewise.
16754         * gcse.c (insert_insn_end_basic_block): Likewise.
16755         * ira-costs.c (scan_one_insn): Likewise.
16756         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
16757         * lra.c (check_rtl): Likewise.
16758         * reload1.c (elimination_costs_in_insn): Likewise.
16759         * reorg.c (follow_jumps): Likewise.
16761         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
16762         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
16763         (thumb_far_jump_used_p): Likewise.
16764         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
16765         (workaround_speculation): Likewise.
16766         (add_sched_insns_for_speculation): Likewise.
16767         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
16768         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
16769         (frv_for_each_packet): Likewise.
16770         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
16771         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
16772         (final_emit_insn_group_barriers): Likewise.
16773         * config/m32r/m32r.c (m32r_is_insn): Likewise.
16774         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
16775         (mips16_insn_length): Likewise.
16776         * config/pa/pa.c (pa_reorg): Likewise.
16777         (pa_combine_instructions): Likewise.
16778         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
16779         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
16780         (sh_reorg): Likewise.
16781         (split_branches): Likewise.
16782         * config/spu/spu.c (get_branch_target): Likewise.
16784         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
16785         JUMP_TABLE_DATA_P.
16787 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
16789         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
16790         Fix declaration name.
16792 2013-03-28  Lawrence Crowl  <crowl@google.com>
16794         * graphds.h (struct graph.indicies): Remove unused.
16795         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
16796         (SCOP_ORIGINAL_PDDRS): Remove unused.
16797         * sese.h (extern insert_loop_close_phis): Removed unused.
16798         (extern insert_guard_phis): Removed unused.
16799         (extern ivtype_map_elt_info): Removed unused.
16800         (new_ivtype_map_elt): Removed unused.
16801         * sese.c (ivtype_map_elt_info): Removed unused.
16803 2013-03-28  Lawrence Crowl  <crowl@google.com>
16805         * Makefile.in: Add several missing include dependences.
16806         (DUMPFILE_H): New.
16807         (test-dump.o): New.  This object is not added to any executable,
16808         but is present for ad-hoc testing.
16809         * bitmap.c
16810         (debug (const bitmap_head_def &)): New.
16811         (debug (const bitmap_head_def *)): New.
16812         * bitmap.h
16813         (extern debug (const bitmap_head_def &)): New.
16814         (extern debug (const bitmap_head_def *)): New.
16815         * cfg.c
16816         (debug (edge_def &)): New.
16817         (debug (edge_def *)): New.
16818         * cfghooks.c
16819         (debug (basic_block_def &)): New.
16820         (debug (basic_block_def *)): New.
16821         * dumpfile.h
16822         (dump_node (const_tree, int, FILE *)): Correct source file.
16823         * dwarf2out.c
16824         (debug (die_struct &)): New.
16825         (debug (die_struct *)): New.
16826         * dwarf2out.h
16827         (extern debug (die_struct &)): New.
16828         (extern debug (die_struct *)): New.
16829         * gimple-pretty-print.c
16830         (debug (gimple_statement_d &)): New.
16831         (debug (gimple_statement_d *)): New.
16832         * gimple-pretty-print.h
16833         (extern debug (gimple_statement_d &)): New.
16834         (extern debug (gimple_statement_d *)): New.
16835         * ira-build.c
16836         (debug (ira_allocno_copy &)): New.
16837         (debug (ira_allocno_copy *)): New.
16838         (debug (ira_allocno &)): New.
16839         (debug (ira_allocno *)): New.
16840         * ira-int.h
16841         (extern debug (ira_allocno_copy &)): New.
16842         (extern debug (ira_allocno_copy *)): New.
16843         (extern debug (ira_allocno &)): New.
16844         (extern debug (ira_allocno *)): New.
16845         * ira-lives.c
16846         (debug (live_range &)): New.
16847         (debug (live_range *)): New.
16848         * lra-int.h
16849         (debug (lra_live_range &)): New.
16850         (debug (lra_live_range *)): New.
16851         * lra-lives.c
16852         (debug (lra_live_range &)): New.
16853         (debug (lra_live_range *)): New.
16854         * omega.c
16855         (debug (omega_pb_d &)): New.
16856         (debug (omega_pb_d *)): New.
16857         * omega.h
16858         (extern debug (omega_pb_d &)): New.
16859         (extern debug (omega_pb_d *)): New.
16860         * print-rtl.c
16861         (debug (const rtx_def &)): New.
16862         (debug (const rtx_def *)): New.
16863         * print-tree.c
16864         (debug_tree (tree): Move within file.
16865         (debug_raw (const tree_node &)): New.
16866         (debug_raw (const tree_node *)): New.
16867         (dump_tree_via_hooks (const tree_node *, int)): New.
16868         (debug (const tree_node &)): New.
16869         (debug (const tree_node *)): New.
16870         (debug_verbose (const tree_node &)): New.
16871         (debug_verbose (const tree_node *)): New.
16872         (debug_head (const tree_node &)): New.
16873         (debug_head (const tree_node *)): New.
16874         (debug_body (const tree_node &)): New.
16875         (debug_body (const tree_node *)): New.
16876         (debug_vec_tree (tree): Move and reimplement in terms of dump.
16877         (debug (vec<tree, va_gc> &)): New.
16878         (debug (vec<tree, va_gc> *)): New.
16879         * rtl.h
16880         (extern debug (const rtx_def &)): New.
16881         (extern debug (const rtx_def *)): New.
16882         * sbitmap.c
16883         (debug_raw (simple_bitmap_def &)): New.
16884         (debug_raw (simple_bitmap_def *)): New.
16885         (debug (simple_bitmap_def &)): New.
16886         (debug (simple_bitmap_def *)): New.
16887         * sbitmap.h
16888         (extern debug (simple_bitmap_def &)): New.
16889         (extern debug (simple_bitmap_def *)): New.
16890         (extern debug_raw (simple_bitmap_def &)): New.
16891         (extern debug_raw (simple_bitmap_def *)): New.
16892         * sel-sched-dump.c
16893         (debug (vinsn_def &)): New.
16894         (debug (vinsn_def *)): New.
16895         (debug_verbose (vinsn_def &)): New.
16896         (debug_verbose (vinsn_def *)): New.
16897         (debug (expr_def &)): New.
16898         (debug (expr_def *)): New.
16899         (debug_verbose (expr_def &)): New.
16900         (debug_verbose (expr_def *)): New.
16901         (debug (vec<rtx> &)): New.
16902         (debug (vec<rtx> *)): New.
16903         * sel-sched-dump.h
16904         (extern debug (vinsn_def &)): New.
16905         (extern debug (vinsn_def *)): New.
16906         (extern debug_verbose (vinsn_def &)): New.
16907         (extern debug_verbose (vinsn_def *)): New.
16908         (extern debug (expr_def &)): New.
16909         (extern debug (expr_def *)): New.
16910         (extern debug_verbose (expr_def &)): New.
16911         (extern debug_verbose (expr_def *)): New.
16912         (extern debug (vec<rtx> &)): New.
16913         (extern debug (vec<rtx> *)): New.
16914         * sel-sched-ir.h
16915         (_list_iter_cond_expr): Make inline instead of static.
16916         * sreal.c
16917         (debug (sreal &)): New.
16918         (debug (sreal *)): New.
16919         * sreal.h
16920         (extern debug (sreal &)): New.
16921         (extern debug (sreal *)): New.
16922         * tree.h
16923         (extern debug_raw (const tree_node &)): New.
16924         (extern debug_raw (const tree_node *)): New.
16925         (extern debug (const tree_node &)): New.
16926         (extern debug (const tree_node *)): New.
16927         (extern debug_verbose (const tree_node &)): New.
16928         (extern debug_verbose (const tree_node *)): New.
16929         (extern debug_head (const tree_node &)): New.
16930         (extern debug_head (const tree_node *)): New.
16931         (extern debug_body (const tree_node &)): New.
16932         (extern debug_body (const tree_node *)): New.
16933         (extern debug (vec<tree, va_gc> &)): New.
16934         (extern debug (vec<tree, va_gc> *)): New.
16935         * tree-cfg.c
16936         (debug (struct loop &)): New.
16937         (debug (struct loop *)): New.
16938         (debug_verbose (struct loop &)): New.
16939         (debug_verbose (struct loop *)): New.
16940         * tree-dump.c: Add header dependence.
16941         * tree-flow.h
16942         (extern debug (struct loop &)): New.
16943         (extern debug (struct loop *)): New.
16944         (extern debug_verbose (struct loop &)): New.
16945         (extern debug_verbose (struct loop *)): New.
16946         * tree-data-ref.c
16947         (debug (data_reference &)): New.
16948         (debug (data_reference *)): New.
16949         (debug (vec<data_reference_p> &)): New.
16950         (debug (vec<data_reference_p> *)): New.
16951         (debug (vec<ddr_p> &)): New.
16952         (debug (vec<ddr_p> *)): New.
16953         * tree-data-ref.h
16954         (extern debug (data_reference &)): New.
16955         (extern debug (data_reference *)): New.
16956         (extern debug (vec<data_reference_p> &)): New.
16957         (extern debug (vec<data_reference_p> *)): New.
16958         (extern debug (vec<ddr_p> &)): New.
16959         (extern debug (vec<ddr_p> *)): New.
16960         * tree-ssa-alias.c
16961         (debug (pt_solution &)): New.
16962         (debug (pt_solution *)): New.
16963         * tree-ssa-alias.h
16964         (extern debug (pt_solution &)): New.
16965         (extern debug (pt_solution *)): New.
16966         * tree-ssa-alias.c
16967         (debug (_var_map &)): New.
16968         (debug (_var_map *)): New.
16969         (debug (tree_live_info_d &)): New.
16970         (debug (tree_live_info_d *)): New.
16971         * tree-ssa-alias.h
16972         (extern debug (_var_map &)): New.
16973         (extern debug (_var_map *)): New.
16974         (extern debug (tree_live_info_d &)): New.
16975         (extern debug (tree_live_info_d *)): New.
16977 2013-03-28  Jan Hubicka  <jh@suse.cz>
16979         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
16981 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
16983         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
16984         record only when desired or required.
16986 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
16988         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
16989         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
16991 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
16993         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
16994         (*andsi3_compare0_uxtw): New pattern.
16995         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
16996         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
16998 2013-03-28  Jan Hubicka  <jh@suse.cz>
17000         * data-streamer-in.c (streamer_read_gcov_count): New function.
17001         * gimple-streamer-out.c: Include value-prof.h.
17002         (output_gimple_stmt): Output histogram.
17003         (output_bb): Use streamer_write_gcov_count.
17004         * value-prof.c: Include data-streamer.h
17005         (dump_histogram_value): Add HIST_TYPE_MAX.
17006         (stream_out_histogram_value): New function.
17007         (stream_in_histogram_value): New function.
17008         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
17009         (stream_out_histogram_value, stream_in_histogram_value): Declare.
17010         * data-streamer-out.c (streamer_write_gcov_count): New function.
17011         (streamer_write_gcov_count_stream): New function.
17012         * lto-cgraph.c (lto_output_edge): Update counter streaming.
17013         (lto_output_node): Likewise.
17014         (input_node, input_edge): Likewise.
17015         * lto-streamer-out.c (output_cfg): Update streaming.
17016         * lto-streamer-in.c (input_cfg): Likewise.
17017         * data-streamer.h (streamer_write_gcov_count,
17018         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
17019         * gimple-streamer-in.c: Include value-prof.h
17020         (input_gimple_stmt): Input histograms.
17021         (input_bb): Update profile streaming.
17023 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
17025         * genmodes.c (emit_max_int): New function.
17026         (emit_insn_modes_h): Added call to emit_max_function.
17027         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
17028         Added doc.
17029         * machmode.def: Fixed comment.
17031 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
17033         * combine.c (try_combine): Removed useless assert.
17034         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
17036 2013-03-28  Marek Polacek  <polacek@redhat.com>
17037             Richard Biener  <rguenther@suse.de>
17039         PR tree-optimization/56695
17040         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
17041         build signed result of a vector comparison.
17042         * tree-cfg.c (verify_gimple_comparison): Check that a result
17043         of a vector comparison has signed type.
17045 2013-03-28  Richard Biener  <rguenther@suse.de>
17047         PR tree-optimization/37021
17048         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
17049         do not restrict gaps between groups.
17050         * tree-vect-stmts.c (vectorizable_load): Properly account for
17051         a gap between groups.
17053 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
17055         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
17056         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
17057         is not enabled.
17059 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
17061         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
17062         * doc/extend.texi (Named Address Spaces): Ditto.
17063         (Variable Attributes): Ditto.
17065 2013-03-27  Kai Tietz  <ktietz@redhat.com>
17067         * config.build: Add support for cygwin x64 target.
17068         * config.gcc: Likewise.
17069         * config.host: Likewise.
17070         * configure.ac: Likewise
17071         * configure: Regenerated.
17073 2013-03-27  Kai Tietz  <ktietz@redhat.com>
17075         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
17076         * config/i386/t-cygwin-w64: New file.
17077         * config/i386/cygwin-w64.h: New file.
17078         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
17079         and add support for x64-cygwin target.
17080         (CPP_SPEC): Likewise.
17081         (CXX_WRAP_SPEC_LIST): Undefine before define.
17082         (LIBGCJ_SONAME): Use 15 as version.
17084 2013-03-27  Richard Biener  <rguenther@suse.de>
17086         PR tree-optimization/56716
17087         * tree-ssa-structalias.c (perform_var_substitution): Adjust
17088         dumping for ref nodes.
17090 2013-03-27  Martin Jambor  <mjambor@suse.cz>
17092         PR tree-optimization/55334
17093         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
17094         restricted pointers to arrays.
17096 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
17098         * Makefile.in (.SUFFIXES): Add .cc.
17099         (.c.o): Apply same recipe for implicit rule .cc.o.
17101 2013-03-27  Richard Biener  <rguenther@suse.de>
17103         PR tree-optimization/37021
17104         * tree-vect-data-refs.c (vect_check_strided_load): Allow
17105         REALPART/IMAGPART_EXPRs around the supported refs.
17106         * tree-ssa-structalias.c (find_func_aliases): Assume that
17107         floating-point values are not used to transfer pointers.
17109 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
17111         * target.def (TARGET_HAS_IFUNC_P): New target hook.
17112         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
17113         * doc/tm.texi: Regenerate.
17114         * targhooks.h (default_has_ifunc_p): New.
17115         * targhooks.c (default_has_ifunc_p): Ditto.
17116         * config/linux-protos.h: New file.
17117         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
17118         hook for linux which disables support of indirect functions in android.
17119         * config/linux-android.c: New file.
17120         * config/t-linux-android.c: Ditto.
17121         * config.gcc: Added new object file linux-android.o.
17122         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
17123         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
17124         * varasm.c (do_assemble_alias): Likewise.
17125         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
17126         doesn't support indirect functions.
17127         * configure: Regenerate.
17129 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
17131         PR target/56102
17132         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
17133         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
17134         mult-word mode.
17136 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17138         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
17140 2013-03-27  Terry Guo  <terry.guo@arm.com>
17142         * config/arm/arm-cores.def: Added core cortex-r7.
17143         * config/arm/arm-tune.md: Regenerated.
17144         * config/arm/arm-tables.opt: Regenerated.
17145         * doc/invoke.texi: Added entry for core cortex-r7.
17147 2013-03-27  Walter Lee  <walt@tilera.com>
17149         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
17150         double-decrement of next_scratch_regno.
17152 2013-03-27  Walter Lee  <walt@tilera.com>
17154         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
17155         input operands.
17156         (insn_v1mulus): Ditto.
17157         (insn_v2muls): Ditto.
17159 2013-03-27  Walter Lee  <walt@tilera.com>
17161         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
17162         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
17164 2013-03-27  Walter Lee  <walt@tilera.com>
17166         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
17167         (*sibcall_value): Ditto.
17169 2013-03-27  Walter Lee  <walt@tilera.com>
17171         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
17172         (insn_mnz_v8qi): ... this ...
17173         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
17174         vector equivalent.
17175         (insn_v<n>mnz): Replaced by ...
17176         (insn_v1mnz): ... this ...
17177         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
17178         equivalent.
17179         (insn_mz_<mode>): Replaced by ...
17180         (insn_mz_v8qi): ... this ...
17181         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
17182         vector equivalent.
17183         (insn_v<n>mz): Replaced by ...
17184         (insn_v1mz): ... this ...
17185         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
17186         equivalent.
17188 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
17190         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
17192 2013-03-26  Roland McGrath  <mcgrathr@google.com>
17194         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
17195         than fprintf with a non-constant, non-format string.
17197 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
17199         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
17200         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
17201         operand 0 predicate.
17202         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
17203         attribute.  Use general_x64nomem_operand as operand 1 predicate.
17204         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
17205         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
17206         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
17207         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
17208         (mov<mode>_insv_1): Remove expander.  Merge insn with
17209         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
17210         Use general_x64nomem_operand as operand 1 predicate.
17211         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
17212         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
17213         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
17214         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
17215         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
17216         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
17217         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
17218         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
17219         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
17220         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
17221         (general_x64nomem_operand): Ditto.
17223 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
17225         * config/rtems.opt: Add -pthread option.
17227 2013-03-26  Richard Biener  <rguenther@suse.de>
17229         * alias.c (find_base_term): Avoid redundant and not used recursion.
17230         (base_alias_check): Get the initial base term from the caller.
17231         (true_dependence_1): Compute and pass base terms to base_alias_check.
17232         (write_dependence_p): Likewise.
17233         (may_alias_p): Likewise.
17235 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
17237         * config/aarch64/aarch64.c (aarch64_classify_address): Support
17238         PC-relative load in SI modes and above only.
17240 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
17242         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
17243         * config/arm/iwmmxt.md (WCGR0): Update.
17244         (WCGR1, WCGR2, WCGR3): Likewise.
17246 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
17248         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
17249         Use x64 and nox64 isa attributes.
17251 2013-03-26  Richard Biener  <rguenther@suse.de>
17253         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
17254         alignment computations and rely on get_object_alignment_1
17255         for the !TYPE_P case.
17256         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
17258 2013-03-26  Walter Lee  <walt@tilera.com>
17260         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
17261         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
17263 2013-03-25  Jeff Law  <law@redhat.com>
17265         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
17266         check for INTEGRAL_TYPE_P that was missing due to checking in
17267         wrong version of prior patch.
17269 2013-03-25  Walter Lee  <walt@tilera.com>
17271         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
17272         TILEGX_INSN_SHUFFLEBYTES1.
17273         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
17274         shufflebytes1.
17275         (tilegx_builtins): Ditto.
17276         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
17278 2013-03-25  Walter Lee  <walt@tilera.com>
17280         * config/tilegx/tilegx.md (floatsisf2): New pattern.
17281         (floatunssisf2): New pattern.
17282         (floatsidf2): New pattern.
17283         (floatunssidf2): New pattern.
17285 2013-03-25  Walter Lee  <walt@tilera.com>
17287         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
17288         tests for constraint J, K, N, P.
17290 2013-03-25  Walter Lee  <walt@tilera.com>
17292         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
17293         Use indirect/pcrel encoding.
17294         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
17295         Ditto.
17297 2013-03-25  Steve Ellcey  <sellcey@mips.com>
17299         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
17300         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
17301         * config/mips/mips.c (mips_option_override): Set IMADD default.
17302         * config/mips/mips.h (PTF_AVOID_IMADD): New.
17303         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
17304         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
17305         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
17306         * doc/invoke.texi (-mimadd/-mno-imadd): New.
17308 2013-03-25  Jeff Law  <law@redhat.com>
17310         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
17311         slightly to avoid creating and folding useless trees.  Simplify
17312         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
17314 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
17316         * config/i386/i386.md (*zero_extendsidi2): Merge with
17317         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
17318         * config/i386/predicates.md (x86_64_zext_operand): Rename from
17319         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
17320         targets.  Clarify comment.
17322 2013-03-25  Martin Jambor  <mjambor@suse.cz>
17324         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
17325         pass-through jump functions differently.
17326         (ipa_read_jump_function): Likewise.  Also use setter functions to set
17327         up jump functions.
17329 2013-03-25  Martin Jambor  <mjambor@suse.cz>
17331         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
17332         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
17333         process it.
17334         (ipa_get_indirect_edge_target): New function.
17335         (devirtualization_time_bonus): New parameter known_aggs, pass it to
17336         ipa_get_indirect_edge_target.  Update all callers.
17337         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
17338         ipa_get_indirect_edge_target_1 instead of calling
17339         ipa_get_indirect_edge_target.
17340         (create_specialized_node): Pass aggvlas to
17341         ipcp_discover_new_direct_edges.
17343 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17345         * config/arm/arm.md (f_sels, f_seld): New types.
17346         (*cmov<mode>): New pattern.
17347         * config/arm/predicates.md (arm_vsel_comparison_operator): New
17348         predicate.
17350 2013-03-25  Kai Tietz  <ktietz@redhat.com>
17352         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
17353         POSIX-printf for mingw-hosted  builds.
17355 2013-03-25  Richard Biener  <rguenther@suse.de>
17357         PR middle-end/56694
17358         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
17359         must-not-throw stmt location.
17361 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17363         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
17364         Emit load-acquire versions when acq is true.
17365         (arm_emit_store_exclusive): Add rel parameter.
17366         Emit store-release versions when rel is true.
17367         (arm_split_compare_and_swap): Use acquire-release instructions
17368         instead.
17369         of barriers when appropriate.
17370         (arm_split_atomic_op): Likewise.
17371         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
17372         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
17373         (VUNSPEC_SLX): Likewise.
17374         (VUNSPEC_LDA): Likewise.
17375         (VUNSPEC_STL): Likewise.
17376         * config/arm/sync.md (atomic_load<mode>): New pattern.
17377         (atomic_store<mode>): Likewise.
17378         (arm_load_acquire_exclusive<mode>): Likewise.
17379         (arm_load_acquire_exclusivesi): Likewise.
17380         (arm_load_acquire_exclusivedi): Likewise.
17381         (arm_store_release_exclusive<mode>): Likewise.
17383 2013-03-25  Catherine Moore  <clm@codesourcery.com>
17385         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
17386         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
17387         * config/mip/predicates.md (lwsp_swsp_operand,
17388         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
17389         sb16_operand, db4_operand, db7_operand, ib3_operand,
17390         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
17391         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
17392         andi16_operand): New predicates.
17393         * config/mips/mips.md (compression): New attribute.
17394         (enabled): New attribute.
17395         (length): Consider compression in computing length.
17396         (shift_compression): New code attribute.
17397         (*add<mode>3): New operands. Record compression.
17398         (sub<mode>3): Likewise.
17399         (one_cmpl<mode>2): Likewise.
17400         (*and<mode>3): Likewise.
17401         (*ior<mode>3): Likewise.
17402         (unnamed pattern for xor): Likewise.
17403         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
17404         (*<optab><mode>3): Likewise.
17405         (*mov<mode>_internal: Likewise.
17406         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
17407         (mips_unsigned_immediate_p): New.
17408         (umips_lwsp_swsp_address_p): New.
17409         (m16_based_address_p): New.
17410         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
17411         (mips_unsigned_immediate_p): New prototype.
17412         (lwsp_swsp_address_p): New prototype.
17413         (m16_based_address_p): New prototype.
17414         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
17415         (mips_signed_immediate_p): New function.
17416         (m16_based_address_p): New function.
17417         (lwsp_swsp_address_p): New function.
17418         (mips_print_operand_punctuation): Recognize short delay slot insns
17419         for microMIPS.add<mode>3"
17421 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17423         PR target/56720
17424         * config/arm/iterators.md (v_cmp_result): New mode attribute.
17425         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
17427 2013-03-25  Richard Biener  <rguenther@suse.de>
17429         PR tree-optimization/56689
17430         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
17431         any edge.
17433 2013-03-25  Richard Biener  <rguenther@suse.de>
17435         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
17436         of bitmap.
17437         (memory_references): Likewise.
17438         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
17439         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
17440         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
17441         (gather_mem_refs_in_loops): Fold into ...
17442         (analyze_memory_references): ... this.  Move initialization
17443         to tree_ssa_lim_initialize.
17444         (fill_always_executed_in): Rename to ...
17445         (fill_always_executed_in_1): ... this.
17446         (fill_always_executed_in): Move contains_call computation to
17447         this new function from ...
17448         (tree_ssa_lim_initialize): ... here.
17449         (tree_ssa_lim): Call fill_always_executed_in.
17451 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
17453         * postreload.c (reload_combine): Fix code detecting returns.
17455 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
17457         * function.c (emit_use_return_register_into_block): On cc0 targets,
17458         do not emit the sequence between cc0 setter and user.
17460 2013-03-25  Kai Tietz  <ktietz@redhat.com>
17462         * config/i386/predicates.md (local_symbolic_operand): Interpret
17463         dll-imported symbols as none-local.
17465 2013-03-25  Richard Biener  <rguenther@suse.de>
17467         * tree-ssa-loop-im.c (struct depend): Remove.
17468         (struct lim_aux_data): Make depends a vec of gimples.
17469         (free_lim_aux_data): Adjust.
17470         (add_dependency): Likewise.
17471         (set_level): Likewise.
17473 2013-03-25  Richard Biener  <rguenther@suse.de>
17475         PR middle-end/56434
17476         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
17477         the pointer returned by calls with ECF_MALLOC set.
17479 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
17481         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
17483 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
17485         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
17486         using MMXMODE mode iterator.
17487         (*move<mode>_internal): Merge with *movv2sf_internal and
17488         *movv2sf_internal_rex64 using MMXMODE mode iterator.
17490 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
17492         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
17493         (record_last_mem_set_info): Likewise.
17495         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
17496         of XNEWVEC followed by memset.
17497         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
17499 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
17501         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
17502         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
17503         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
17504         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
17505         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
17506         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
17507         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
17508         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
17509         BARRIER_P instead of GET_CODE.
17511 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
17513         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
17514         inaccuracy in the probing code.
17516         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
17517         (ctrapdi4): Likewise.
17519 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
17521         * calls.c (expand_call): Add missing guard to code handling return
17522         of non-BLKmode structures in MSB.
17523         * function.c (expand_function_end): Likewise.
17525 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
17527         * combine.c (try_combine): Adjust comment.  Do not add the set of
17528         insn #0 if the destination indirectly is set or dies in insn #2.
17529         Tidy up code to distribute a new note.
17531 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
17533         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
17534         also for alternatives 16 and 17.
17536 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
17538         * config/i386/sse.md (*mov<mode>_internal): Merge with
17539         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
17540         Emit insn template depending on type attribute.  Use
17541         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
17542         movd instead of movq mnemonic for interunit moves.  Rewrite mode
17543         attribute calculation.  Remove unit attribute calculation.
17544         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
17545         Set prefix_data16 attribute for DImode ssemov types.
17546         Use Ym instead of y for SSE-MMX conversion alternatives.
17547         Reorder operand constraints.
17549 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
17551         * df.h (df_insn_delete): Adjust prototype.
17552         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
17553         and let it decide whether mark the basic block dirty.
17554         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
17555         * df-scan.c (df_insn_info_delete): New helper function, split
17556         off from df_insn_delete.
17557         (df_scan_free_bb_info): Use it.
17558         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
17559         Likewise.
17560         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
17561         that the insn is actually an insn and it has a non-NULL basic block.
17562         Do not mark basic block dirty if only deleting a DEBUG_INSN.
17564 2013-03-22  Richard Biener  <rguenther@suse.de>
17566         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
17567         dep_ref members.
17568         (mem_ref_alloc): Do not allocate them.
17569         (refs_independent_p): Do not query or maintain a cache.
17571 2013-03-22  Richard Biener  <rguenther@suse.de>
17573         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
17574         (gather_mem_refs_in_loops): Do not compute it.
17575         (analyze_memory_references): Do not allocate it.
17576         (tree_ssa_lim_finalize): Do not free it.
17577         (for_all_locs_in_loop): Do not query all_refs_in_loop.
17579 2013-03-22  Richard Biener  <rguenther@suse.de>
17581         * is-a.h (as_a): Use gcc_checking_assert.
17583 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
17585         * config/aarch64/aarch64.c (aarch64_print_operand): New
17586         format specifier for printing a constant in hex.
17587         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
17588         format specifier for printing second operand.
17590 2013-03-22  Richard Biener  <rguenther@suse.de>
17592         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
17593         bitmaps.
17594         (gather_mem_refs_in_loops): Perform store accumulation here.
17595         (create_vop_ref_mapping_loop): Remove.
17596         (create_vop_ref_mapping): Likewise.
17597         (analyze_memory_references): Initialize refs_stored_in_loop.
17598         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
17599         (record_indep_loop): Remove.
17600         (record_dep_loop): New function.
17601         (ref_indep_loop_p_1): Adjust to only walk over references
17602         in the loop, not its subloops.
17603         (ref_indep_loop_p): Rename to ...
17604         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
17605         maintaining a more fine-grained cache.
17606         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
17607         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
17609 2013-03-22  Richard Biener  <rguenther@suse.de>
17611         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
17612         (struct mem_ref): Make accesses_in_loop a vec of a vec of
17613         aggregate mem_ref_loc.
17614         (free_mem_ref_locs): Inline into ...
17615         (memref_free): ... this and adjust.
17616         (mem_ref_alloc): Adjust.
17617         (mem_ref_locs_alloc): Remove.
17618         (record_mem_ref_loc): Adjust.
17619         (get_all_locs_in_loop): Rewrite into ...
17620         (for_all_locs_in_loop): ... this iterator.
17621         (rewrite_mem_ref_loc): New functor.
17622         (rewrite_mem_refs): Use for_all_locs_in_loop.
17623         (sm_set_flag_if_changed): New functor.
17624         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
17625         (ref_always_accessed): New functor.
17626         (ref_always_accessed_p): Use for_all_locs_in_loop.
17628 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
17630         * tree-pass.h (PROP_gimple_lvec): New.
17631         * passes.c (dump_properties): Handle PROP_gimple_lvec.
17632         (init_optimization_passes): Move pass_lower_vector.
17633         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
17634         PROP_gimple_lvec.
17635         (pass_lower_vector): Provide PROP_gimple_lvec.
17636         (pass_lower_vector_ssa): Likewise.
17637         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
17639 2013-03-21  Mark Wielaard  <mjw@redhat.com>
17641         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
17643 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
17645         * config/i386/i386.md (*movdi_internal): Disparage slightly
17646         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
17647         conversion alternatives.
17649 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
17651         PR middle-end/48087
17652         * diagnostic.def (DK_WERROR): New kind.
17653         * diagnostic.h (werrorcount): Define.
17654         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
17655         promoted to DK_ERROR, increment DK_WERROR counter instead of
17656         DK_ERROR counter.
17657         * toplev.c (toplev_main): Call print_ignored_options even if
17658         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
17659         even if just werrorcount is non-zero.
17661         PR debug/55608
17662         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
17663         on failure.
17664         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
17665         (string_cst_pool_decl): New function.
17666         (optimize_one_addr_into_implicit_ptr): New function.
17667         (resolve_addr_in_expr): Optimize DWARF location expression
17668         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
17669         which doesn't live in memory, but has DW_AT_location or
17670         DW_AT_const_value, or refers to a string literal, into
17671         DW_OP_GNU_implicit_pointer.
17672         (optimize_location_into_implicit_ptr): New function.
17673         (resolve_addr): If removing DW_AT_location of a variable because
17674         it was DW_OP_addr of address of the variable, but the variable doesn't
17675         live in memory, try to emit const value attribute for the initializer.
17677 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
17679         * tree.h (VECTOR_TYPE_P): New macro.
17680         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
17681         TYPE_MODE): Use it.
17682         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
17683         VEC_COND_EXPR cannot be lvalues.
17684         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
17686 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
17688         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
17689         Restrict the transformation to equal modes.
17691 2013-03-21  Richard Biener  <rguenther@suse.de>
17693         PR tree-optimization/39326
17694         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
17695         (MEM_ANALYZABLE): Adjust.
17696         (record_mem_ref_loc): Move bitmap ops ...
17697         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
17698         unanalyzable refs, do not record locations for it.
17699         (analyze_memory_references): Allocate ref zero as shared
17700         unanalyzable ref.
17701         (refs_independent_p): Do not test for unanalyzed mems here.
17702         (ref_indep_loop_p_1): Special-case disambiguation against
17703         the unanalyzed ref.
17704         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
17706 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
17708         * config/arm/arm-protos.h (tune_params): Add
17709         prefer_neon_for_64bits field.
17710         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
17711         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
17712         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
17713         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
17714         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
17715         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
17716         (arm_option_override): Handle -mneon-for-64bits new option.
17717         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
17718         (prefer_neon_for_64bits): Declare new variable.
17719         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
17720         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
17721         (arch_enabled): Handle new arch types. Remove support for onlya8
17722         and nota8.
17723         (one_cmpldi2): Use new arch names.
17724         (zero_extend<mode>di2, extend<mode>di2): Ditto.
17725         * config/arm/arm.opt (mneon-for-64bits): Add option.
17726         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
17727         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
17728         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
17729         of onlya8.
17730         * doc/invoke.texi (-mneon-for-64bits): Document.
17732 2013-03-21  Richard Biener  <rguenther@suse.de>
17734         PR tree-optimization/39326
17735         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
17736         (sort_bbs_in_loop_postorder_cmp): New function.
17737         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
17739 2013-03-21  Richard Biener  <rguenther@suse.de>
17741         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
17742         (vect_insert_into_interleaving_chain): Likewise.
17743         (vect_drs_dependent_in_basic_block): Inline ...
17744         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
17745         split out from ...
17746         (vect_analyze_data_ref_dependence): ... here.  Simplify.
17747         (vect_check_interleaving): Simplify.
17748         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
17749         (vect_slp_analyze_data_ref_dependences): ... this new function.
17750         (dr_group_sort_cmp): New function.
17751         (vect_analyze_data_ref_accesses): Compute data-reference groups
17752         here instead of in vect_analyze_data_ref_dependence.  Use
17753         a more efficient algorithm.
17754         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
17755         vect_slp_analyze_data_ref_dependences.  Call
17756         vect_analyze_data_ref_accesses earlier.
17757         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
17758         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
17759         (vect_slp_analyze_data_ref_dependences): New prototype.
17761 2013-03-21  Richard Biener  <rguenther@suse.de>
17763         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
17764         ref is stored in the loop.
17765         (find_refs_for_sm): Walk only over all stores.
17766         (store_motion_loop): Allocate from lim_bitmap_obstack.
17767         (store_motion): Likewise.
17769 2013-03-21  Richard Biener  <rguenther@suse.de>
17771         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
17772         Update virtual SSA form.
17774 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17776         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
17777         * configure: Regenerate.
17778         * config.in: Regenerate.
17779         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
17780         if !HAVE_LD_EH_FRAME_CIEV3.
17782 2013-03-21  Richard Biener  <rguenther@suse.de>
17784         * tree-cfg.c (verify_expr_no_block): New function.
17785         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
17786         nor DECL_VALUE_EXPR have locations with associated blocks.
17787         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
17788         (clear_unused_block_pointer): Remove code dealing with
17789         blocks in DECL_DEBUG_EXPR locations.
17791 2013-03-21  Richard Biener  <rguenther@suse.de>
17793         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
17794         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
17795         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
17796         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
17797         instead of DECL_DEBUG_EXPR_IS_FROM.
17798         * gimplify.c (gimplify_modify_expr): Likewise.
17799         * tree-cfg.c (verify_expr_location_1): Likewise.
17800         * tree-complex.c (create_one_component_var): Likewise.
17801         * tree-sra.c (create_access_replacement): Likewise.
17802         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
17803         (clear_unused_block_pointer): Likewise.
17804         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
17805         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
17806         * var-tracking.c (var_debug_decl): Likewise.
17807         (track_expr_p): Likewise.
17808         * tree-inline.c (add_local_variables): Likewise.  Set
17809         DECL_HAS_DEBUG_EXPR_P after copying it.
17810         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
17811         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
17813 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
17815         PR bootstrap/56656
17816         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
17817         * configure: Regenerate.
17818         * config.in: Regenerate.
17819         * config/i386/i386.md (*movdf_internal): Use
17820         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
17821         movd instead of movq mnemonic for interunit moves.
17822         (*movdi_internal): Ditto.
17824 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17826         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
17827         (abd<mode>_3): New pattern.
17828         (aba<mode>_3): New pattern.
17829         (fabd<mode>_3): New pattern.
17831 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17833         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
17834         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
17835         occurrence of REGISTER_PREFIX as its empty string.
17837 2013-03-20  Jeff Law  <law@redhat.com>
17839         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
17840         addititional equivalences for equality comparisons between an SSA_NAME
17841         and a constant where the SSA_NAME was set from a widening conversion.
17843 2013-03-20  Walter Lee  <walt@tilera.com>
17845         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
17847 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
17849         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
17850         depending on type attribute.
17851         (*movti_internal): Ditto.
17852         (*movtf_internal): Ditto.
17853         (*movxf_internal): Ditto.
17854         (*movdf_internal): Ditto.
17855         (*movsf_internal): Ditto.
17857 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
17859         * config/i386/i386.md (*movti_internal): Set prefix attribute to
17860         maybe_vex for sselog1 and ssemov types.
17861         (*movdi_internal): Reorder operand constraints.
17862         (*movsi_internal): Ditto.  Set prefix attribute to
17863         maybe_vex for sselog1 and ssemov types.
17864         (*movtf_internal): Set prefix attribute to maybe_vex
17865         for sselog1 and ssemov types.
17866         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
17867         DImode ssemov types.  Reorder operand constraints.
17868         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
17869         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
17870         attribute for SImode ssemov types.  Reorder operand constraints.
17872 2013-03-20  Martin Jambor  <mjambor@suse.cz>
17874         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
17875         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
17877 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
17879         * config/rs6000/predicates.md (indexed_address, update_address_mem
17880         update_indexed_address_mem): New predicates.
17881         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
17882         attribute for load/store instructions.
17883         * config/rs6000/dfp.md (movsd_store): Likewise.
17884         (movsd_load): Likewise.
17885         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
17886         (unnamed HI->DI extend define_insn): Likewise.
17887         (unnamed SI->DI extend define_insn): Likewise.
17888         (unnamed QI->SI extend define_insn): Likewise.
17889         (unnamed QI->HI extend define_insn): Likewise.
17890         (unnamed HI->SI extend define_insn): Likewise.
17891         (unnamed HI->SI extend define_insn): Likewise.
17892         (extendsfdf2_fpr): Likewise.
17893         (movsi_internal1): Likewise.
17894         (movsi_internal1_single): Likewise.
17895         (movhi_internal): Likewise.
17896         (movqi_internal): Likewise.
17897         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
17898         attribute for load/store instructions.
17899         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
17900         instructions.
17901         (mov<mode>_softfloat): Likewise.
17902         (mov<mode>_hardfloat32): Likewise.
17903         (mov<mode>_hardfloat64): Likewise.
17904         (mov<mode>_softfloat64): Likewise.
17905         (movdi_internal32): Likewise.
17906         (movdi_internal64): Likewise.
17907         (probe_stack_<mode>): Likewise.
17909 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
17911         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
17912         floating point, and decimal floating point to reload iterator.
17914         * config/rs6000/constraints.md (wl constraint): New constraints to
17915         return FLOAT_REGS if certain options are used to reduce the number
17916         of separate patterns that exist in the file.
17917         (wx constraint): Likewise.
17918         (wz constraint): Likewise.
17920         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
17921         -mdebug=reg, print wg, wl, wx, and wz constraints.
17922         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
17923         Initialize the reload functions for 64-bit binary/decimal floating
17924         point types.
17925         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
17926         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
17927         create the buffer on the stack to overcome not having a 32-bit
17928         load and store.
17929         (rs6000_emit_move): Likewise.
17930         (rs6000_secondary_memory_needed_rtx): Likewise.
17931         (rs6000_alloc_sdmode_stack_slot): Likewise.
17932         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
17933         via xxlxor, just like DFmode 0.0.
17935         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
17936         define as 1 if we are running on a power7 or newer.
17937         (enum r6000_reg_class_enum): Add new constraints.
17939         * config/rs6000/dfp.md (movsd): Delete, combine with binary
17940         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
17941         with other moves by using conditional constraits (wg).  Use LFIWZX
17942         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
17943         (movsd splitter): Likewise.
17944         (movsd_hardfloat): Likewise.
17945         (movsd_softfloat): Likewise.
17947         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
17948         binary and decimal floating point moves.
17949         (fmove_ok): New attributes to combine binary and decimal floating
17950         point moves, and to combine power6x (mfpgpr) moves along normal
17951         floating moves.
17952         (real_value_to_target): Likewise.
17953         (f32_lr): Likewise.
17954         (f32_lm): Likewise.
17955         (f32_li): Likewise.
17956         (f32_sr): Likewise.
17957         (f32_sm): Likewise.
17958         (f32_si): Likewise.
17959         (movsf): Combine binary and decimal floating point moves.  Combine
17960         power6x (mfpgpr) moves with other moves by using conditional
17961         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
17962         (mov<mode> for SFmode/SDmode); Likewise.
17963         (SFmode/SDmode splitters): Likewise.
17964         (movsf_hardfloat): Likewise.
17965         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
17966         (movsf_softfloat): Likewise.
17967         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
17969         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
17970         wx and wz constraints.
17972         * config/rs6000/constraints.md (wg constraint): New constraint to
17973         return FLOAT_REGS if -mmfpgpr (power6x) was used.
17975         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
17976         constraint.
17978         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
17979         -mdebug=reg, print wg, wl, wx, and wz constraints.
17980         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
17981         Initialize the reload functions for 64-bit binary/decimal floating
17982         point types.
17983         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
17984         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
17985         create the buffer on the stack to overcome not having a 32-bit
17986         load and store.
17987         (rs6000_emit_move): Likewise.
17988         (rs6000_secondary_memory_needed_rtx): Likewise.
17989         (rs6000_alloc_sdmode_stack_slot): Likewise.
17990         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
17991         via xxlxor, just like DFmode 0.0.
17993         * config/rs6000/dfp.md (movdd): Delete, combine with binary
17994         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
17995         with other moves by using conditional constraits (wg).  Use LFIWZX
17996         and STFIWX for loading SDmode on power7.
17997         (movdd splitters): Likewise.
17998         (movdd_hardfloat32): Likewise.
17999         (movdd_softfloat32): Likewise.
18000         (movdd_hardfloat64_mfpgpr): Likewise.
18001         (movdd_hardfloat64): Likewise.
18002         (movdd_softfloat64): Likewise.
18004         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
18005         64-bit binary and decimal floating point moves.
18006         (FMOVE64X): Likewise.
18007         (movdf): Combine 64-bit binary and decimal floating point moves.
18008         Combine power6x (mfpgpr) moves with other moves by using
18009         conditional constraits (wg).
18010         (mov<mode> for DFmode/DDmode): Likewise.
18011         (DFmode/DDmode splitters): Likewise.
18012         (movdf_hardfloat32): Likewise.
18013         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
18014         (movdf_softfloat32): Likewise.
18015         (movdf_hardfloat64_mfpgpr): Likewise.
18016         (movdf_hardfloat64): Likewise.
18017         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
18018         (movdf_softfloat64): Likewise.
18019         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
18020         (reload_<mode>_load): Move to later in the file so they aren't in
18021         the middle of the floating point move insns.
18022         (reload_<mode>_store): Likewise.
18024         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
18025         constraint.
18027         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
18028         constraint if -mdebug=reg.
18029         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
18030         Enable using dd reload support if needed.
18032         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
18033         binary and decimal floating point moves in rs6000.md.
18034         (movtd_internal): Likewise.
18036         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
18037         decimal floating point moves.
18038         (movtf): Likewise.
18039         (movtf_internal): Likewise.
18040         (mov<mode>_internal, TDmode/TFmode): Likewise.
18041         (movtf_softfloat): Likewise.
18042         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
18044         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
18045         movdi_internal64, using wg constraint for move direct operations.
18046         (movdi_internal64): Likewise.
18048         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
18049         MODES_TIEABLE_P for selected modes.  Print the numerical value of
18050         the various virtual registers. Use GPR/FPR first/last values,
18051         instead of hard coding the register numbers.  Print which modes
18052         have reload functions registered.
18053         (rs6000_option_override_internal): If -mdebug=reg, trace the options
18054         settings before/after setting cpu, target and subtarget settings.
18055         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
18056         and for secondary reload failures in rs6000_secondary_reload_inner.
18057         (rs6000_secondary_reload_fail): Likewise.
18058         (rs6000_secondary_reload_inner): Likewise.
18060         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
18061         macros for first/last GPR and FPR registers.
18062         (LAST_GPR_REGNO): Likewise.
18063         (FIRST_FPR_REGNO): Likewise.
18064         (LAST_FPR_REGNO): Likewise.
18066         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
18067         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
18068         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
18069         (vcond<mode><mode>): Likewise.
18070         (vcondu<mode><mode>): Likewise.
18071         (vector_gtu<mode>): Likewise.
18072         (vector_gte<mode>): Likewise.
18073         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
18074         to prevent the compiler from converting DImode operations to TImode.
18075         (ior<mode>3): Likewise.
18076         (and<mode>3): Likewise.
18077         (one_cmpl<mode>2): Likewise.
18078         (nor<mode>3): Likewise.
18079         (andc<mode>3): Likewise.
18081         * config/rs6000/constraints.md (wt constraint): New constraint
18082         that returns VSX_REGS if TImode is allowed in VSX registers.
18084         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
18085         constant under VSX.
18087         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
18088         similar to TImode, but it is restricted to being in the GPRs.
18090         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
18091         TImode to occupy a single VSX register.
18093         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
18094         -mvsx-timode for power7/power8.
18095         (power7 cpu): Likewise.
18096         (power8 cpu): Likewise.
18098         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
18099         sure that TFmode/TDmode take up two registers if they are ever
18100         allowed in the upper VSX registers.
18101         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
18102         registers.
18103         (rs6000_init_hard_regno_mode_ok): Likewise.
18104         (rs6000_debug_reg_global): Add debugging for PTImode and wt
18105         constraint.  Print if LRA is turned on.
18106         (rs6000_option_override_internal): Give an error if -mvsx-timode
18107         and VSX is not enabled.
18108         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
18109         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
18110         to reg+offset addressing.  Use PTImode when checking offset
18111         addresses for validity.
18112         (reg_offset_addressing_ok_p): Likewise.
18113         (rs6000_legitimate_offset_address_p): Likewise.
18114         (rs6000_legitimize_address): Likewise.
18115         (rs6000_legitimize_reload_address): Likewise.
18116         (rs6000_legitimate_address_p): Likewise.
18117         (rs6000_eliminate_indexed_memrefs): Likewise.
18118         (rs6000_emit_move): Likewise.
18119         (rs6000_secondary_reload): Likewise.
18120         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
18121         reloads to fpr registers to continue to use reg+offset addressing,
18122         but 64-bit reloads to altivec registers need reg+reg addressing.
18123         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
18124         it.  Treat LO_SUM like a PLUS operation.
18125         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
18126         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
18127         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
18128         registers to share a register with a smaller sized type, since VSX
18129         puts scalars in the upper 64-bits.
18130         (print_operand): Add support for PTImode.
18131         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
18132         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
18133         registers, but don't have arithmetic support.
18134         (rs6000_memory_move_cost): Add test for VSX.
18135         (rs6000_opt_masks): Add -mvsx-timode.
18137         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
18138         for TImode.
18139         (VSs): Likewise.
18140         (VSr): Use wt constraint for TImode.
18141         (VSv): Drop TImode support.
18142         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
18143         (vsx_movti_64bit): Likewise.
18144         (vsx_movti_32bit): Likewise.
18145         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
18146         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
18147         one '?' on the appropriate output constraint.  Do not allow TImode
18148         logical operations on 32-bit systems.
18149         (vsx_ior<mode>3): Likewise.
18150         (vsx_xor<mode>3): Likewise.
18151         (vsx_one_cmpl<mode>2): Likewise.
18152         (vsx_nor<mode>3): Likewise.
18153         (vsx_andc<mode>3): Likewise.
18154         (vsx_concat_<mode>): Likewise.
18155         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
18157         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
18158         OPTION_MASK_VSX_TIMODE.
18159         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
18160         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
18162         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
18163         (TI2 iterator): New iterator for TImode, PTImode.
18164         (wd mode attribute): Add values for vector types.
18165         (movti_string): Replace TI move operations with operations for TImode
18166         and PTImode.  Add support for TImode being allowed in VSX registers.
18167         (mov<mode>_string, TImode/PTImode): Likewise.
18168         (movti_ppc64): Likewise.
18169         (mov<mode>_ppc64, TImode/PTImode): Likewise.
18170         (TI mode splitters): Likewise.
18172         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
18173         constraint.
18175 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
18177         PR tree-optimization/56355
18178         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
18179         Also handle integers with undefined overflow.
18181 2013-03-20  Catherine Moore  <clm@codesourcery.com>
18182             Maciej W. Rozycki  <macro@codesourcery.com>
18183             Tom de Vries  <tom@codesourcery.com>
18184             Nathan Sidwell  <nathan@codesourcery.com>
18185             Iain Sandoe  <iain@codesourcery.com>
18186             Nathan Froyd  <froydnj@codesourcery.com>
18187             Chao-ying Fu  <fu@mips.com>
18189         * doc/extend.texi: (micromips, nomicromips, nocompression):
18190         Document new function attributes.
18191         * doc/invoke.texi (minterlink-compressed, mmicromips,
18192         m14k, m14ke, m14kec): Document new options.
18193         (minterlink-mips16): Update documentation.
18194         * doc/md.texi (ZC, ZD): Document new constraints.
18195         * configure.ac (gcc_cv_as_micromips): Check if linker
18196         supports the .set micromips directive.
18197         * configure: Regenerate.
18198         * config.in: Regenerate.
18199         * config/mips/mips-tables.opt: Regenerate.
18200         * config/mips/micromips.md: New file.
18201         * constraints.md (ZC, ZD): New constraints.
18202         * config/mips/predicates.md (movep_src_register): New predicate.
18203         (movep_src_operand): New predicate.
18204         (non_volatile_mem_operand): New predicate.
18205         * config/mips/mips.md (multimem): New type.
18206         (length): Differentiate between 17-bit and 18-bit branch offsets.
18207         (MOVEP1, MOVEP2): New mode iterator.
18208         (mov_<load>l): Use ZC constraint.
18209         (mov_<load>r): Likewise.
18210         (mov_<store>l): Likewise.
18211         (mov_<store>r): Likewise.
18212         (*branch_equality<mode>_inverted): Add microMIPS support.
18213         (*branch_equality<mode>): Likewise.
18214         (*jump_absolute): Likewise.
18215         (indirect_jump_<mode>): Likewise.
18216         (tablejump_<mode>): Likewise.
18217         (<optab>_internal): Likewise.
18218         (sibcall_internal): Likewise.
18219         (sibcall_value_internal): Likewise.
18220         (prefetch): Use constraint ZD.
18221         * config/mips/mips.opt (minterlink-compressed): New option.
18222         (minterlink-mips16): Now an alias for minterlink-compressed.
18223         (mmicromips): New option.
18224         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
18225         (compare_and_swap_12): Likewise.
18226         (sync_add<mode>): Likewise.
18227         (sync_<optab>_12): Likewise.
18228         (sync_old_<optab>_12): Likewise.
18229         (sync_new_<optab>_12): Likewise.
18230         (sync_nand_12): Likewise.
18231         (sync_old_nand_12): Likewise.
18232         (sync_new_nand_12): Likewise.
18233         (sync_sub<mode>): Likewise.
18234         (sync_old_add<mode>): Likewise.
18235         (sync_old_sub<mode>): Likewise.
18236         (sync_new_add<mode>): Likewise.
18237         (sync_new_sub<mode>): Likewise.
18238         (sync_<optab><mode>): Likewise.
18239         (sync_old_<optab><mode>): Likewise.
18240         (sync_new_<optab><mode>): Likewise.
18241         (sync_nand<mode>): Likewise.
18242         (sync_old_nand<mode>): Likewise.
18243         (sync_new_nand<mode>): Likewise.
18244         (sync_lock_test_and_set<mode>): Likewise.
18245         (test_and_set_12): Likewise.
18246         (atomic_compare_and_swap<mode>): Likewise.
18247         (atomic_exchange<mode>_llsc): Likewise.
18248         (atomic_fetch_add<mode>_llsc): Likewise.
18249         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
18250         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
18251         (umips_save_restore_pattern_p): Likewise.
18252         (umips_load_store_pair_p): Likewise.
18253         (umips_output_load_store_pair): Likewise.
18254         (umips_movep_target_p): Likewise.
18255         (umips_12bit_offset_address_p): Likewise.
18256         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
18257         (mips_base_mips16): Rename this...
18258         (mips_base_compression_flags): ...to this. Update all uses.
18259         (mips_attribute_table): Add micromips, nomicromips and nocompression.
18260         (mips_mips16_decl_p): Delete.
18261         (mips_nomips16_decl_p): Delete.
18262         (mips_get_compress_on_flags): New function.
18263         (mips_get_compress_off_flags): New function.
18264         (mips_get_compress_mode): New function.
18265         (mips_get_compress_on_name): New function.
18266         (mips_get_compress_off_name): New function.
18267         (mips_insert_attributes): Support multiple compression types.
18268         (mips_merge_decl_attributes): Likewise.
18269         (umips_12bit_offset_address_p): New function.
18270         (mips_start_function_definition): Emit .set micromips directive.
18271         (mips_call_may_need_jalx_p): New function.
18272         (mips_function_ok_for_sibcall): Add microMIPS support.
18273         (mips_print_operand_punctuation): Support short delay slots and
18274         compact jumps.
18275         (umips_swm_mask, umips_swm_encoding): New.
18276         (umips_build_save_restore): New function.
18277         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
18278         (was_mips16_p): Remove.
18279         (old_compression_mode): New.
18280         (mips_set_compression_mode): New function.
18281         (mips_set_current_function): Add microMIPS support.
18282         (mips_option_override): Likewise.
18283         (umips_save_restore_pattern_p): New function.
18284         (umips_output_save_restore): New function.
18285         (umips_load_store_pair_p_1): New function.
18286         (umips_load_store_pair_p): New function.
18287         (umips_output_load_store_pair_1): New function.
18288         (umips_output_load_store_pair): New function.
18289         (umips_movep_target_p) New function.
18290         (mips_prepare_pch_save): Add microMIPS support.
18291         * config/mips/mips.h (TARGET_COMPRESSION): New.
18292         (TARGET_CPU_CPP_BUILTINS): Update macro
18293         to use new compression flags and to support microMIPS.
18294         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
18295         (MIPS_ARCH_FLOAT_SPEC): Likewise.
18296         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
18297         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
18298         (ASM_SPEC): Support mmicromips and mno-micromips.
18299         (M16STORE_REG_P): New macro.
18300         (MIPS_CALL): Support TARGET_MICROMIPS.
18301         (MICROMIPS_J): New macro.
18302         (mips_base_mips16): Rename this...
18303         (mips_base_compression_flags): ...to this.
18304         (UMIPS_12BIT_OFFSET_P): New macro.
18305         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
18306         (MULTILIB_DIRNAMES): Likewise.
18307 2013-03-20  Richard Biener  <rguenther@suse.de>
18309         PR tree-optimization/56661
18310         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
18311         the result does not have to be distinct.
18313 2013-03-20  Richard Biener  <rguenther@suse.de>
18315         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
18316         remap_gimple_op_r.
18318 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18319             Steven Bosscher  <steven@gcc.gnu.org>
18321         PR rtl-optimization/56605
18322         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
18324 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
18326         PR bootstrap/56656
18327         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
18328         that require movd instead of movq.
18330 2013-03-20  Richard Biener  <rguenther@suse.de>
18332         * tree-ssa-structalias.c (struct variable_info): Add pointer
18333         to the first field of an aggregate with sub-vars.  Make
18334         this and the pointer to the next subfield its ID.
18335         (vi_next): New function.
18336         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
18337         storedanything_id, integer_id): Increment by one.
18338         (new_var_info, get_call_vi, lookup_call_clobber_vi,
18339         get_call_clobber_vi): Adjust.
18340         (solution_set_expand): Simplify and speedup.
18341         (solution_set_add): Inline into ...
18342         (set_union_with_increment): ... this.  Adjust accordingly.
18343         (do_sd_constraint): Likewise.
18344         (do_ds_constraint): Likewise.
18345         (do_complex_constraint): Simplify.
18346         (build_pred_graph): Adjust.
18347         (solve_graph): Likewise.  Simplify and speedup.
18348         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
18349         get_constraint_for_component_ref, get_constraint_for_1,
18350         first_vi_for_offset, first_or_preceding_vi_for_offset,
18351         create_function_info_for, create_variable_info_for_1,
18352         create_variable_info_for, intra_create_variable_infos): Adjust.
18353         (init_base_vars): Push NULL for ID zero.
18354         (compute_points_to_sets): Adjust.
18356 2013-03-20  Richard Biener  <rguenther@suse.de>
18358         * cfgloop.c (verify_loop_structure): Streamline and avoid
18359         ICEing on corrupt loop tree.
18360         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
18361         loop tree.
18363 2013-03-20  Richard Biener  <rguenther@suse.de>
18365         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
18366         check whether an SSA update is needed.
18368 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
18370         * config/mips/constraints.md (T): Rename to...
18371         (Yf): ...this.
18372         (U): Rename to...
18373         (Yd): ...this.
18374         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
18375         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
18377 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
18379         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
18380         (*subsi3_carryin_uxtw): Likewise.
18382 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
18384         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
18385         (*rorsi3_insn_uxtw): Likewise.
18387 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
18389         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
18390         (*extrsi5_insn_uxtw): Likewise.
18392 2013-03-19  Richard Biener  <rguenther@suse.de>
18394         PR tree-optimization/56273
18395         * passes.c (init_optimization_passes): Move second VRP after DOM.
18397 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
18399         * config/i386/i386.md (*movti_internal): Merge from
18400         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
18401         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
18402         nox64 isa attributes.
18404 2013-03-18  Richard Biener  <rguenther@suse.de>
18406         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
18407         (unite): Likewise.
18408         (merge_node_constraints): Likewise.
18409         (build_succ_graph): Likewise.
18410         (valid_graph_edge): Inline into single caller.
18411         (unify_nodes): Likewise.  Use bitmap_set_bit return value
18412         and cache varinfo.
18413         (scc_visit): Fix formatting and variable use.
18414         (do_sd_constraint): Use gcc_checking_assert.
18415         (do_ds_constraint): Likewise.
18416         (do_complex_constraint): Likewise.
18417         (condense_visit): Likewise.  Cleanup.
18418         (dump_pred_graph): New function.
18419         (perform_var_substitution): Dump the pred-graph before
18420         variable substitution.
18421         (find_equivalent_node): Use gcc_checking_assert.
18422         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
18424 2013-03-18  Richard Biener  <rguenther@suse.de>
18426         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
18427         Remove cond_expr_stmt_list argument and do not gimplify the
18428         built expression.
18429         (vect_loop_versioning): Adjust.
18430         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
18431         Cleanup to use less temporaries.
18432         (vect_create_data_ref_ptr): Cleanup.
18434 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
18436         PR tree-optimization/56635
18437         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
18438         require types_compatible_p types.
18440 2013-03-18  Nick Clifton  <nickc@redhat.com>
18442         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
18443         spurious backslash.
18445         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
18446         Add missing line to comment describing function.
18448 2013-03-18  Richard Biener  <rguenther@suse.de>
18450         PR tree-optimization/56210
18451         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18452         Handle string / character search functions.
18453         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
18455 2013-03-18  Richard Biener  <rguenther@suse.de>
18457         PR middle-end/56483
18458         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
18459         and implement properly.
18460         * gimple.h (gimple_cond_single_var_p): Remove.
18462 2013-03-18  Richard Biener  <rguenther@suse.de>
18464         * tree-data-ref.h (find_data_references_in_loop): Declare.
18465         * tree-data-ref.c (get_references_in_stmt): Use a stack
18466         vector pre-allocated in the callers.
18467         (find_data_references_in_stmt): Adjust.
18468         (graphite_find_data_references_in_stmt): Likewise.
18469         (create_rdg_vertices): Likewise.
18470         (find_data_references_in_loop): Export.
18471         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
18472         Compute dependences here...
18473         (vect_analyze_data_refs): ...not here.  When we encounter
18474         a non-vectorizable data reference in basic-block vectorization
18475         truncate the data reference vector.  Do not bother to
18476         fixup data-dependence information for gather loads.
18477         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
18478         of data references, as reported.
18480 2013-03-18  Richard Biener  <rguenther@suse.de>
18482         PR tree-optimization/3713
18483         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
18484         has_constants and expr.
18485         (stmt_has_constants): Properly valueize SSA names when deciding
18486         whether the stmt has constants.
18488 2013-03-18  Richard Biener  <rguenther@suse.de>
18490         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
18491         whole function when there is nothing to do.
18492         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
18493         * tree-vectorizer.c (vectorize_loops): Update virtual and
18494         loop-closed SSA once.
18495         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
18497 2013-03-18  Richard Biener  <rguenther@suse.de>
18499         PR middle-end/56113
18500         * domwalk.c (bb_postorder): New global static.
18501         (cmp_bb_postorder): New function.
18502         (walk_dominator_tree): Replace scheme imposing an order for
18503         visiting dominator sons by one sorting them at the time they
18504         are pushed on the stack.
18506 2013-03-18  Richard Biener  <rguenther@suse.de>
18508         PR tree-optimization/39326
18509         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
18510         (struct mem_ref): Replace mem member with ao_ref typed member.
18511         (MEM_ANALYZABLE): Adjust.
18512         (memref_eq): Likewise.
18513         (mem_ref_alloc): Likewise.
18514         (gather_mem_refs_stmt): Likewise.
18515         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
18516         (execute_sm_if_changed_flag_set): Adjust.
18517         (execute_sm): Likewise.
18518         (ref_always_accessed_p): Likewise.
18519         (refs_independent_p): Likewise.
18520         (can_sm_ref_p): Likewise.
18522 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
18524         PR c/56566
18525         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
18526         return 1 even for !unsignedp.
18528 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
18530         * config/i386/i386.md (isa): Add x64 and nox64.
18531         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
18532         (*pushtf): Enable *roF alternative for x64 isa only.
18533         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
18534         mode attribute of integer alternatives to DImode for TARGET_64BIT.
18535         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
18536         (*movtf_internal): Merge from *movtf_internal_rex64 and
18537         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
18538         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
18539         nox64 isa attributes.
18540         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
18541         nox64 isa attributes.
18542         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
18544 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
18546         * config/alpha/alpha.c (TARGET_LRA_P): New define.
18548 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
18550         PR target/56640
18551         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
18552         class names.  Remove trailing comma after "ALL_REGS".
18554 2013-03-16  Jan Hubicka  <jh@suse.cz>
18556         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
18557         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
18558         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
18559         of cgraph_get_create_node.
18560         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
18562 2013-03-16  Jason Merrill  <jason@redhat.com>
18564         PR debug/49090
18565         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
18566         with DW_AT_default_value.
18568 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
18570         * BASE-VER: Set to 4.9.0.
18572 2013-03-14  Andi Kleen  <ak@linux.intel.com>
18574         PR target/56619
18575         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
18576         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
18577         Document _x* TSX intrinsics.
18579 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
18580             David Holsgrove  <david.holsgrove@xilinx.com>
18582         * configure.ac: Add MicroBlaze TLS support detection.
18583         * configure: Regenerate.
18584         * config/microblaze/microblaze-protos.h
18585         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
18586         symbol_mentioned_p, label_mentioned_p): Add prototypes.
18587         * config/microblaze/microblaze.c (microblaze_address_type): Add
18588         ADDRESS_TLS and tls_reloc address types.
18589         (microblaze_address_info): Add tls_reloc.
18590         (TARGET_HAVE_TLS): Define.
18591         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
18592          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
18593          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
18594         load_tls_operand,  microblaze_call_tls_get_addr,
18595         microblaze_legitimize_tls_address): New functions.
18596         (microblaze_classify_unspec): Handle UNSPEC_TLS.
18597         (get_base_reg): Use microblaze_tls_symbol_p.
18598         (microblaze_classify_address): Handle TLS.
18599         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
18600         label_mentioned_p and microblaze_tls_referenced_p.
18601         (microblaze_legitimize_address): Handle TLS.
18602         (microblaze_address_insns): Handle ADDRESS_TLS.
18603         (pic_address_needs_scratch): Handle TLS.
18604         (print_operand_address): Handle TLS.
18605         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
18606         (microblaze_expand_move): Handle TLS.
18607         (microblaze_legitimate_constant_p): Check
18608         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
18609         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
18610         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
18611         (PIC_OFFSET_TABLE_REGNUM): Set.
18612         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
18613         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
18614         (addsi3, movsi_internal2, movdf_internal): Update constraints
18615         * config/microblaze/predicates.md (arith_plus_operand): Define
18616         (move_operand): Redefine as move_src_operand,
18617         check microblaze_tls_referenced_p.
18619 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
18621         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
18622         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
18624 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
18626         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
18627         CC mode for AND.
18629 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
18631         PR tree-optimization/53265
18632         * common.opt (Waggressive-loop-optimizations): New option.
18633         * tree-ssa-loop-niter.c: Include tree-pass.h.
18634         (do_warn_aggressive_loop_optimizations): New function.
18635         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
18636         if number_of_latch_executions returned constant.
18637         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
18638         early.  If number_of_latch_executions returned constant, set
18639         nb_iterations_upper_bound back to it.
18640         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
18641         field.
18642         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
18643         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
18645         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
18646         (MULTILIB_OSDIRNAMES): Set.
18647         * genmultilib: If defaultosdirname doesn't start with :: , set
18648         defaultosdirname2 instead, clear it and emit two . multilib_raw
18649         entries instead of just one.
18651 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
18653         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
18654         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
18655         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
18656         (SUBTARGET_OVERRIDE_OPTIONS): New.
18658 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
18660         PR target/49880
18661         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
18662         (musermode): Convert to Var(TARGET_USERMODE).
18663         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
18664         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
18665         * config/sh/sh.c (sh_option_override): Use
18666         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
18667         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
18668         condition.
18669         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
18670         TARGET_SH4.
18671         (udivsi3_i4_single, divsi3_i4_single): Use
18672         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
18674 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
18676         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
18677         default setting.
18679 2013-03-13  Richard Biener  <rguenther@suse.de>
18681         PR tree-optimization/56608
18682         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
18683         calls when vectorizing basic-blocks.
18685 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
18687         PR plugins/45078
18688         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
18689         tm_file.
18691 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
18693         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
18695 2013-03-11  Jan Hubicka  <jh@suse.cz>
18697         PR lto/56557
18698         * lto-streamer-out.c (output_symbol_p): Skip references from
18699         constructors of external variables.
18701 2013-03-11  Jan Hubicka  <jh@suse.cz>
18703         PR middle-end/56571
18704         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
18705         from pseudos.
18706         * emit-rtl.c (verify_rtx_sharing): Likewise.
18707         (copy_insn_1): Likewise.
18708         * rtl.c (copy_rtx): Likewise.
18710 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
18712         PR target/56591
18713         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
18714         output_operand_lossage message.
18716 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
18718         PR target/56470
18719         * arm.c (shift_op): Validate RTL pattern on the fly.
18720         (arm_print_operand, case 'S'): Don't use shift_operator to validate
18721         the RTL.
18723 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18725         PR target/56347
18726         * config/pa/pa.md (call_value): Check for calls to powf and direct to
18727         new call patterns that clobber %fr12.
18728         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
18729         split and postreload patterns.
18730         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
18731         registers %fr12 and %fr12R as call used.
18733 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
18735         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
18736         (canon_address, record_store, replace_read, check_mem_read_rtx,
18737         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
18738         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
18739         rest_of_handle_dse): Likewise.
18741 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
18743         PR middle-end/56524
18744         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
18745         Add base_optabs.
18746         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
18747         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
18748         (save_optabs_if_changed): Replace with...
18749         (init_tree_optimization_optabs): ...this.
18750         * optabs.c (save_optabs_if_changed): Rename to...
18751         (init_tree_optimization_optabs): ...this.  Take the optimization node
18752         as argument.  Do nothing if the base optabs are already correct.
18753         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
18754         to recompute optabs.
18755         * function.h (function): Remove optabs field.
18756         * function.c (invoke_set_current_function_hook): Call
18757         init_tree_optimization_optabs.  Use the result to initialize
18758         this_fn_optabs.
18760 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
18762         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
18763         if GTMA_HAS_NO_INSTRUMENTATION.
18764         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
18765         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
18766         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
18767         * gimple-pretty-print.c (dump_gimple_transaction): Handle
18768         GTMA_HAS_NO_INSTRUMENTATION.
18770 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
18772         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
18773         libasan_preinit.o.
18775 2013-03-08  Marek Polacek  <polacek@redhat.com>
18776             Jakub Jelinek  <jakub@redhat.com>
18778         PR tree-optimization/56478
18779         * predict.c (is_comparison_with_loop_invariant_p): Change the
18780         type of loop_step to tree.
18781         (predict_loops): Adjust.
18782         (predict_iv_comparison): Perform the computations on double_ints.
18784 2013-03-08  Richard Biener  <rguenther@suse.de>
18786         PR tree-optimization/56570
18787         * tree-cfg.c (verify_expr_location_1): Verify locations for
18788         DECL_DEBUG_EXPR.
18789         * tree-sra.c (create_access_replacement): Strip locations
18790         from DECL_DEBUG_EXPRs.
18792 2013-03-08  Richard Biener  <rguenther@suse.de>
18794         * tree-inline.c (expand_call_inline): Do not associate
18795         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
18796         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
18798 2013-03-08  Richard Biener  <rguenther@suse.de>
18800         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
18801         or block changes with -Og.  Fix for location / block encoding
18802         changes and PHI arguments with locations.
18804 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
18806         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
18807         for all counters.
18808         (struct output_info): Likewise.
18809         (register_overhead): Remove bad gcc_assert.
18810         (bitmap_find_bit): If there is only a single bitmap element, do not
18811         count a miss as a search.
18812         (print_statistics): Update for counter type changes.
18813         (dump_bitmap_statistics): Likewise.  Print headers such that they
18814         are properly lined up with the printed counters.
18816 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
18818         PR tree-optimization/56559
18819         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
18820         check that it has only a single use.
18822 2013-03-07  Richard Biener  <rguenther@suse.de>
18824         * doc/invoke.texi (fwhole-program): Discourage use in combination
18825         with -flto.
18827 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
18829         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
18831         PR tree-optimization/56539
18832         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
18833         instead of GSI_CONTINUE_LINKING as last argument to
18834         force_gimple_operand_gsi.  Adjust function comment.
18836         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
18837         aarch64-cores.def.
18839         PR middle-end/56548
18840         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
18841         promoted mode, convert the result back to the original mode.
18843 2013-03-06  Richard Biener  <rguenther@suse.de>
18845         PR middle-end/56294
18846         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
18847         (insert_updated_phi_nodes_compare_uids): New function.
18848         (update_ssa): Sort symbols_to_rename after UID before
18849         traversing it to insert PHI nodes.
18851 2013-03-06  Richard Biener  <rguenther@suse.de>
18853         PR middle-end/50494
18854         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
18855         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
18857         Revert
18858         2013-02-13  Richard Biener  <rguenther@suse.de>
18860         PR lto/50494
18861         * varasm.c (output_constant_def_1): Get the decl representing
18862         the constant as argument.
18863         (output_constant_def): Wrap output_constant_def_1.
18864         (make_decl_rtl): Use output_constant_def_1 with the decl
18865         representing the constant.
18866         (build_constant_desc): Optionally re-use a decl already
18867         representing the constant.
18868         (tree_output_constant_def): Adjust.
18870 2013-03-06  Joey Ye  <joey.ye@arm.com>
18872         PR lto/50293
18873         * gcc.c (convert_white_space): New function.
18874         (main): Handles white space in function name.
18876 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
18878         PR target/56529
18879         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
18880         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
18881         to SH_DIV_CALL_TABLE for TARGET_SH2.
18882         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
18883         list.
18884         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
18885         call-table options.
18887 2013-03-05  Sterling Augustine  <saugustine@google.com>
18888             Cary Coutant  <ccoutant@google.com>
18890         PR debug/55364
18891         * dwarf2out.c (resolve_addr): Don't call
18892         remove_loc_list_addr_table_entries a second time for the same
18893         expression.
18895 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18897         PR debug/56510
18898         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
18899         (avoid_complex_debug_insns): New function.
18900         (expand_debug_locations): Call it.
18902         PR rtl-optimization/56484
18903         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
18904         lifetimes of hard registers on small register class machines.
18906 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
18908         * config/microblaze/microblaze-protos.h: Rename
18909         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
18910         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
18911         fast_interrupt.
18912         (microblaze_fast_interrupt_function_p): New function.
18913         (microblaze_is_interrupt_handler): Rename to
18914         microblaze_is_interrupt_variant and add fast_interrupt check.
18915         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
18916         (save_restore_insns): Likewise.
18917         (compute_frame_size): Likewise.
18918         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
18919         (microblaze_globalize_label): Likewise.
18920         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
18921         * config/microblaze/microblaze.md: Use wrapper
18922         microblaze_is_interrupt_variant.
18924 2013-03-05  Kai Tietz  <ktietz@redhat.com>
18926         * sdbout.c (sdbout_one_type): Switch to current function's section
18927         supporting cold/hot.
18929 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
18931         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
18932         -mxl-reorder.
18934 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18936         PR middle-end/56461
18937         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
18938         if VALGRIND_GET_VBITS is defined, temporarily make object
18939         memory all defined, and restore previous valgrind addressability
18940         and definability afterwards.  Free this_object at the end.
18942         PR middle-end/56461
18943         * lra.c (lra): Call lra_clear_live_ranges if live_p,
18944         right before calling lra_create_live_ranges, also call it
18945         when clearing live_p.  Only call lra_clear_live_ranges
18946         at the end if live_p.
18948         PR middle-end/56461
18949         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
18951 2013-03-05  Richard Biener  <rguenther@suse.de>
18953         PR tree-optimization/56521
18954         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
18955         value-id.
18957 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
18959         PR c++/55135
18960         * except.h (remove_unreachable_eh_regions): New prototype.
18961         * except.c (remove_eh_handler_splicer): New function, split out
18962         of remove_eh_handler.
18963         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
18964         warning about running it on many EH regions one at a time.
18965         (remove_unreachable_eh_regions_worker): New function, walk the
18966         EH tree in depth-first order and remove non-marked regions.
18967         (remove_unreachable_eh_regions): New function.
18968         * tree-eh.c (mark_reachable_handlers): New function, split out
18969         from remove_unreachable_handlers.
18970         (remove_unreachable_handlers): Use mark_reachable_handlers and
18971         remove_unreachable_eh_regions.
18972         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
18973         and remove_unreachable_eh_regions.
18975 2013-03-05  Richard Biener  <rguenther@suse.de>
18977         PR middle-end/56525
18978         * loop-init.c (fix_loop_structure): Remove loops in two stages,
18979         not freeing them until the end.
18981 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
18983         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
18985 2013-03-05  Richard Biener  <rguenther@suse.de>
18987         PR tree-optimization/56270
18988         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
18989         of loads after scheduling an SLP instance.
18991 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
18993         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
18994         tic6x.exp.
18995         (check_gcc_parallelize): Run guality.exp as a separate job from
18996         vect.exp with unsorted.exp and $(dg_target_exps) separately from
18997         struct-layout-1.exp with stackalign.exp.
18999         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
19001         PR middle-end/56461
19002         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
19003         load_index sbitmap even if some bit in it isn't set.
19005         PR middle-end/56461
19006         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
19007         (discover_iteration_bound_by_body_walk): Change queues to
19008         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
19009         spelling in comment.  Call safe_push on queues[bound_index] directly.
19010         Release queues[queue_index] in every iteration unconditionally.
19011         Release bounds vector.
19013         PR middle-end/56461
19014         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
19015         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
19016         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
19017         inner_phis vector.
19019 2013-03-05  Richard Biener  <rguenther@suse.de>
19021         PR lto/56515
19022         * tree-inline.c (remap_blocks_to_null): New function.
19023         (expand_call_inline): When expanding a call stmt without
19024         an associated block inline remap all callee blocks to NULL.
19026 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
19028         PR rtl-optimization/56494
19029         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
19030         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
19031         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
19033         PR middle-end/56461
19034         * sel-sched-ir.c (free_sched_pools): Release
19035         succs_info_pool.stack[succs_info_pool.max_top] vectors too
19036         if succs_info_pool.max_top isn't -1.
19038         PR bootstrap/56509
19039         * opts.c (opts_obstack, opts_concat): Moved to...
19040         * opts-common.c (opts_obstack, opts_concat): ... here.
19042 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
19044         PR middle-end/56461
19045         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
19047 2013-03-04  Martin Jambor  <mjambor@suse.cz>
19049         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
19050         all appropriate places.
19052 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
19054         PR tree-optimization/56424
19055         * ipa-split.c (split_function): Do not set the RSO flag if result is
19056         not by reference and its type is a register type.
19058 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
19060         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
19061         (microblaze_legitimate_pic_operand): Likewise
19062         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
19063         new function microblaze_legitimate_pic_operand
19064         * config/microblaze/microblaze-protos.h
19065         (microblaze_legitimate_pic_operand): Declare.
19067 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19069         * config/microblaze/predicates.md (call_insn_simple_operand):
19070         New predicate for supported rtx code types.
19071         * config/microblaze/microblaze.md (call_internal1): Use
19072         call_insn_simple_operand predicate.
19074 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
19076         PR middle-end/56461
19077         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
19078         partitions.ordered_remove.
19080         PR middle-end/56461
19081         * tree-vect-stmts.c (vectorizable_conversion): Don't call
19082         vec_oprnds0.create (1) for modifier == NONE.
19084         PR middle-end/56461
19085         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
19086         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
19087         vec_oprnds1 right before pushing anything to it for
19088         scalar_shift_arg.
19090         PR middle-end/56461
19091         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
19092         set nbbs to 0 instead of having separate code path.
19093         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
19094         instead of false as last argument if returning NULL.
19096 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
19098         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
19099         the attribute is now called "target" instead of "option".
19100         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
19101         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
19102         attribute/pragma name for TARGET_OPTION_VALID_P and
19103         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
19104         * doc/tm.texi: Regenerated.
19106 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
19108         * config/microblaze/microblaze.c:
19109         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
19110         * config/microblaze/microblaze.h: Add -mxl-reorder to
19111         DRIVER_SELF_SPECS.
19112         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
19113         instructions emitted if TARGET_REORDER.
19114         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
19115         or 0 for -m/-mno case, but initialises as 2 to detect default use case
19116         separately.
19118 2013-03-01  Xinliang David Li  <davidxl@google.com>
19120         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
19121         walk length.
19123 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
19125         PR middle-end/56461
19126         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
19127         vector even when returning true.  Fix up function comment formatting.
19129         PR middle-end/56461
19130         * ira-build.c (ira_loop_nodes_count): New variable.
19131         (create_loop_tree_nodes): Initialize it.
19132         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
19134         PR middle-end/56461
19135         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
19136         method on dr_chain and result_chain.
19137         * tree-vect-stmts.c (vectorizable_store): Only call
19138         result_chain.create if j == 0.
19140         PR middle-end/56461
19141         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
19142         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
19143         before overwriting it.
19145 2013-03-01  Tobias Burnus  <burnus@net-b.de>
19147         * doc/extended.texi (C Extensions): Change order in @menu
19148         to match @node.
19149         (Other MIPS Built-in Functions): Move last MIPS entry before
19150         "picoChip Built-in Functions".
19151         (SH Built-in Functions): Move after RX Built-in Functions.
19152         * doc/gcc.texi (Introduction): Change order in @menu
19153         to match @node.
19154         * doc/md.texi (Constraints): Ditto.
19155         * gty.texi (Type Information): Ditto.
19156         (User-provided marking routines for template types): Make
19157         subsection.
19158         * doc/invoke.texi (AArch64 Options): Move before
19159         "Adapteva Epiphany Options".
19161 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
19162             Jakub Jelinek  <jakub@redhat.com>
19164         PR sanitizer/56454
19165         * asan.c (gate_asan): Lookup no_sanitize_address instead of
19166         no_address_safety_analysis attribute.
19167         * doc/extend.texi (no_address_safety_attribute): Rename to
19168         no_sanitize_address attribute, mention no_address_safety_analysis
19169         attribute as deprecated alias.
19171 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
19173         PR middle-end/56461
19174         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
19175         type to vec<vec<tree> > *.
19176         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
19177         to be vec<tree> instead of vec<tree> *, set vec_defs
19178         to vNULL and call vec_defs.create (number_of_vects), adjust other
19179         uses of vec_defs.
19180         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
19181         vectorizable_condition): Adjust vect_get_slp_defs callers.
19183 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
19185         * config/aarch64/aarch64.c
19186         (aarch64_float_const_representable): Remove unused variable.
19188 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
19190         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
19192 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
19194         * config/aarch64/aarch64-builtins.c
19195         (aarch64_init_simd_builtins): Make static.
19197 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
19199         * config/aarch64/aarch64.c
19200         (aarch64_simd_make_constant): Make static.
19202 2013-02-28  Martin Jambor  <mjambor@suse.cz>
19204         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
19205         with no initialization to the RHS of debug statements.
19207 2013-02-28  Martin Jambor  <mjambor@suse.cz>
19209         PR tree-optimization/56294
19210         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
19211         Adjust dumping.
19212         (get_access_replacement): Do not call create_access_replacement.
19213         Assert a replacement exists.
19214         (get_repl_default_def_ssa_name): Create the replacement declaration
19215         itself.
19217 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
19219         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
19220         final_end_function.
19222 2013-02-28  Marek Polacek  <polacek@redhat.com>
19224         PR rtl-optimization/56466
19225         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
19226         if we're changing a loop.
19227         (peel_loops_completely): Likewise.
19229 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
19231         PR c++/55813
19232         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
19234 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
19236         PR target/56445
19237         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
19238         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
19239         INTX_FTYPE_FX, FX_FTYPE_INTX.
19240         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
19242 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
19244         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
19245         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
19246         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
19247         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
19248         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
19249         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
19250         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
19251         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
19252         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
19253         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
19254         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
19255         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
19256         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
19257         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
19258         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
19259         (avrxmega6): Increase max flash segments from 5 to 6.
19260         * config/avr/t-multilib: Regenerate.
19261         * config/avr/avr-tables.opt: Regenerate.
19262         * doc/avr-mmcu.texi: Regenerate.
19264 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
19266         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
19267         (avr_device_to_arch): Rename to avr_device_to_ld.
19268         (avr_device_to_as): New prototype.
19269         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
19270         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
19271         * config/avr/driver-avr.c (avr_device_to_as): New.
19272         (avr_device_to_arch): Rename to avr_device_to_ld.
19274 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
19276         PR middle-end/56461
19277         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
19278         method on dr_chain and result_chain.
19280         PR middle-end/56461
19281         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
19282         pointer_set_destroy on not_executed_last_iteration.
19284         PR middle-end/56461
19285         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
19287         PR middle-end/56461
19288         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
19289         FOR_EACH_DEFINED_FUNCTION when freeing state.
19291         PR middle-end/56461
19292         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
19293         pool_free.
19294         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
19295         overwriting it.
19297         PR middle-end/56461
19298         * ipa-cp.c (decide_whether_version_node): Call vec_free on
19299         known_aggs[i].items and release known_aggs vector.
19301         PR middle-end/56461
19302         * ipa-reference.c (propagate): Free node_info even for alias nodes.
19304 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19306         * config/microblaze/microblaze.c (microblaze_emit_compare):
19307         Use xor for EQ/NE comparisions.
19308         * config/microblaze/microblaze.md (cstoresf4): Add constraints
19309         (cbranchsf4): Adjust operator to comparison_operator.
19311 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
19313         PR middle-end/56461
19314         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
19315         vector.
19316         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
19317         vec_safe_push, always update *slot.
19318         (redirect_edge_var_map_clear): Use vec_free.
19319         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
19320         (free_var_map_entry): Use vec_free.
19321         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
19322         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
19324 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
19326         PR middle-end/45472
19327         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
19328         when the may_trap_p bit of the exprs being merged differs.
19329         Reorder tests for speculativeness in the logical and operator.
19331 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
19333         * incpath.c (add_standard_paths): Use reconcat instead of concat
19334         where appropriate and avoid leaking memory.
19336         * opts.h: Include obstack.h.
19337         (opts_concat): New prototype.
19338         (opts_obstack): New declaration.
19339         * opts.c (opts_concat): New function.
19340         (opts_obstack): New variable.
19341         (init_options_struct): Call gcc_init_obstack on opts_obstack.
19342         (finish_options): Use opts_concat instead of concat
19343         and XOBNEWVEC instead of XNEWVEC.
19344         * opts-common.c (generate_canonical_option, decode_cmdline_option,
19345         generate_option): Likewise.
19346         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
19347         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
19349         PR target/56455
19350         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
19351         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
19353 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
19355         PR middle-end/56461
19356         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
19358 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
19360         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
19361         (arm_block_move_unaligned_straight): Likewise.
19362         (arm_adjust_block_mem): Likewise.
19364 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
19366         PR target/48901
19367         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
19368         temp, cond and label.
19369         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
19371         PR target/52500
19372         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
19373         * config/c6x/c6x.h (dbx_register_map): Update declaration.
19375         PR target/52501
19376         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
19377         of prologue/epilogue functions.
19379         PR target/52550
19380         * config/tilegx/tilegx.c (tilegx_expand_prologue):
19381         Remove unused variable cfa_offset.
19382         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
19384         PR target/54639
19385         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
19386         type promotion to unsigned.
19388         PR target/54640
19389         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
19390         for HOST_WIDE_INT of 32 bit / same size as int.
19391         (arm_block_move_unaligned_straight): Likewise.
19392         (arm_adjust_block_mem): Likewise.
19394         PR target/54662
19395         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
19396         ALL_CFLAGS.
19398 2013-02-26  Marek Polacek  <polacek@redhat.com>
19400         PR tree-optimization/56426
19401         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
19403 2013-02-26  Richard Biener  <rguenther@suse.de>
19405         PR target/56444
19406         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
19407         unused variable loops.
19409 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
19411         PR tree-optimization/56448
19412         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
19413         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
19414         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
19415         later operands of the references, or even first operand for
19416         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
19418         PR tree-optimization/56443
19419         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
19420         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
19421         to type_for_mode langhook.
19423 2013-02-25  Matt Turner  <mattst88@gmail.com>
19425         * doc/invoke.texi: Document r4700.
19427 2013-02-25  Richard Biener  <rguenther@suse.de>
19429         PR tree-optimization/56175
19430         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
19431         split out from ...
19432         (simplify_bitwise_binary): ... here.  Also guard the conversion
19433         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
19435 2013-02-25  Catherine Moore  <clm@codesourcery.com>
19437         Revert:
19438         2013-02-24  Catherine Moore  <clm@codesourcery.com>
19439             Maciej W. Rozycki  <macro@codesourcery.com>
19440             Tom de Vries  <tom@codesourcery.com>
19441             Nathan Sidwell  <nathan@codesourcery.com>
19442             Iain Sandoe  <iain@codesourcery.com>
19443             Nathan Froyd  <froydnj@codesourcery.com>
19444             Chao-ying Fu  <fu@mips.com>
19446         * doc/extend.texi: (micromips, nomicromips, nocompression):
19447         Document new function attributes.
19448         * doc/invoke.texi (minterlink-compressed, mmicromips,
19449         m14k, m14ke, m14kec): Document new options.
19450         (minterlink-mips16): Update documentation.
19451         * doc/md.texi (ZC, ZD): Document new constraints.
19452         * configure.ac (gcc_cv_as_micromips): Check if linker
19453         supports the .set micromips directive.
19454         * configure: Regenerate.
19455         * config.in: Regenerate.
19456         * config/mips/mips-tables.opt: Regenerate.
19457         * config/mips/micromips.md: New file.
19458         * constraints.md (ZC, AD): New constraints.
19459         * config/mips/predicates.md (movep_src_register): New predicate.
19460         (movep_src_operand): New predicate.
19461         (non_volatile_mem_operand): New predicate.
19462         * config/mips/mips.md (multimem): New type.
19463         (length): Differentiate between 17-bit and 18-bit branch offsets.
19464         (MOVEP1, MOVEP2): New mode iterator.
19465         (mov_<load>l): Use ZC constraint.
19466         (mov_<load>r): Likewise.
19467         (mov_<store>l): Likewise.
19468         (mov_<store>r): Likewise.
19469         (*branch_equality<mode>_inverted): Add microMIPS support.
19470         (*branch_equality<mode>): Likewise.
19471         (*jump_absolute): Likewise.
19472         (indirect_jump_<mode>): Likewise.
19473         (tablejump_<mode>): Likewise.
19474         (<optab>_internal): Likewise.
19475         (sibcall_internal): Likewise.
19476         (sibcall_value_internal): Likewise.
19477         (prefetch): Use constraint ZD.
19478         * config/mips/mips.opt (minterlink-compressed): New option.
19479         (minterlink-mips16): Now an alias for minterlink-compressed.
19480         (mmicromips): New option.
19481         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
19482         (compare_and_swap_12): Likewise.
19483         (sync_add<mode>): Likewise.
19484         (sync_<optab>_12): Likewise.
19485         (sync_old_<optab>_12): Likewise.
19486         (sync_new_<optab>_12): Likewise.
19487         (sync_nand_12): Likewise.
19488         (sync_old_nand_12): Likewise.
19489         (sync_new_nand_12): Likewise.
19490         (sync_sub<mode>): Likewise.
19491         (sync_old_add<mode>): Likewise.
19492         (sync_old_sub<mode>): Likewise.
19493         (sync_new_add<mode>): Likewise.
19494         (sync_new_sub<mode>): Likewise.
19495         (sync_<optab><mode>): Likewise.
19496         (sync_old_<optab><mode>): Likewise.
19497         (sync_new_<optab><mode>): Likewise.
19498         (sync_nand<mode>): Likewise.
19499         (sync_old_nand<mode>): Likewise.
19500         (sync_new_nand<mode>): Likewise.
19501         (sync_lock_test_and_set<mode>): Likewise.
19502         (test_and_set_12): Likewise.
19503         (atomic_compare_and_swap<mode>): Likewise.
19504         (atomic_exchange<mode>_llsc): Likewise.
19505         (atomic_fetch_add<mode>_llsc): Likewise.
19506         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
19507         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
19508         (umips_save_restore_pattern_p): Likewise.
19509         (umips_load_store_pair_p): Likewise.
19510         (umips_output_load_store_pair): Likewise.
19511         (umips_movep_target_p): Likewise.
19512         (umips_12bit_offset_address_p): Likewise.
19513         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
19514         (mips_base_mips16): Rename this...
19515         (mips_base_compression_flags): ...to this. Update all uses.
19516         (mips_attribute_table): Add micromips, nomicromips and nocompression.
19517         (mips_mips16_decl_p): Delete.
19518         (mips_nomips16_decl_p): Delete.
19519         (mips_get_compress_on_flags): New function.
19520         (mips_get_compress_off_flags): New function.
19521         (mips_get_compress_mode): New function.
19522         (mips_get_compress_on_name): New function.
19523         (mips_get_compress_off_name): New function.
19524         (mips_insert_attributes): Support multiple compression types.
19525         (mips_merge_decl_attributes): Likewise.
19526         (umips_12bit_offset_address_p): New function.
19527         (mips_start_function_definition): Emit .set micromips directive.
19528         (mips_call_may_need_jalx_p): New function.
19529         (mips_function_ok_for_sibcall): Add microMIPS support.
19530         (mips_print_operand_punctuation): Support short delay slots and
19531         compact jumps.
19532         (umips_swm_mask, umips_swm_encoding): New.
19533         (umips_build_save_restore): New function.
19534         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
19535         (was_mips16_p): Remove.
19536         (old_compression_mode): New.
19537         (mips_set_compression_mode): New function.
19538         (mips_set_current_function): Add microMIPS support.
19539         (mips_option_override): Likewise.
19540         (umips_save_restore_pattern_p): New function.
19541         (umips_output_save_restore): New function.
19542         (umips_load_store_pair_p_1): New function.
19543         (umips_load_store_pair_p): New function.
19544         (umips_output_load_store_pair_1): New function.
19545         (umips_output_load_store_pair): New function.
19546         (umips_movep_target_p) New function.
19547         (mips_prepare_pch_save): Add microMIPS support.
19548         * config/mips/mips.h (TARGET_COMPRESSION): New.
19549         (TARGET_CPU_CPP_BUILTINS): Update macro
19550         to use new compression flags and to support microMIPS.
19551         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
19552         (MIPS_ARCH_FLOAT_SPEC): Likewise.
19553         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
19554         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
19555         (ASM_SPEC): Support mmicromips and mno-micromips.
19556         (M16STORE_REG_P): New macro.
19557         (MIPS_CALL): Support TARGET_MICROMIPS.
19558         (MICROMIPS_J): New macro.
19559         (mips_base_mips16): Rename this...
19560         (mips_base_compression_flags): ...to this.
19561         (UMIPS_12BIT_OFFSET_P): New macro.
19562         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
19563         (MULTILIB_DIRNAMES): Likewise.
19565 2013-02-25  Tom de Vries  <tom@codesourcery.com>
19567         PR rtl-optimization/56131
19568         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
19569         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
19570         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
19572 2013-02-25  Tobias Burnus  <burnus@net-b.de>
19574         * doc/invoke.texi (-fsanitize=): Move from optimization
19575         to debugging options.
19577 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
19579         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
19581 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
19582             Alexander Monakov  <amonakov@ispras.ru>
19584         PR middle-end/56077
19585         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
19586         flush pending lists also on non-jumps.  Adjust comment.
19588 2013-02-24  Catherine Moore  <clm@codesourcery.com>
19589             Maciej W. Rozycki  <macro@codesourcery.com>
19590             Tom de Vries  <tom@codesourcery.com>
19591             Nathan Sidwell  <nathan@codesourcery.com>
19592             Iain Sandoe  <iain@codesourcery.com>
19593             Nathan Froyd  <froydnj@codesourcery.com>
19594             Chao-ying Fu  <fu@mips.com>
19596         * doc/extend.texi: (micromips, nomicromips, nocompression):
19597         Document new function attributes.
19598         * doc/invoke.texi (minterlink-compressed, mmicromips,
19599         m14k, m14ke, m14kec): Document new options.
19600         (minterlink-mips16): Update documentation.
19601         * doc/md.texi (ZC, ZD): Document new constraints.
19602         * configure.ac (gcc_cv_as_micromips): Check if linker
19603         supports the .set micromips directive.
19604         * configure: Regenerate.
19605         * config.in: Regenerate.
19606         * config/mips/mips-tables.opt: Regenerate.
19607         * config/mips/micromips.md: New file.
19608         * constraints.md (ZC, AD): New constraints.
19609         * config/mips/predicates.md (movep_src_register): New predicate.
19610         (movep_src_operand): New predicate.
19611         (non_volatile_mem_operand): New predicate.
19612         * config/mips/mips.md (multimem): New type.
19613         (length): Differentiate between 17-bit and 18-bit branch offsets.
19614         (MOVEP1, MOVEP2): New mode iterator.
19615         (mov_<load>l): Use ZC constraint.
19616         (mov_<load>r): Likewise.
19617         (mov_<store>l): Likewise.
19618         (mov_<store>r): Likewise.
19619         (*branch_equality<mode>_inverted): Add microMIPS support.
19620         (*branch_equality<mode>): Likewise.
19621         (*jump_absolute): Likewise.
19622         (indirect_jump_<mode>): Likewise.
19623         (tablejump_<mode>): Likewise.
19624         (<optab>_internal): Likewise.
19625         (sibcall_internal): Likewise.
19626         (sibcall_value_internal): Likewise.
19627         (prefetch): Use constraint ZD.
19628         * config/mips/mips.opt (minterlink-compressed): New option.
19629         (minterlink-mips16): Now an alias for minterlink-compressed.
19630         (mmicromips): New option.
19631         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
19632         (compare_and_swap_12): Likewise.
19633         (sync_add<mode>): Likewise.
19634         (sync_<optab>_12): Likewise.
19635         (sync_old_<optab>_12): Likewise.
19636         (sync_new_<optab>_12): Likewise.
19637         (sync_nand_12): Likewise.
19638         (sync_old_nand_12): Likewise.
19639         (sync_new_nand_12): Likewise.
19640         (sync_sub<mode>): Likewise.
19641         (sync_old_add<mode>): Likewise.
19642         (sync_old_sub<mode>): Likewise.
19643         (sync_new_add<mode>): Likewise.
19644         (sync_new_sub<mode>): Likewise.
19645         (sync_<optab><mode>): Likewise.
19646         (sync_old_<optab><mode>): Likewise.
19647         (sync_new_<optab><mode>): Likewise.
19648         (sync_nand<mode>): Likewise.
19649         (sync_old_nand<mode>): Likewise.
19650         (sync_new_nand<mode>): Likewise.
19651         (sync_lock_test_and_set<mode>): Likewise.
19652         (test_and_set_12): Likewise.
19653         (atomic_compare_and_swap<mode>): Likewise.
19654         (atomic_exchange<mode>_llsc): Likewise.
19655         (atomic_fetch_add<mode>_llsc): Likewise.
19656         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
19657         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
19658         (umips_save_restore_pattern_p): Likewise.
19659         (umips_load_store_pair_p): Likewise.
19660         (umips_output_load_store_pair): Likewise.
19661         (umips_movep_target_p): Likewise.
19662         (umips_12bit_offset_address_p): Likewise.
19663         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
19664         (mips_base_mips16): Rename this...
19665         (mips_base_compression_flags): ...to this. Update all uses.
19666         (mips_attribute_table): Add micromips, nomicromips and nocompression.
19667         (mips_mips16_decl_p): Delete.
19668         (mips_nomips16_decl_p): Delete.
19669         (mips_get_compress_on_flags): New function.
19670         (mips_get_compress_off_flags): New function.
19671         (mips_get_compress_mode): New function.
19672         (mips_get_compress_on_name): New function.
19673         (mips_get_compress_off_name): New function.
19674         (mips_insert_attributes): Support multiple compression types.
19675         (mips_merge_decl_attributes): Likewise.
19676         (umips_12bit_offset_address_p): New function.
19677         (mips_start_function_definition): Emit .set micromips directive.
19678         (mips_call_may_need_jalx_p): New function.
19679         (mips_function_ok_for_sibcall): Add microMIPS support.
19680         (mips_print_operand_punctuation): Support short delay slots and
19681         compact jumps.
19682         (umips_swm_mask, umips_swm_encoding): New.
19683         (umips_build_save_restore): New function.
19684         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
19685         (was_mips16_p): Remove.
19686         (old_compression_mode): New.
19687         (mips_set_compression_mode): New function.
19688         (mips_set_current_function): Add microMIPS support.
19689         (mips_option_override): Likewise.
19690         (umips_save_restore_pattern_p): New function.
19691         (umips_output_save_restore): New function.
19692         (umips_load_store_pair_p_1): New function.
19693         (umips_load_store_pair_p): New function.
19694         (umips_output_load_store_pair_1): New function.
19695         (umips_output_load_store_pair): New function.
19696         (umips_movep_target_p) New function.
19697         (mips_prepare_pch_save): Add microMIPS support.
19698         * config/mips/mips.h (TARGET_COMPRESSION): New.
19699         (TARGET_CPU_CPP_BUILTINS): Update macro
19700         to use new compression flags and to support microMIPS.
19701         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
19702         (MIPS_ARCH_FLOAT_SPEC): Likewise.
19703         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
19704         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
19705         (ASM_SPEC): Support mmicromips and mno-micromips.
19706         (M16STORE_REG_P): New macro.
19707         (MIPS_CALL): Support TARGET_MICROMIPS.
19708         (MICROMIPS_J): New macro.
19709         (mips_base_mips16): Rename this...
19710         (mips_base_compression_flags): ...to this.
19711         (UMIPS_12BIT_OFFSET_P): New macro.
19712         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
19713         (MULTILIB_DIRNAMES): Likewise.
19715 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
19717         PR target/52555
19718         * target-globals.c (save_target_globals): For init_reg_sets and
19719         target_reinit remporarily set this_fn_optabs to this_target_optabs.
19721 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
19723         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
19724         * config/aarch64/t-aarch64
19725         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
19727 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
19729         PR inline-asm/56148
19730         * lra-constraints.c (process_alt_operands): Reload operand
19731         conflicting with earlier clobber only if no more other conflicting
19732         operands.
19734 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
19736         PR sanitizer/56393
19737         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
19738         if not linking a shared library.
19740 2013-02-22  Seth LaForge  <sethml@google.com>
19742         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
19744 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
19746         * config/arm/arm.md (split for extendsidi): Update condition.
19747         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
19748         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
19749         (qhs_zextenddi_cstr): Likewise.
19751 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
19753         PR middle-end/56420
19754         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
19755         avoid signed wrapping.
19756         (expand_mult): Handle properly multiplication by
19757         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
19758         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
19759         in the compiler if coeff is HOST_WIDE_INT_MIN.
19760         (expand_divmod): Don't make ext_op1 static, change it's type to
19761         uhwi.  Avoid undefined behavior in -INTVAL (op1).
19763         PR rtl-optimization/50339
19764         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
19765         field.
19766         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
19767         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
19768         into splitting_ashiftrt field.
19769         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
19770         ASHIFTRT.
19771         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
19772         choices.
19774 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
19776         PR middle-end/56108
19777         * trans-mem.c (execute_tm_mark): Do not expand transactions that
19778         are sure to go irrevocable.
19780 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
19782         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
19783         scalars are valid operands.
19785 2013-02-21  Martin Jambor  <mjambor@suse.cz>
19787         PR tree-optimization/56310
19788         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
19789         only matching indices and non-negative final offsets.
19790         (intersect_aggregates_with_edge): Pass src_idx to
19791         agg_replacements_to_vector.  Pass src_idx insstead of index to
19792         intersect_with_agg_replacements.
19794 2013-02-21  Martin Jambor  <mjambor@suse.cz>
19796         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
19797         instead of hard-wired defaults.
19799 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
19801         * doc/invoke.texi (MIPS Options): Update documentation of the
19802         floating-point multiply-accumulate instruction restrictions.
19804 2013-02-21  Kostya Serebryany  <kcc@google.com>
19806         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
19807         asan_shadow_offset on x86_64 linux.
19809 2013-02-21  Richard Biener  <rguenther@suse.de>
19811         PR tree-optimization/56415
19812         Revert
19813         2013-02-11  Richard Biener  <rguenther@suse.de>
19815         PR tree-optimization/56273
19816         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
19817         first VRP run.
19819 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
19821         PR bootstrap/56258
19822         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
19823         instead of @itemx.
19825         PR inline-asm/56405
19826         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
19827         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
19829 2013-02-20  Jan Hubicka  <jh@suse.cz>
19831         PR tree-optimization/56265
19832         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
19833         when target is referenced for first time.
19835 2013-02-20  Richard Biener  <rguenther@suse.de>
19837         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
19838         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
19839         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
19840         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
19841         not return anything.
19842         (rename_ssa_copies): Do not remove unused locals.
19843         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
19844         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
19845         * passes.c (execute_function_todo): Do not schedule unused locals
19846         removal if cleanup_tree_cfg did something.
19847         * tree-ssa-live.c (remove_unused_locals): Dump statistics
19848         about the number of removed locals.
19850 2013-02-20  Richard Biener  <rguenther@suse.de>
19852         PR tree-optimization/56398
19853         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
19855 2013-02-20  Martin Jambor  <mjambor@suse.cz>
19857         PR tree-optimization/55334
19858         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
19859         restricted pointers to arrays.
19861 2013-02-20  Richard Biener  <rguenther@suse.de>
19862             Jakub Jelinek  <jakub@redhat.com>
19864         PR tree-optimization/56396
19865         * tree-ssa-ccp.c (n_const_val): New static variable.
19866         (get_value): Return NULL for SSA names we don't have a lattice
19867         entry for.
19868         (ccp_initialize): Initialize n_const_val.
19869         * tree-ssa-copy.c (n_copy_of): New static variable.
19870         (init_copy_prop): Initialize n_copy_of.
19871         (get_value): Return NULL_TREE for SSA names we don't have a
19872         lattice entry for.
19874 2013-02-20  Martin Jambor  <mjambor@suse.cz>
19876         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
19878 2013-02-20  Richard Biener  <rguenther@suse.de>
19880         * genpreds.c (write_lookup_constraint): Do not compare first
19881         letter of the constraint again.
19883 2013-02-20  Richard Biener  <rguenther@suse.de>
19885         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
19886         and ceil_log2.
19887         (get_use_iv_cost): Terminate hashtable walk when coming across
19888         an empty entry.
19890 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
19892         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
19893         reassociation for avx2 targets.
19895 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19897         * config/microblaze/microblaze.c: microblaze_has_clz = 0
19898         Add version check for v8.10.a to enable microblaze_has_clz
19899         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
19900         version and TARGET_PATTERN_COMPARE check
19901         * config/microblaze/microblaze.md: New clzsi2 instruction
19903 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
19905         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
19906         function before branching.
19908 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
19910         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
19911         DUMP_INSN_RTX_UID.
19912         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
19914 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
19916         PR middle-end/55889
19917         * sel-sched.c: Include ira.h.
19918         (implicit_clobber_conflict_p): New function.
19919         (moveup_expr): Use it.
19920         * Makefile.in (sel-sched.o): Depend on ira.h.
19922 2013-02-19  Richard Biener  <rguenther@suse.de>
19924         PR tree-optimization/56384
19925         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
19926         (vn_hash_type): Split out from ...
19927         (vn_hash_constant_with_type): ... here.
19928         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
19929         (vn_phi_eq): Compare types from vn_phi_s structure.
19930         (vn_phi_lookup): Populate vn_phi_s type.
19931         (vn_phi_insert): Likewise.
19933 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
19935         PR tree-optimization/56350
19936         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
19937         if haven't found reduction or nested cycle operand, rather than
19938         asserting we must find it.
19940         PR tree-optimization/56381
19941         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
19942         to fold_build3.
19944 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
19945             Jakub Jelinek  <jakub@redhat.com>
19947         PR target/52555
19948         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
19949         (swap_optab_enable): Same.
19950         (init_all_optabs): Use argument instead of global.
19951         * tree.h (struct tree_optimization_option): New field target_optabs.
19952         * expr.h (init_all_optabs): Add argument to prototype.
19953         (TREE_OPTIMIZATION_OPTABS): New.
19954         (save_optabs_if_changed): Protoize.
19955         * optabs.h: Declare this_fn_optabs.
19956         * optabs.c (save_optabs_if_changed): New.
19957         Declare this_fn_optabs.
19958         (init_optabs): Add argument to init_all_optabs() call.
19959         * function.c (invoke_set_current_function_hook): Handle per
19960         function optabs.
19961         * function.h (struct function): New field optabs.
19962         * config/mips/mips.c (mips_set_mips16_mode): Handle when
19963         optimization_current_node has changed.
19964         * target-globals.h (save_target_globals_default_opts): Protoize.
19965         * target-globals.c (save_target_globals_default_opts): New.
19967 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19969         PR target/56347
19970         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
19971         registers %fr12 and %fr12R as call used.
19973         PR target/56214
19974         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
19975         and HImode, require all displacements to be an integer multiple of
19976         their mode size.
19977         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
19978         only allow QImode and HImode when reload is in progress and strict is
19979         true.  Likewise for symbolic addresses.  Use base14_operand to check
19980         displacements in REG+BASE addresses.
19982 2013-02-18  Richard Biener  <rguenther@suse.de>
19984         PR tree-optimization/56366
19985         * tree-vect-loop.c (get_initial_def_for_induction): Properly
19986         handle sign-conversion of outer-loop initial induction value.
19988 2013-02-18  Richard Biener  <rguenther@suse.de>
19990         PR middle-end/56349
19991         * cfghooks.c (merge_blocks): If we merge a latch into another
19992         block adjust references to it.
19993         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
19994         (verify_loop_structure): Verify that a recorded latch is in fact
19995         a latch.
19997 2013-02-18  Richard Biener  <rguenther@suse.de>
19999         PR tree-optimization/56321
20000         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
20001         order SSA name release and virtual operand unlinking.
20003 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20005         * config/microblaze/microblaze.md (save_stack_block): Define.
20006         (restore_stack_block): Likewise.
20008 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20010         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
20011         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
20012         * config/microblaze/microblaze.c (microblaze_option_override):
20013         Bail out early for PIC modes when target does not support PIC.
20015 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20017         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
20018         Replace with a microblaze version.
20019         (microblaze_trampoline_init): Adapt for microblaze.
20020         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
20021         microblaze.
20023 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
20024             Dodji Seketeli  <dodji@redhat.com>
20026         PR asan/56330
20027         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
20028         (instrument_mem_region_access): Do not forget to always put
20029         instrumentation of the of 'base' and 'base + len' in a "if (len !=
20030         0) statement, even for cases where either 'base' or 'base + len'
20031         are not instrumented -- because they have been previously
20032         instrumented.  Simplify the logic by putting all the statements
20033         instrument 'base + len' inside a sequence, and then insert that
20034         sequence right before the current insertion point.  Then, to
20035         instrument 'base + len', just get an iterator on that statement.
20036         And do not forget to update the pointer to iterator the function
20037         received as argument.
20039 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
20041         PR rtl-optimization/56348
20042         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
20044 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
20046         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
20047         (clean_graph_dump_file): Pass base to start_graph_dump.
20049 2013-02-14  Richard Henderson  <rth@redhat.com>
20051         PR target/55941
20052         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
20054 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
20056         * collect2-aix.h: Define F_LOADONLY.
20058 2013-02-14  Richard Biener  <rguenther@suse.de>
20060         PR lto/50494
20061         * varasm.c (output_constant_def_1): Get the decl representing
20062         the constant as argument.
20063         (output_constant_def): Wrap output_constant_def_1.
20064         (make_decl_rtl): Use output_constant_def_1 with the decl
20065         representing the constant.
20066         (build_constant_desc): Optionally re-use a decl already
20067         representing the constant.
20068         (tree_output_constant_def): Adjust.
20070 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
20072         Fix an asan crash
20073         * asan.c (instrument_builtin_call):  Really put the length of the
20074         second source argument into src1_len.
20076 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
20078         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
20079         argument.  If it is false, don't create edge from then_bb to
20080         fallthru_bb.
20081         (insert_if_then_before_iter): Pass true to it.
20082         (build_check_stmt): Pass false to it.
20083         (transform_statements): Flush hash table only on extended basic
20084         block boundaries, rather than at the beginning of every bb.
20085         Don't flush hash table on nonfreeing_call_p calls.
20086         * tree-flow.h (nonfreeing_call_p): New prototype.
20087         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
20089 2013-02-13  David S. Miller  <davem@davemloft.net>
20091         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
20093 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
20095         PR target/56184
20096         * ira.c (max_regno_before_ira): Move from ...
20097         (ira): ... here.
20098         (fix_reg_equiv_init): Use max_regno_before_ira instead of
20099         vec_safe_length.
20101 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
20103         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
20105 2013-02-13  Richard Biener  <rguenther@suse.de>
20107         PR lto/56295
20108         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
20109         globals in MEM_REFs.
20111 2013-02-13  Richard Biener  <rguenther@suse.de>
20113         * loop-init.c (loop_optimizer_init): Clear loop state when
20114         re-initializing preserved loops.
20115         * loop-unswitch.c (unswitch_single_loop): Return whether
20116         we unswitched the loop.  Do not verify loop state here.
20117         (unswitch_loops): When we unswitched a loop discover new loops.
20119 2013-02-13  Kostya Serebryany  <kcc@google.com>
20121         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
20122         on x86_64 linux.
20123         * sanitizer.def: Rename __asan_init to __asan_init_v1.
20125 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
20127         Avoid instrumenting duplicated memory access in the same basic block
20128         * Makefile.in (asan.o): Add new dependency on hash-table.h
20129         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
20130         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
20131         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
20132         (free_mem_ref_resources, has_mem_ref_been_instrumented)
20133         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
20134         (get_mem_ref_of_assignment): New functions.
20135         (get_mem_refs_of_builtin_call): Extract from
20136         instrument_builtin_call and tweak a little bit to make it fit with
20137         the new signature.
20138         (instrument_builtin_call): Use the new
20139         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
20140         of is_gimple_builtin_call.
20141         (instrument_derefs, instrument_mem_region_access): Insert the
20142         instrumented memory reference into the hash table.
20143         (maybe_instrument_assignment): Renamed instrument_assignment into
20144         this, and change it to advance the iterator when instrumentation
20145         actually happened and return true in that case.  This makes it
20146         homogeneous with maybe_instrument_assignment, and thus give a
20147         chance to callers to be more 'regular'.
20148         (transform_statements): Clear the memory reference hash table
20149         whenever we enter a new BB, when we cross a function call, or when
20150         we are done transforming statements.  Use
20151         maybe_instrument_assignment instead of instrumentation.  No more
20152         need to special case maybe_instrument_assignment and advance the
20153         iterator after calling it; it's now handled just like
20154         maybe_instrument_call.  Update comment.
20156 2013-02-13  Richard Biener  <rguenther@suse.de>
20158         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
20159         Fix loop discovery code.
20161 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
20163         PR inline-asm/56148
20164         * lra-constraints.c (process_alt_operands): Match early clobber
20165         operand with itself.  Check conflicts with earlyclobber only if
20166         the operand is not reloaded.  Prefer to reload conflicting operand
20167         if earlyclobber and matching operands are the same.
20169 2013-02-12  Richard Biener  <rguenther@suse.de>
20171         PR lto/56297
20172         * lto-streamer-out.c (write_symbol): Do not output symbols
20173         for hard register variables.
20175 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
20177         PR target/54222
20178         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
20179         (umulsidi3_insn, mulsidi3_insn): New insns.
20181 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
20183         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
20184         (struct tune_params): Add vec_costs field.
20185         * config/arm/arm.c (arm_builtin_vectorization_cost)
20186         (arm_add_stmt_cost): New functions.
20187         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
20188         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
20189         (arm_default_vec_cost): New struct of type cpu_vec_costs.
20190         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
20191         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
20192         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
20193         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
20195 2013-02-12  Richard Biener  <rguenther@suse.de>
20197         PR lto/56295
20198         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
20199         decls again if possible.
20201 2013-02-12  Richard Biener  <rguenther@suse.de>
20203         PR middle-end/56288
20204         * tree-ssa.c (verify_ssa_name): Fix check, move
20205         SSA_NAME_IN_FREE_LIST check up.
20207 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
20208             Steven Bosscher   <steven@gcc.gnu.org>
20210         PR rtl-optimization/56151
20211         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
20212         equal to op0 or op1, and last_insn pattern is CODE operation
20213         with MEM dest and one of the operands matches that MEM.
20215 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
20217         * doc/extend.texi: Document Function Multiversioning and "default"
20218         parameter string to target attribute.
20219         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
20220         target attribute parameter is "default".
20221         (ix86_compare_version_priority): Remove checks for target attribute.
20222         (ix86_mangle_function_version_assembler_name): Change error to sorry.
20223         Remove check for target attribute equal to NULL. Add assert.
20224         (ix86_generate_version_dispatcher_body): Change error to sorry.
20226 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
20227             Jack Howarth  <howarth@bromo.med.uc.edu>
20228             Patrick Marlier  <patrick.marlier@gmail.com>
20230         PR libitm/55693
20231         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
20232         define ENDFILE_SPEC as TM_DESTRUCTOR.
20233         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
20235 2013-02-11  Alexander Potapenko  <glider@google.com>
20236             Jack Howarth  <howarth@bromo.med.uc.edu>
20237             Jakub Jelinek  <jakub@redhat.com>
20239         PR sanitizer/55617
20240         * config/darwin.c (cdtor_record): Rename ctor_record.
20241         (sort_cdtor_records): Rename sort_ctor_records.
20242         (finalize_dtors): New routine to sort destructors by
20243         priority before use in assemble_integer.
20244         (machopic_asm_out_destructor): Use finalize_dtors if needed.
20246 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
20248         PR rtl-optimization/56275
20249         * simplify-rtx.c (avoid_constant_pool_reference): Check that
20250         offset is non-negative and less than cmode size before
20251         calling simplify_subreg.
20253 2013-02-11  Richard Biener  <rguenther@suse.de>
20255         PR tree-optimization/56264
20256         * cfgloop.h (fix_loop_structure): Adjust prototype.
20257         * loop-init.c (fix_loop_structure): Return the number of
20258         newly discovered loops.
20259         * tree-cfgcleanup.c (repair_loop_structures): When new loops
20260         are discovered, do a full loop-closed SSA rewrite.
20262 2013-02-11  Richard Biener  <rguenther@suse.de>
20264         PR tree-optimization/56273
20265         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
20266         first VRP run.
20267         (check_array_ref): Fix missing newline in dumps.
20268         (search_for_addr_array): Likewise.
20270 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
20272         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
20274 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
20276         PR target/56256
20277         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
20279 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
20281         PR rtl-optimization/56246
20282         * lra-constraints.c (simplify_operand_subreg): Try to reuse
20283         reload pseudo.
20284         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
20285         constraints are satisfied.
20287 2013-02-08  Jeff Law  <law@redhat.com>
20289         PR debug/53948
20290         * emit-rtl.c (reg_is_parm_p): New function.
20291         * regs.h (reg_is_parm_p): New prototype.
20292         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
20293         callee-clobbered registers.
20295 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
20297         PR target/56043
20298         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
20299         If there is no implicit builtin declaration, just return NULL.
20301 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
20303         * config/i386/sse.md (FMAMODEM): New mode iterator.
20304         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
20305         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
20307 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
20309         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
20310         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
20311         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
20313 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
20315         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
20316         (microblaze*-*-elf): Likewise.
20317         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
20318         LINK_SPEC.
20319         * config/microblaze/microblaze-c.c: Add builtin defines for
20320         _LITTLE_ENDIAN and _BIG_ENDIAN.
20321         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
20322         add to TARGET_DEFAULT flags.
20323         Expand ASM_SPEC and LINK_SPEC.
20324         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
20325         * config/microblaze/microblaze.md: Update extendsidi2 and
20326         movdi_internal instructions to use low-order / high-order reg
20327         print_operands.
20328         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
20329         options and inversemask / mask of LITTLE_ENDIAN.
20330         * config/microblaze/t-microblaze: Expand multilib options to
20331         include mlittle-endian (le) and update exceptions patterns.
20333 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
20335         PR rtl-optimization/56195
20336         * lra-constraints.c (get_reload_reg): Don't reuse regs
20337         if they have smaller mode than requested, if they have
20338         wider mode than requested, try to return a SUBREG.
20340         PR tree-optimization/56250
20341         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
20342         if type is unsigned and code isn't MULT_EXPR.
20344 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
20346         PR tree-optimization/56064
20347         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
20348         bits according to mode.
20349         * fixed-value.h (fixed_from_double_int)
20350         (const_fixed_from_double_int): Adjust comments.
20352 2013-02-08  Richard Biener  <rguenther@suse.de>
20354         PR lto/56231
20355         * lto-streamer.h (struct data_in): Remove current_file, current_line
20356         and current_col members.
20357         * lto-streamer-out.c (lto_output_location): Stream changed bits
20358         en-block for efficiency.
20359         * lto-streamer-in.c (clear_line_info): Remove.
20360         (lto_input_location): Cache current file, line and column
20361         globally via local statics.  Read changed bits en-block.
20362         (input_function): Do not call clear_line_info.
20363         (lto_read_body): Likewise.
20364         (lto_input_toplevel_asms): Likewise.
20366 2013-02-08  Michael Matz  <matz@suse.de>
20368         PR tree-optimization/52448
20369         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
20370         (nt_call_phase): New static.
20371         (add_or_mark_expr): Only mark accesses with newer phase than any
20372         call seen.
20373         (nonfreeing_call_p): New.
20374         (nt_init_block): Update nt_call_phase, mark blocks as visited.
20375         (nt_fini_block): Keep blocks marked as visited.
20376         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
20378 2013-02-08  Richard Biener  <rguenther@suse.de>
20380         * ira.c (ira): Free broken dominator information.
20382 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
20384         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
20386 2013-02-08  Marek Polacek  <polacek@redhat.com>
20388         * cfgloop.c (verify_loop_structure): Add more checking of headers.
20390 2013-02-08  Richard Biener  <rguenther@suse.de>
20392         PR middle-end/56181
20393         * cfgloop.h (flow_loops_find): Adjust.
20394         (bb_loop_header_p): Declare.
20395         * cfgloop.c (bb_loop_header_p): New function split out from ...
20396         (flow_loops_find): ... here.  Adjust function signature,
20397         support incremental loop structure update.
20398         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
20399         * cfgloopmanip.c (fix_loop_structure): Move ...
20400         * loop-init.c (fix_loop_structure): ... here.
20401         (apply_loop_flags): Split out from ...
20402         (loop_optimizer_init): ... here.
20403         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
20404         in incremental mode, only remove dead loops here.
20406 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
20408         PR target/54222
20409         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
20410         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
20411         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
20412         (*round<mode>3.libgcc): New insns for fixed-modes.
20413         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
20414         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
20415         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
20416         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
20417         implementations.  Define to __builtin_avr_absFX,
20418         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
20419         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
20420         __builtin_avr_countlsFX, respectively.
20421         * config/avr/avr-c.c (target.h): Include it.
20422         (enum avr_builtin_id): New enum.
20423         (avr_resolve_overloaded_builtin): New static function.
20424         (avr_register_target_pragmas): Use it to set
20425         targetm.resolve_overloaded_builtin.
20426         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
20427         tree nodes used by DEF_BUILTIN.
20428         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
20429         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
20430         <AVR_BUILTIN_xxBITS>: Same.
20432 2013-02-08  Richard Biener  <rguenther@suse.de>
20434         * cfgloop.c (verify_loop_structure): Properly handle
20435         a loop exiting to another loop header.
20436         * ira-int.h (ira_loops): Remove.
20437         * ira.c (ira_loops): Remove.
20438         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
20439         (do_reload): Use loop_optimizer_finalize.
20440         * ira-build.c (create_loop_tree_nodes): Use get_loops and
20441         number_of_loops to access the loop tree.
20442         (more_one_region_p): Likewise.
20443         (finish_loop_tree_nodes): Likewise.
20444         (rebuild_regno_allocno_maps): Likewise.
20445         (mark_loops_for_removal): Likewise.
20446         (mark_all_loops_for_removal): Likewise.
20447         (remove_unnecessary_regions): Likewise.
20448         (ira_build): Likewise.
20449         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
20451 2013-02-08  Richard Biener  <rguenther@suse.de>
20453         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
20454         * ipa-pure-const.c (analyze_function): Avoid calling
20455         mark_irreducible_loops twice.
20456         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
20458 2013-02-07  David S. Miller  <davem@davemloft.net>
20460         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
20461         on 'reg'.
20462         * var-tracking.c (vt_add_function_parameter): Test the presence of
20463         HAVE_window_save properly and do not remap argument registers when
20464         we have a leaf function.
20466 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
20468         PR bootstrap/56227
20469         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
20470         instead of "ll".
20471         * config/i386/i386.c (ix86_print_operand): Ditto.
20473 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
20475         * lra-constraints.c (process_alt_operands): Fix recently added comment.
20477 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
20479         PR rtl-optimization/56225
20480         * lra-constraints.c (process_alt_operands): Check that reload hard
20481         reg can hold value for strict_low_part.
20483 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
20485         PR debug/56154
20486         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
20487         dwarf2out_end_function.
20488         (in_first_function_p, maybe_at_text_label_p,
20489         first_loclabel_num_not_at_text_label): New variables.
20490         (dwarf2out_var_location): In the first function find out
20491         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
20492         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
20493         functions.
20495 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
20497         PR rtl-optimization/56178
20498         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
20499         SUBREG of a register.  Tidy up related block of code.
20500         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
20501         note if the source is a register or a SUBREG of a register.
20503 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
20505         PR target/56228
20506         * config/rs6000/rs6000.md (ptrm): New mode attr.
20507         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
20508         call_value_indirect_aix<pttrsize>,
20509         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
20510         m in constraints.
20512 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
20514         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
20515         if -bnortl. Convert to strcmp and strncmp.
20517 2013-02-07  Alan Modra  <amodra@gmail.com>
20519         PR target/54009
20520         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
20521         addresses won't wrap when offsetting.
20522         (rs6000_secondary_reload): Provide secondary reloads needed for
20523         wrapping LO_SUM addresses.
20525 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
20527         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
20528         MACH, just __MACH__.
20530 2013-02-06  Richard Biener  <rguenther@suse.de>
20532         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
20533         instead of calling fix_loop_structure.
20535 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
20537         PR middle-end/56217
20538         * omp-low.c (use_pointer_for_field): Return false if
20539         lower_send_shared_vars doesn't generate any copy-out code.
20541 2013-02-06  Tom de Vries  <tom@codesourcery.com>
20543         PR rtl-optimization/56131
20544         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
20545         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
20546         of the label is NULL.  Add comment.
20548 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
20550         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
20552         PR sanitizer/55374
20553         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
20554         (STATIC_LIBTSAN_LIBS): Likewise.
20555         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
20556         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
20557         is defined, don't add anything else beyond that.
20558         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
20559         (LINK_COMMAND_SPEC): Use them.
20561         PR tree-optimization/56205
20562         * tree-stdarg.c (check_all_va_list_escapes): Return true if
20563         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
20564         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
20566 2013-02-05  Richard Biener  <rguenther@suse.de>
20568         PR tree-optimization/53342
20569         PR tree-optimization/53185
20570         * tree-vectorizer.h (vect_check_strided_load): Remove.
20571         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
20572         not disallow peeling for vectorized strided loads.
20573         (vect_check_strided_load): Make static and simplify.
20574         (vect_analyze_data_refs): Adjust.
20575         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
20576         correctly when vectorizing strided loads.
20578 2013-02-05  Richard Biener  <rguenther@suse.de>
20580         * doc/install.texi: Refer to ISL, not PPL.
20582 2013-02-05  Jan Hubicka  <jh@suse.cz>
20584         PR tree-optimization/55789
20585         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
20587 2013-02-05  Jan Hubicka  <jh@suse.cz>
20589         PR tree-optimization/55789
20590         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
20591         the dead call anyway.
20593 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
20595         PR sanitizer/55374
20596         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
20598 2013-02-04  Alexander Potapenko  <glider@google.com>
20599             Jack Howarth  <howarth@bromo.med.uc.edu>
20600             Jakub Jelinek  <jakub@redhat.com>
20602         PR sanitizer/55617
20603         * config/darwin.c (sort_ctor_records): Stabilized qsort
20604         on constructor priority by using original position.
20605         (finalize_ctors): New routine to sort constructors by
20606         priority before use in assemble_integer.
20607         (machopic_asm_out_constructor): Use finalize_ctors if needed.
20609 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
20611         PR libstdc++/54314
20612         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
20613         about visibility on artificial decls.
20614         * config/sol2.c (solaris_assemble_visibility): Likewise.
20616 2013-02-04  Kai Tietz  <ktietz@redhat.com>
20618         PR target/56186
20619         * config/i386/i386.c (function_value_ms_64): Add additional valtype
20620         argument and improve checking of return-argument types for 16-byte
20621         modes.
20622         (ix86_function_value_1): Add additional valtype argument on call
20623         of function_value_64.
20624         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
20625         handling infunction_value_64 function.
20627 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
20629         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
20631 2013-02-04  Richard Biener  <rguenther@suse.de>
20633         PR tree-optimization/56188
20634         * tree-ssa-structalias.c (label_visit): Consider case with
20635         initially non-empty points-to set.
20636         (perform_var_substitution): Dump node mapping and clean up.
20638 2013-02-04  Richard Guenther  <rguenther@suse.de>
20640         PR lto/56168
20641         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
20642         node prevail as last resort.
20643         (lto_symtab_merge_decls): Remove guard on LTRANS here.
20644         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
20646 2013-02-04  Richard Biener  <rguenther@suse.de>
20648         PR tree-optimization/56113
20649         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
20650         Merge into ...
20651         (equiv_class_lookup_or_add): ... this.
20652         (label_visit): Adjust and fix error in previous patch.
20653         (perform_var_substitution): Adjust.
20655 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
20657         * config/sh/divtab.c: Fix formatting and comments throughout the file.
20658         * config/sh/sh4-300.md: Likewise.
20659         * config/sh/sh4a.md: Likewise.
20660         * config/sh/constraints.md: Likewise.
20661         * config/sh/sh.md: Likewise.
20662         * config/sh/netbsd-elf.h: Likewise.
20663         * config/sh/predicates.md: Likewise.
20664         * config/sh/sh-protos.h: Likewise.
20665         * config/sh/ushmedia.h: Likewise.
20666         * config/sh/linux.h: Likewise.
20667         * config/sh/sh.c: Likewise.
20668         * config/sh/superh.h: Likewise.
20669         * config/sh/elf.h: Likewise.
20670         * config/sh/sh4.md: Likewise.
20671         * config/sh/sh.h: Likewise.
20673 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20675         * config/pa/constraints.md: Adjust unused letters.  Change "T"
20676         constraint to match_test floating_point_store_memory_operand().
20677         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
20678         (base14_operand): New.
20679         (floating_point_store_memory_operand): New.
20680         (integer_store_memory_operand): Revise to use base14_operand and
20681         reg_plus_base_memory_operand.
20682         (move_dest_operand): Allow symbolic_memory_operands.
20683         (symbolic_memory_operand): Check for LO_SOM.
20684         (symbolic_operand): Change default case to break.
20685         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
20686         CONST_DOUBLE values to be reloaded by putting them into memory when
20687         the destination is a floating point register.
20688         (movdf): Remove code to handle CONST_DOUBLE.
20689         (movsf): Likewise.
20690         (reload_indf_r1): New.
20691         (reload_insf_r1): New.
20692         Consistently use "Q" and "T" constraints with integer and floating
20693         point move instructions, respectively.
20694         (movdi): Remove FAIL.
20695         Change predicate for source operand unamed DImode move from
20696         general_operand to move_src_operand.
20697         (umulsidi3): Change predicate for destination operand to
20698         register_operand.
20699         Likewise for similar unamed patterns.
20700         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
20701         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
20702         (hppa_legitimize_address): Simplify mask calculation.
20703         (pa_emit_move_sequence): Revised handling of secondary reloads from
20704         REG+D addresses for floating point loads and stores.  Directly handle
20705         loading CONST0_RTX (mode) to a floating point register.
20706         (pa_secondary_reload): Handle reloading DF and SFmode constant values
20707         to floating point registers.  Don't restrict secondary reloads to
20708         floating point registers to integer modes.  Revise some comments and
20709         cleanup some code.
20710         (TARGET_LEGITIMATE_ADDRESS_P): Define.
20711         (pa_legitimate_address_p): New.
20712         (pa_legitimize_reload_address): New.
20713         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
20714         (STRICT_REG_OK_FOR_BASE_P): New.
20715         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
20716         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
20718 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
20719             Andrew Dixie  <andrewd@gentrack.com>
20721         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
20722         flag set.
20724 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
20726         * expmed.c (extract_bit_field_1): Pass the full width of the
20727         structure to get_best_reg_extraction_insn.
20729 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
20731         PR target/54601
20732         * configure.ac (use_cxa_atexit): Add AIX.
20733         * configure: Regenerate.
20735         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
20737 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
20739         PR debug/54793
20740         * final.c (need_profile_function): New variable.
20741         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
20742         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
20743         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
20744         notes, targetm.asm_out.function_prologue doesn't emit anything,
20745         HAVE_prologue and profiler should be emitted before prologue,
20746         set need_profile_function instead of emitting it.
20747         (final_scan_insn): If need_profile_function, emit
20748         profile_function on the first NOTE_INSN_BASIC_BLOCK or
20749         NOTE_INSN_FUNCTION_BEG note.
20751 2013-02-01  Richard Henderson  <rth@redhat.com>
20753         * config/rs6000/rs6000.md (smulditi3): New.
20754         (umulditi3): New.
20756         * config/alpha/alpha.md (umulditi3): New.
20758 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
20760         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
20761         (ASM_OUTPUT_ALIGNED_LOCAL): New.
20763 2013-02-01  Richard Biener  <rguenther@suse.de>
20765         PR tree-optimization/56113
20766         * tree-ssa-structalias.c (label_visit): Reduce work for
20767         single-predecessor nodes.
20769 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
20771         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
20772         range isn't testing for zero.
20774 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
20776         PR middle-end/56113
20777         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
20779 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
20780             Nick Clifton  <nickc@redhat.com>
20782         * config/v850/constraints.md (Q): Define as a memory constraint.
20783         * config/v850/predicates.md (label_ref_operand): New predicate.
20784         (e3v5_shift_operand): New predicate.
20785         (ior_operator): New predicate.
20786         * config/v850/t-v850: Add e3v5 multilib.
20787         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
20788         (v850_gen_movdi): Prototype.
20789         * config/v850/v850.c: Add support for e3v5 architecture.
20790         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
20791         TARGET_V850E_UP.
20792         (construct_save_jarl): Add e3v5 long JARL support.
20793         (v850_adjust_insn_length): New function.  Adjust length of call
20794         insns when using e3v5 instructions.
20795         (v850_gen_movdi): New function: Generate instructions to move a
20796         DImode value.
20797         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
20798         (CPP_SPEC): Define __v850e3v5__ as appropriate.
20799         (TARGET_USE_FPU): Enable for e3v5.
20800         (CONST_OK_FOR_W): New macro.
20801         (ADJUST_INSN_LENGTH): Define.
20802         * config/v850/v850.md (UNSPEC_LOOP): Define.
20803         (attr cpu): Add v850e3v5.
20804         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
20805         (movdi): New pattern.
20806         (movdi_internal): New pattern.
20807         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
20808         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
20809         (cstoresf4): Likewise.
20810         (cstoredf4): Likewise.
20811         (insv): New pattern.
20812         (rotlso3_a): New pattern.
20813         (rotlsi3_b): New pattern
20814         (rotlsi3_v850e3v5): New pattern.
20815         (doloop_begin): New pattern.
20816         (fix_loop_counter): New pattern.
20817         (doloop_end): New pattern.
20818         (branch_normal): Add e3v5 long branch support.
20819         (branch_invert): Likewise.
20820         (branch_z_normal): Likewise.
20821         (branch_z_invert): Likewise.
20822         (branch_nz_normal): Likewise.
20823         (branch_nz_invert): Likewise.
20824         (call_internal_short): Add e3v5 register-indirect JARL support.
20825         (call_internal_long): Likewise.
20826         (call_value_internal_short): Likewise.
20827         (call_value_internal_long): Likewise.
20828         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
20829         (mloop): New option.
20830         * config.gcc: Add support for configuring v840e3v5 target.
20831         * doc/invoke.texi: Document new v850 specific command line options.
20833 2013-01-31  Paul Koning  <ni1d@arrl.net>
20835         PR debug/55059
20836         PR debug/54508
20837         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
20838         children if parent is a class.
20839         (prune_unused_types_prune): Don't add DW_AT_declaration.
20841 2013-01-31  Richard Biener  <rguenther@suse.de>
20843         PR tree-optimization/56157
20844         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
20845         match up operand with SLP child.
20847 2013-01-31  Jason Merrill  <jason@redhat.com>
20849         PR debug/54410
20850         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
20851         parameters the first time.
20852         (gen_scheduled_generic_parms_dies): Check completeness here.
20854 2013-01-31  Richard Biener  <rguenther@suse.de>
20856         PR middle-end/53073
20857         * common.opt (faggressive-loop-optimizations): New flag,
20858         enabled by default.
20859         * doc/invoke.texi (faggressive-loop-optimizations): Document.
20860         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
20861         infer_loop_bounds_from_undefined by it.
20863 2013-01-31  Richard Biener  <rguenther@suse.de>
20865         PR tree-optimization/56150
20866         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
20867         visit virtual operands.
20868         (find_uses_to_rename_bb): Likewise.
20870 2013-01-31  Richard Biener  <rguenther@suse.de>
20872         PR tree-optimization/56150
20873         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
20874         mixed store non-store stmts.
20876 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
20878         PR sanitizer/55374
20879         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
20880         LIBASAN_EARLY_SPEC is defined.
20881         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
20882         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
20883         before %o.
20884         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
20886         PR c++/55742
20887         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
20888         invalid args instead of ICEing on it.
20889         (ix86_valid_target_attribute_tree): Return error_mark_node if
20890         ix86_valid_target_attribute_inner_p failed.
20891         (ix86_valid_target_attribute_p): Return false only if
20892         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
20893         target("default") attribute.
20894         (sorted_attr_string): Change argument from const char * to tree,
20895         merge in all target attribute arguments rather than just one.
20896         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
20897         instead of free.  Avoid using strcat.
20898         (ix86_mangle_function_version_assembler_name): Mangle
20899         target("default") as if no target attribute is present.  Adjust
20900         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
20901         instead of xmalloc and XDELETEVEC instead of free.
20902         (ix86_function_versions): Don't return true if one of the decls
20903         doesn't have target attribute.  If they don't and one of the decls
20904         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
20905         sorted_attr_string caller.  Use XDELETEVEC instead of free.
20906         (ix86_supports_function_versions): Remove.
20907         (make_name): Fix up formatting.
20908         (make_dispatcher_decl): Remove resolver_name and its initialization.
20909         Avoid leaking memory.
20910         (is_function_default_version): Return true if there is
20911         target("default") attribute rather than no target attribute at all.
20912         (make_resolver_func): Avoid leaking memory.
20913         (ix86_generate_version_dispatcher_body): Likewise.
20914         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
20915         * target.def (supports_function_versions): Remove.
20916         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
20917         * doc/tm.texi: Regenerated.
20919 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
20921         PR rtl-optimization/56144
20922         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
20923         for values with side effects.
20925 2013-01-30  Richard Biener  <rguenther@suse.de>
20927         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
20928         (sparseset_pop): Likewise.
20929         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
20930         to be able to use quick_push in the worker loop.
20932 2013-01-30  Marek Polacek  <polacek@redhat.com>
20934         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
20936 2013-01-30  Richard Biener  <rguenther@suse.de>
20938         PR lto/56147
20939         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
20941 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
20943         PR tree-optimization/56064
20944         * fixed-value.c (fixed_from_double_int): New function.
20945         * fixed-value.h (fixed_from_double_int): New prototype.
20946         (const_fixed_from_double_int): New static inline function.
20947         * fold-const.c (native_interpret_fixed): New static function.
20948         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
20949         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
20950         (native_encode_fixed): New static function.
20951         (native_encode_expr) <FIXED_CST>: Use it.
20952         (native_interpret_int): Move double_int worker code to...
20953         * double-int.c (double_int::from_buffer): ...this new static method.
20954         * double-int.h (double_int::from_buffer): Prototype it.
20956 2013-01-30  Richard Biener  <rguenther@suse.de>
20958         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
20959         New pointer-map and obstack.
20960         (init_alias_vars): Allocate pointer-map and obstack.
20961         (delete_points_to_sets): Free them.
20962         (find_what_var_points_to): Cache result.
20963         (find_what_p_points_to): Adjust for changed interface of
20964         find_what_var_points_to.
20965         (compute_points_to_sets): Likewise.
20966         (ipa_pta_execute): Likewise.
20968 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20970         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
20971         * configure: Regenerate.
20972         * config.in: Regenerate.
20973         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
20974         #nobits/#progbits if supported.
20976 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
20978         PR target/56121
20979         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
20980         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
20981         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
20983 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20985         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
20986         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
20988 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20990         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
20991         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
20993 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
20995         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
20996         declaration.
20997         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
20998         * config/arm/cortex-a7.md: New bypasses using
20999         arm_mac_accumulator_is_result.
21001 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
21003         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
21004         (cortex_a7_neon_mla): Likewise.
21005         (cortex_a7_fpfmad): New reservation.
21006         (cortex_a7_fpmacs): Use ffmas and update required units.
21007         (cortex_a7_fpmuld): Update required units and latency.
21008         (cortex_a7_fpmacd): Likewise.
21009         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
21010         (cortex_a7_neon). Likewise.
21011         (bypass) Update participating units.
21013 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
21015         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
21016         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
21017         from fmac to ffma.
21018         * config/arm/vfp11.md (vfp_farith): Use ffmas.
21019         (vfp_fmul): Use ffmad.
21020         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
21021         (cortex_r4_fmacd): Use ffmad.
21022         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
21023         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
21024         (cortex_a9_fmacd): Use ffmad.
21025         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
21026         (cortex_a8_vfp_macd): Use ffmad.
21027         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
21028         (cortex_a5_fpmacd): Use ffmad.
21029         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
21030         (cortex_a15_vfp_macd): Use ffmad.
21031         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
21033 2013-01-29  Jason Merrill  <jason@redhat.com>
21035         PR libstdc++/54314
21036         * varasm.c (default_assemble_visibility): Don't warn about
21037         visibility on artificial decls.
21039 2013-01-29  Richard Biener  <rguenther@suse.de>
21041         PR tree-optimization/56113
21042         * tree-ssa-structalias.c (equiv_class_lookup): Also return
21043         the bitmap leader.
21044         (label_visit): Free duplicate bitmaps and record the leader instead.
21045         (perform_var_substitution): Adjust.
21047 2013-01-29  Richard Biener  <rguenther@suse.de>
21049         PR tree-optimization/55270
21050         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
21051         the CFG, schedule loops for fixup.
21053 2013-01-29  Nick Clifton  <nickc@redhat.com>
21055         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
21056         SP_REG.
21058 2013-01-28  Leif Ekblad  <leif@rdos.net>
21060         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
21061         * config/i386/i386.h (TARGET_RDOS): New macro.
21062         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
21063         * config/i386/i386.c (ix86_option_override_internal): For 64bit
21064         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
21065         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
21066         DEFAULT_LARGE_SECTION_THRESHOLD.
21067         * config/i386/i386.md (R14_REG, R15_REG): New constants.
21068         * config/i386/rdos.h: New file.
21069         * config/i386/rdos64.h: New file.
21071 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
21073         PR other/54814
21074         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
21075         TEST_HARD_REG_BIT.
21077 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
21079         PR rtl-optimization/56117
21080         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
21081         call cselib_lookup_from_insn on the MEM before calling
21082         add_insn_mem_dependence.
21084 2013-01-28  Richard Biener  <rguenther@suse.de>
21086         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
21087         to a stmt that didn't have one.
21088         (copy_phis_for_bb): Likewise for PHI arguments.
21089         (copy_debug_stmt): Likewise for debug stmts.
21091 2013-01-28  Richard Biener  <rguenther@suse.de>
21093         PR tree-optimization/56034
21094         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
21095         (partition_builtin_p): Adjust.
21096         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
21097         it is the last partition.
21098         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
21099         up the vertex for the definition.
21100         (classify_partition): Classify whether a partition is a
21101         PKIND_REDUCTION, thus has uses outside of the loop.
21102         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
21103         Merge all PKIND_REDUCTION partitions into the last partition.
21104         (tree_loop_distribution): Seed partitions from reductions as well.
21106 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
21108         PR tree-optimization/56125
21109         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
21110         pow(x,c) into sqrt(x) * powi(x, n/2) or
21111         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
21112         optimizing for size.
21113         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
21114         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
21115         integer.
21117         PR tree-optimization/56094
21118         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
21119         to UNKNOWN_LOCATION while gimplifying expr.
21121 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
21123         PR target/56114
21124         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
21125         operand 0 in movabs insn template for -masm=intel asm alternative.
21126         (*movabs<mode>_2): Ditto for operand 1.
21128 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
21130         PR target/54663
21131         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
21132         of microblaze-c.o
21134 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
21136         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
21137         tm_file.
21139 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21141         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
21142         Undef to avoid warning.
21144 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
21146         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
21147         * configure: Regenerate.
21149 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
21151         PR tree-optimization/56098
21152         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
21153         for stmts with volatile ops.
21154         (cond_store_replacement): Don't optimize if assign has volatile ops.
21155         (cond_if_else_store_replacement_1): Don't optimize if either
21156         then_assign or else_assign have volatile ops.
21157         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
21158         volatile ops.
21160 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
21162         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
21164 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
21166         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
21167         missing ':' in asm example.
21169 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
21171         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
21172         entries into lane and laneq entries.
21173         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
21174         Remove AdvSIMD scalar modes.
21175         (aarch64_sq<r>dmulh_laneq<mode>): New.
21176         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
21177         modes.
21178         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
21179         builtin implementations to relfect changes in RTL in aarch64-simd.md.
21180         * config/aarch64/iterators.md (VCOND): New.
21181         (VCONQ): New.
21183 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
21185         PR target/54222
21186         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
21187         Add NULL LIBNAME argument to existing definitions.
21188         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
21189         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
21190         * config/avr/avr.c (DEF_BUILTIN): Same.
21191         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
21192         (avr_expand_builtin): Expand to a vanilla call if a libgcc
21193         implementation is available (DECL_ASSEMBLER_NAME is set).
21194         (avr_fold_absfx): New static function.
21195         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
21196         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
21197         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
21198         AVR_BUILTIN_ABSLLK.
21199         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
21200         (abshk, absk, abslk, absllk): Provide as static inline functions.
21202 2013-01-25  Marek Polacek  <polacek@redhat.com>
21204         PR tree-optimization/56035
21205         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
21207 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
21209         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
21210         (*movtf_internal_rex64): Add (!o,C) alternative
21211         (*movxf_internal_rex64): Ditto.
21212         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
21214 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
21216         * doc/invoke.texi: fix typo.
21217         * doc/objc.texi: fix typo.
21219 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
21221         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
21222         for the first two alternatives.
21224 2013-01-24  Diego Novillo  <dnovillo@google.com>
21226         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
21227         (ggc-zone.o): Remove.
21228         * configure.ac: Remove option --with-gc.
21229         * configure: Re-generate.
21230         * doc/install.texi: Remove documentation for --with-gc.
21231         * gengtype.c (write_enum_defn): Remove.  Update all users.
21232         (write_Types_process_field): Remove generation of gt_e_* argument.
21233         (output_type_enum): Remove.  Update all users.
21234         (write_enum_defn): Remove.  Update all users.
21235         (enum alloc_zone): Remove.  Update all users.
21236         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
21237         * ggc-common.c (ggc_splay_alloc): Remove first argument.
21238         Update all callers.
21239         (struct ptr_data): Remove field TYPE.  Update all users.
21240         (gt_pch_note_object): Remove argument TYPE.  Update all users.
21241         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
21242         Update all users.
21243         * ggc-none.c (ggc_alloc_typed_stat): Remove.
21244         (struct alloc_zone): Remove.
21245         (ggc_internal_alloc_zone_stat): Remove.
21246         (ggc_internal_cleared_alloc_zone_stat): Remove.
21247         * ggc-page.c (ggc_alloc_typed_stat): Remove.
21248         (ggc_pch_count_object): Remove last argument.  Update all users.
21249         (ggc_pch_alloc_object): Remove last argument.  Update all users.
21250         (struct alloc_zone): Remove.
21251         * ggc-zone.c: Remove.
21252         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
21253         (struct alloc_zone): Remove.
21254         (ggc_alloc_typed_stat): Remove.
21255         (ggc_alloc_typed): Remove.
21256         (ggc_splay_alloc): Remove first argument.
21257         (rtl_zone): Remove.  Update all users.
21258         (tree_zone): Remove.  Update all users.
21259         (tree_id_zone): Remove.  Update all users.
21260         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
21261         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
21262         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
21263         * tree-ssanames.c: Remove references to zone allocator in comments.
21265 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
21267         * config/avr/avr.c (avr_out_fract): Make register numbers that
21268         might be outside of source operand signed.
21270 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
21272         * config/i386/constraints.md (Yf): New constraint.
21273         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
21274         of f constraint to conditionaly disable x87 register preferences.
21275         (*movdf_internal): Ditto.
21276         (*movsf_internal): Ditto.
21278 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
21280         PR inline-asm/55934
21281         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
21282         that have operands with impossible constraints.
21283         Add a FIXME for a speed-up opportunity.
21284         * lra-constraints.c (process_alt_operands): Verify that a class
21285         selected from constraints on asms is valid for the operand mode.
21286         (curr_insn_transform): Remove incorrect comment.
21288 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
21290         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
21291         TOC operand is a valid symbol ref in the constant pool.
21293 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
21295         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
21297 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
21299         PR target/54222
21300         * config/avr/stdfix.h: New file.
21301         * t-avr (stdfix-gcc.h): New rule to build it.
21302         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
21304 2013-01-23  Kostya Serebryany  <kcc@google.com>
21306         * config/darwin.h: remove dependency on
21307         CoreFoundation (asan on Mac OS).
21309 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
21311         PR target/49069
21312         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
21313         instead of cmpdi_operand for first comparison operand.
21314         Don't assert that comparison operands aren't both constants.
21316 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
21318         * doc/install.texi (Downloading the Source): Update references to
21319         downloading separate components.
21321 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
21323         * doc/extend.texi (__int128): Improve grammar.
21325 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
21327         PR target/56028
21328         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
21329         alternative to (o,r).
21330         (*movdi_internal_rex64): Remove (!o,n) alternative.
21331         (DImode immediate->memory splitter): Remove.
21332         (DImode immediate->memory peephole2): Remove.
21333         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
21334         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
21335         alternative to (!o,*r).
21336         (*movtf_internal_sse): New pattern.
21337         (*movxf_internal_rex64): New pattern.
21338         (*movxf_internal): Disable for TARGET_64BIT.
21339         (*movdf_internal_rex64): Remove (!o,F) alternative.
21341 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
21343         PR middle-end/56074
21344         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
21345         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
21346         * tree-vect-loop-manip.c (find_loop_location): Also ignore
21347         stmt locations where LOCATION_LOCUS of the stmt location is
21348         UNKNOWN_LOCATION or BUILTINS_LOCATION.
21350         PR target/55686
21351         * config/i386/i386.md (UNSPEC_STOS): New.
21352         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
21353         *strsetqi_1): Add UNSPEC_STOS.
21355 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
21357         PR c++/56067
21358         * doc/invoke.texi: Remove left over -Wsynth example.
21360 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
21362         PR tree-optimization/56051
21363         * fold-const.c (fold_binary_loc): Don't fold
21364         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
21365         a narrowing conversion, or widening conversion from signed
21366         to unsigned.
21368 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
21370         PR rtl-optimization/56023
21371         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
21372         dependent on debug instruction.
21374 2013-01-21  Martin Jambor  <mjambor@suse.cz>
21376         PR middle-end/56022
21377         * function.c (allocate_struct_function): Call
21378         invoke_set_current_function_hook earlier.
21380 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
21382         * reload1.c (init_reload): Only initialize reload_obstack
21383         during the first call.
21385 2013-01-21  Marek Polacek  <polacek@redhat.com>
21387         * cfgloop.c (verify_loop_structure): Fix up grammar.
21389 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
21391         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
21392         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
21394 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
21396         PR target/56058
21397         * config/arm/marvell-pj4.md: Update copyright year.
21398         Fix up use of alu to alu_reg and simple_alu_imm.
21400 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
21402         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
21404 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
21406         PR target/55433
21407         * lra-constraints.c (curr_insn_transform): Don't reuse original
21408         insn for secondary memory move when memory mode should be different.
21410 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
21412         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
21413         atomic_storedi_1): New patterns.
21415 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
21417         btver2 pipeline descriptions.
21418         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
21419         descriptions.
21420         * config/i386/i386.md (btver2_decode): New type attributes.
21421         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
21422         type attributes.
21423         * config/i386/btver2.md: New file describing btver2 pipelines.
21425 2013-01-19  Andrew Pinski  <apinski@cavium.com>
21427         PR tree-optimization/52631
21428         * tree-ssa-sccvn (visit_use): Before looking up the original
21429         statement, try looking up the simplified expression.
21431 2013-01-19  Anthony Green  <green@moxielogic.com>
21433         * config/moxie/moxie.c (moxie_expand_prologue): Set
21434         current_function_static_stack_size.
21436 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
21438         PR tree-optimization/56029
21439         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
21440         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
21442 2013-01-18  Sharad Singhai  <singhai@google.com>
21444         PR tree-optimization/55995
21445         * dumpfile.c (dump_loc): Print location only if available.
21446         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
21448 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
21450         PR target/55433
21451         * lra-constraints.c (curr_insn_transform): Reuse original insn for
21452         secondary memory move.
21453         (inherit_reload_reg): Use rclass instead of cl for
21454         check_secondary_memory_needed_p.
21456 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
21458         PR middle-end/56015
21459         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
21460         the case where writing real complex part of target modifies op1.
21462 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
21464         * config/aarch64/aarch64-simd.md
21465         (aarch64_vcond_internal<mode>): Handle unordered cases.
21466         * config/aarch64/iterators.md (v_cmp_result): New.
21468 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
21469             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
21471         * config/arm/marvell-pj4.md: New file.
21472         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
21473         * config/arm/arm.md (generic_sched): Add marvell_pj4.
21474         (generic_vfp): Likewise.
21475         * config/arm/arm-cores.def: Add marvell-pj4.
21476         * config/arm/arm-tune.md: Regenerate.
21477         * config/arm/arm-tables.opt: Regenerate.
21478         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
21479         * doc/invoke.texi: Document marvell-pj4.
21481 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
21483         * config/aarch64/arm_neon.h: Map scalar types to standard types.
21485 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
21487         PR debug/54114
21488         PR debug/54402
21489         PR debug/49888
21490         * var-tracking.c (negative_power_of_two_p): New.
21491         (global_get_addr_cache, local_get_addr_cache): New.
21492         (get_addr_from_global_cache, get_addr_from_local_cache): New.
21493         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
21494         heading comment.
21495         (vt_stack_offset_p): Remove.
21496         (vt_canon_true_dep): Always canonicalize loc's address.
21497         (clobber_overlapping_mems): Make sure we have a MEM.
21498         (local_get_addr_clear_given_value): New.
21499         (val_reset): Clear local cached entries.
21500         (compute_bb_dataflow): Create and release the local cache.
21501         Disable duplicate MEMs clobbering.
21502         (emit_notes_in_bb): Clobber MEMs likewise.
21503         (vt_emit_notes): Create and release the local cache.
21504         (vt_initialize, vt_finalize): Create and release the global
21505         cache, respectively.
21506         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
21508 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
21510         PR libmudflap/53359
21511         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
21512         not found in the symtab.
21514 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
21516         PR debug/56006
21517         PR rtl-optimization/55547
21518         PR rtl-optimization/53827
21519         PR debug/53671
21520         PR debug/49888
21521         * alias.c (offset_overlap_p): New, factored out of...
21522         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
21523         the conservative special case for symbolic constants.  Don't
21524         adjust zero sizes on alignment.
21526 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
21528         PR rtl-optimization/52573
21529         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
21530         REG_UNUSED for the same register.
21532 2013-01-17  Richard Biener  <rguenther@suse.de>
21533             Marek Polacek  <polacek@redhat.com>
21535         PR rtl-optimization/55833
21536         * loop-unswitch.c (unswitch_loops): Move loop verification...
21537         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
21538         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
21539         Set it to true when we're removing a loop from hierarchy tree in
21540         an irreducible region.
21541         (fix_bb_placements): Adjust caller.
21542         (fix_loop_placements): Likewise.
21544 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
21546         * config/avr/builtins.def (DEF_BUILTIN): Factor out
21547         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
21548         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
21549         Remove ID.  Adjust comments.
21550         * config/avr/avr-c.c (avr_builtin_name): Remove.
21551         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
21552         * config/avr/avr.c (avr_tolower): New static function.
21553         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
21554         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
21555         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
21556         default expansion.
21558 2013-01-17  Jan Hubicka  <jh@suse.cz>
21560         PR tree-optimization/55273
21561         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
21563 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
21565         PR target/55981
21566         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
21567         store through atomic_store<mode>_1.
21568         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
21570 2013-01-17  Martin Jambor  <mjambor@suse.cz>
21572         PR tree-optimizations/55264
21573         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
21574         for virtual methods.
21575         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
21576         virtual methods before inlining is over.
21577         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
21578         virtual functions.
21579         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
21580         non-virtual.
21582 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
21584         PR rtl-optimization/56005
21585         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
21586         pending reads for prefetch.
21588 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
21590         * config/aarch64/aarch64.md
21591         (*cstoresi_neg_uxtw): New pattern.
21592         (*cmovsi_insn_uxtw): New pattern.
21593         (*<optab>si3_uxtw): New pattern.
21594         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
21595         (*<optab>si3_insn_uxtw): New pattern.
21596         (*bswapsi2_uxtw): New pattern.
21598 2013-01-16  Richard Biener  <rguenther@suse.de>
21600         * tree-inline.c (tree_function_versioning): Remove set but
21601         never used variable.
21603 2013-01-16  Richard Biener  <rguenther@suse.de>
21605         PR tree-optimization/55964
21606         * tree-flow.h (rename_variables_in_loop): Remove.
21607         (rename_variables_in_bb): Likewise.
21608         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
21609         (copy_loop_before): Adjust and delete update-ssa status.
21610         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
21611         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
21612         (rename_variables_in_loop): Remove.
21613         (slpeel_update_phis_for_duplicate_loop): Likewise.
21614         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
21615         use available cfg machinery instead of duplicating it.
21616         Update PHI nodes and perform poor-mans SSA update here.
21617         (slpeel_tree_peel_loop_to_edge): Adjust.
21619 2013-01-16  Richard Biener  <rguenther@suse.de>
21621         PR tree-optimization/54767
21622         PR tree-optimization/53465
21623         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
21624         (vrp_visit_phi_node): For PHI arguments coming via backedges
21625         drop all symbolical range information.
21626         (execute_vrp): Compute backedges.
21628 2013-01-16  Richard Biener  <rguenther@suse.de>
21630         * doc/install.texi: Update CLooG and ISL requirements to
21631         0.18.0 and 0.11.1.
21633 2013-01-16  Christian Bruel  <christian.bruel@st.com>
21635         PR target/55301
21636         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
21637         (broken_move): Handle UNSPECV_SP_SWITCH_B.
21638         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
21640 2013-01-16  DJ Delorie  <dj@redhat.com>
21642         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
21643         (UNSPECV_SP_SWITCH_E): New.
21644         (sp_switch_1): Change to an unspec.
21645         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
21646         replace $r15.
21648 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
21650         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
21651         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
21652         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
21653         (expand_mem_thread_fence): Ditto.
21654         (expand_mem_signal_fence): Ditto.
21655         (expand_atomic_load): Ditto.
21656         (expand_atomic_store): Ditto.
21658 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
21660         PR rtl-optimization/55547
21661         PR rtl-optimization/53827
21662         PR debug/53671
21663         PR debug/49888
21664         * alias.c (memrefs_conflict_p): Set sizes to negative after
21665         AND adjustments.
21667 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
21669         PR target/55940
21670         * function.c (thread_prologue_and_epilogue_insns): Always
21671         add crtl->drap_reg to set_up_by_prologue.set, even if
21672         stack_realign_drap is false.
21674 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
21676         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
21677         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
21678         *call): Fix indention.
21680 2013-01-15  Tom de Vries  <tom@codesourcery.com>
21682         PR target/55876
21683         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
21684         Update comment.
21686 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
21688         PR rtl-optimization/55153
21689         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
21691 2013-01-15  Martin Jambor  <mjambor@suse.cz>
21693         PR tree-optimization/55920
21694         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
21695         accesses as grp_to_be_debug_replaced.
21697 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
21699         PR tree-optimization/55920
21700         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
21701         there is non-useless type conversion needed from debug rhs to lhs,
21702         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
21704 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
21705             Mikael Pettersson  <mikpe@it.uu.se>
21707         PR target/43961
21708         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
21709         Thumb.
21710         (ASM_OUTPUT_CASE_LABEL): Remove.
21711         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
21712         * final.c (shorten_branches): Update alignment of labels before
21713         jump tables if CASE_VECTOR_SHORTEN_MODE.
21715 2013-01-15  Richard Biener  <rguenther@suse.de>
21717         PR bootstrap/55961
21718         * system.h: Do not include gmp.h for building host tools.
21720 2013-01-15  Richard Biener  <rguenther@suse.de>
21722         PR middle-end/55882
21723         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
21724         account for bitpos when computing alignment.
21726 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
21728         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
21729         (ix86_target_macros_internal): Likewise.
21731         * config/i386/i386.c (m_CORE2I7): Removed.
21732         (m_CORE_HASWELL): New macro.
21733         (m_CORE_ALL): Likewise.
21734         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
21735         (initial_ix86_arch_features): Likewise.
21736         (processor_target_table): Initializations for Core avx2.
21737         (cpu_names): New names "core-avx2".
21738         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
21739         PROCESSOR_CORE_HASWELL.
21740         (ix86_issue_rate): New case.
21741         (ia32_multipass_dfa_lookahead): Likewise.
21742         (ix86_sched_init_global): Likewise.
21744         * config/i386/i386.h (TARGET_HASWELL): New macro.
21745         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
21746         (processor_type): New PROCESSOR_HASWELL.
21748 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
21750         PR tree-optimization/55955
21751         * tree-vect-loop.c (vectorizable_reduction): Give up early on
21752         *SHIFT_EXPR and *ROTATE_EXPR codes.
21754         PR tree-optimization/48766
21755         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
21756         -ftrapv disable -fwrapv.
21758 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
21760         PR target/55974
21761         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
21762         etc. to 1 and not to __flash.
21763         Use LL suffix for __INT24_MAX__ with -mint8.
21764         Use ULL suffix for __UINT24_MAX__ with -mint8.
21766 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
21768         * config/avr/avr-arch.h
21769         (struct base_arch_s): Use typedef avr_arch_t instead.
21770         (struct arch_info_s): Use typedef avr_arch_info_t instead.
21771         (struct mcu_type_s): Use typedef avr_mcu_t instead.
21772         * config/avr/avr.c: Same.
21773         * config/avr/avr-devices.c: Same.
21774         * config/avr/driver-avr.c: Same.
21775         * config/avr/gen-avr-mmcu-texi.c: Same.
21776         * config/avr/avr-mcus.def: Adjust comment.
21778 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
21780         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
21781         * config/aarch64/iterators.md (VALLDI): New.
21783 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
21784             Andi Kleen  <ak@linux.intel.com>
21786         PR target/55948
21787         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
21788         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
21789         memmodel flag.
21791 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
21793         * config/avr/avr-stdint.h: Remove trailing blanks.
21794         * config/avr/avr-log.h: Same.
21795         * config/avr/avr-arch.h: Same.
21796         * config/avr/avr-devices.c: Same.
21797         * config/avr/avr-dimode.md: Same.
21798         * config/avr/predicates.md: Same.
21799         * config/avr/avr-c.c: Same.  And fix typo.
21801         * config/avr/avr-protos.h: Same.  And:
21802         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
21803         (init_cumulative_args): Rename to avr_init_cumulative_args.
21804         (expand_prologue): Rename to avr_expand_prologue.
21805         (expand_epilogue): Rename to avr_expand_epilogue.
21806         (adjust_insn_length): Rename to avr_adjust_insn_length.
21807         (notice_update_cc): Rename to avr_notice_update_cc.
21808         (final_prescan_insn): Rename to avr_final_prescan_insn.
21809         * config/avr/avr.c: Same.
21810         * config/avr/avr.h: Same.
21811         * config/avr/avr.md: Remove trailing blanks.
21812         (prologue): Use avr_expand_prologue.
21813         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
21815 2013-01-14  Richard Biener  <rguenther@suse.de>
21817         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
21818         verify_location, collect_subblocks): New functions.
21819         (verify_gimple_in_cfg): Verify that locations only reference
21820         BLOCKs in the functions BLOCK tree.
21822 2013-01-14  Richard Biener  <rguenther@suse.de>
21824         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
21825         PHI argument.
21826         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
21827         unshare reference.
21828         (insert_out_of_ssa_copy_on_edge): Likewise.
21829         (rewrite_close_phi_out_of_ssa): Likewise.
21830         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
21831         debug expressions.
21832         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
21833         propagated constants.
21834         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
21835         can not be shared.
21837 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
21839         * config/avr/avr-modes.def: Add GPL copyright notice.
21841 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
21843         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
21844         MEMMODEL_MASK to determine memory model.
21845         (atomic_store<mode>): Ditto from operands[2].
21846         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
21848 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
21850         PR fortran/55935
21851         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
21852         (fold_gimple_assign): Don't call unshare_expr here.
21853         (fold_ctor_reference): Call unshare_expr.
21855 2013-01-13  Terry Guo  <terry.guo@arm.com>
21857         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
21858         * doc/fragments.texi: Document MULTILIB_REUSE.
21859         * gcc.c (multilib_reuse): New internal spec.
21860         (set_multilib_dir): Also search multilib from multilib_reuse.
21861         * genmultilib (tmpmultilib3): Refactor code.
21862         (tmpmultilib4): Ditto.
21863         (multilib_reuse): New multilib argument.
21865 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
21867         * Makefile.in: Update copyright.
21869 2013-01-12  Tom de Vries  <tom@codesourcery.com>
21871         PR middle-end/55890
21872         * calls.c (expand_call): Check if arg_nr is valid.
21874 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
21876         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
21877         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
21878         documentation.  Add missing '__' in front of
21879         __builtin_ia32_packssdw256.
21881 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21883         PR target/55719
21884         * config/s390/s390.c (s390_preferred_reload_class): Do not return
21885         NO_REGS for larl operands.
21886         (s390_reload_larl_operand): Use s390_load_address instead of
21887         emit_move_insn.
21889 2013-01-11  Richard Biener  <rguenther@suse.de>
21891         * tree-cfg.c (verify_node_sharing_1): Split out from ...
21892         (verify_node_sharing): ... here.
21893         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
21895 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
21897         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
21898         Substitute TREECHECKING.
21899         * configure: Regenerate.
21900         * Makefile.in (TREECHECKING): New.
21902 2013-01-11  Richard Guenther  <rguenther@suse.de>
21904         PR tree-optimization/44061
21905         * tree-vrp.c (extract_range_basic): Compute zero as
21906         value-range for __builtin_constant_p of function parameters.
21908 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
21910         Update copyright years.
21912 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
21914         PR rtl-optimization/55672
21915         * lra-eliminations.c (mark_not_eliminable): Permit addition with
21916         const to be eliminable.
21918 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
21920         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
21921         * configure: Regenerate.
21923 2013-01-10  Richard Biener  <rguenther@suse.de>
21925         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
21927 2013-01-10  Richard Biener  <rguenther@suse.de>
21929         PR bootstrap/55792
21930         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
21931         locations for virtual PHI arguments.
21932         (rewrite_update_phi_arguments): Likewise.
21934 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
21936         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
21937         on to assembler.
21939 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
21941         PR tree-optimization/55921
21942         * tree-complex.c (expand_complex_asm): New function.
21943         (expand_complex_operations_1): Call it for GIMPLE_ASM.
21945 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21947         PR target/55718
21948         * config/s390/s390.c (s390_symref_operand_p)
21949         (s390_loadrelative_operand_p): Merge the two functions.
21950         (s390_check_qrst_address, print_operand_address): Add parameters
21951         to s390_loadrelative_operand_p invokation.
21952         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
21953         (s390_reload_larl_operand, s390_secondary_reload): Use
21954         s390_loadrelative_operand_p instead of s390_symref_operand_p.
21955         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
21957 2013-01-09  Mike Stump  <mikestump@comcast.net>
21959         * dse.c (record_store): Remove unnecessary assert.
21961 2013-01-09  Jan Hubicka  <jh@suse.cz>
21963         PR tree-optimization/55569
21964         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
21965         * cfgloop.h (scale_loop_profile): Likewise.
21967 2013-01-09  Jan Hubicka  <jh@suse.cz>
21969         PR lto/45375
21970         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
21971         functions.
21972         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
21974 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
21976         PR middle-end/55114
21977         * expr.h (maybe_emit_group_store): Declare.
21978         * expr.c (maybe_emit_group_store): New function.
21979         * builtins.c (expand_builtin_int_roundingfn): Call it.
21980         (expand_builtin_int_roundingfn_2): Likewise.
21982 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
21984         PR rtl-optimization/55829
21985         * lra-constraints.c (match_reload): Add code for absent output.
21986         (curr_insn_transform): Add code for reloads of matched inputs
21987         without output.
21989 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
21991         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
21992         attribute of movddup insn to DF.
21993         (*vec_interleave_lowv2df): Ditto.
21994         (vec_dupv2df): Ditto.
21996 2013-01-09  Jan Hubicka  <jh@suse.cz>
21998         PR tree-optimiation/55875
21999         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
22000         EVERY_ITERATION parameter.
22001         (number_of_iterations_exit): Check if exit is executed every iteration.
22002         (idx_infer_loop_bounds): Similarly here.
22003         (n_of_executions_at_most): Simplify
22004         to only test for cases where statement is dominated by the
22005         particular bound; handle correctly the "postdominance" test.
22006         (scev_probably_wraps_p): Use max loop iterations info
22007         as a global bound first.
22009 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
22010             Nick Clifton  <nickc@redhat.com>
22012         * config/v850/v850.md (cbranchsf4): New pattern.
22013         (cstoresf4): New pattern.
22014         (cbranchdf4): New pattern.
22015         (cstoredf4): New pattern.
22016         (movsicc): Disallow floating point comparisons.
22017         (cmpsf_le_insn): Fix order of operators.
22018         (cmpsf_lt_insn): Likewise.
22019         (cmpsf_eq_insn): Likewise.
22020         (cmpdf_le_insn): Likewise.
22021         (cmpdf_lt_insn): Likewise.
22022         (cmpdf_eq_insn): Likewise.
22023         (cmpsf_ge_insn): Use LE comparison.
22024         (cmpdf_ge_insn): Likewise.
22025         (cmpsf_gt_insn): Use LT comparison.
22026         (cmpdf_gt_insn): Likewise.
22027         (cmpsf_ne_insn): Delete pattern.
22028         (cmpdf_ne_insn): Delete pattern.
22029         * config/v850/v850.c (v850_gen_float_compare): Use
22030         gen_cmpdf_eq_insn for NE comparison.
22031         (v850_float_z_comparison_operator)
22032         (v850_float_nz_comparison_operator): Move from here ...
22033         * config/v850/predicates.md: ... to here.  Move GT and GE
22034         comparisons into v850_float_z_comparison_operator.
22035         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
22036         Delete prototype.
22037         (v850_float_nz_comparison_operator): Likewise.
22039 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
22041         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
22042         with calls to gen_insvsi/gen_insvdi.
22044 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
22046         * config/i386/i386.c (initial_ix86_tune_features): Set up
22047         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
22049 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
22050             Jakub Jelinek  <jakub@redhat.com>
22052         PR tree-optimization/48189
22053         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
22054         If nitercst is 0, don't predict the exit edge.
22056 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22058         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
22059         in asm_fprintf with reg_names.
22060         (aarch64_print_operand_address): Likewise.
22061         (aarch64_return_addr): Likewise.
22062         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
22064 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
22066         * config/pa/pa.h (VAL_U6_BITS_P): Define.
22067         (INT_U6_BITS): Likewise.
22068         * config/pa/predicates.md (uint6_operand): New predicate.
22069         (shift5_operand, shift6_operand): Likewise.
22070         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
22071         arith32_operand.
22072         (lshrdi3): Use shift6_operand.
22073         (shrpsi4, shrpdi4): New insn patterns.
22074         (extzv): Delete expander.
22075         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
22076         predicates in unamed zero extract patterns.  Tighten common constraint.
22077         (extv): Delete expander.
22078         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
22079         predicates in unamed sign extract patterns.  Tighten common constraint.
22080         (insv): Delete expander.
22081         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
22082         predicates in unamed insert patterns.  Tighten common constraint.
22083         Change uint32_operand predicate to uint6_operand predicate in unamed
22084         DImode pattern to insert constant values of type 1...1xxxx.
22086 2013-01-04  Jan Hubicka  <jh@suse.cz>
22088         PR tree-optimization/55823
22089         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
22090         issue.
22092 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
22093             Uros Bizjak  <ubizjak@gmail.com>
22095         PR rtl-optimization/55845
22096         * df-problems.c (can_move_insns_across): Stop scanning at
22097         volatile_insn_p source instruction or give up if
22098         across_from .. across_to range contains any volatile_insn_p
22099         instructions.
22101 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
22103         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
22104         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
22105         Declare.
22106         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
22107         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
22109 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
22111         PR fortran/55341
22112         * asan.c (asan_clear_shadow): New function.
22113         (asan_emit_stack_protection): Use it.
22115 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
22117         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
22118         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
22119         with tab instead of space.
22121 2013-01-08  Nick Clifton  <nickc@redhat.com>
22123         * config/rl78/rl78.c (rl78_expand_prologue): Always select
22124         register bank 0 at the start of an interrupt handler.
22125         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
22126         MDBH registers.
22128 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
22130         * config/aarch64/aarch64-simd.md
22131         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
22132         (aarch64_simd_bsl): Likewise.
22133         (aarch64_vcond_internal<mode>): Likewise.
22134         (vcond<mode><mode>): Likewise.
22135         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
22136         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
22138 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
22140         * config/aarch64/aarch64-builtins.c
22141         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
22143 2013-01-08  Martin Jambor  <mjambor@suse.cz>
22145         PR debug/55579
22146         * tree-sra.c (analyze_access_subtree): Return true also after
22147         potentially creating a debug-only replacement.
22149 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
22151         PR middle-end/55890
22152         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
22154         PR tree-optimization/54120
22155         * tree-vrp.c (range_fits_type_p): Don't allow
22156         src_precision < precision from signed vr to unsigned_p
22157         if vr->min or vr->max is negative.
22158         (simplify_float_conversion_using_ranges): Test can_float_p
22159         against CODE_FOR_nothing.
22161 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
22162             Richard Biener  <rguenther@suse.de>
22164         PR middle-end/55851
22165         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
22166         types instead of just INTEGER_TYPE types.
22168 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
22170         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
22171         TF_SIZE): Define.
22173 2013-01-07  Steve Ellcey  <sellcey@mips.com>
22175         PR target/42661
22176         * config/mips/mips.opt: Change mad to mmad to match documentation.
22178 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
22180         PR target/55897
22181         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
22182         .progmemx.data now.
22184 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
22186         PR target/55897
22187         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
22188         (avr_addrspace_t): Add .section_name field.
22189         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
22190         array size.
22191         (avr_addrspace): Same.  Initialize .section_name.  Remove last
22192         NULL entry.  Put __memx into .progmemx.data.
22193         (progmem_section_prefix): Remove.
22194         (avr_asm_init_sections): No need to initialize progmem_section.
22195         (avr_asm_named_section): Use avr_addrspace[].section_name to get
22196         section name prefix.
22197         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
22198         retrieve the progmem section.
22199         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
22200         boundary to run over avr_addrspace[].
22201         (avr_register_target_pragmas): Ditto.
22203 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
22205         * varasm.c (output_constant_def_contents): For asan_protect_global
22206         protected strings, adjust DECL_ALIGN if needed, before testing for
22207         anchored symbols.
22208         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
22209         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
22210         normal decls.
22211         (output_object_block): For asan protected decls, emit asan padding
22212         after their contents.
22213         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
22214         (asan_finish_file): Test it here instead.
22216 2013-01-07  Nick Clifton  <nickc@redhat.com>
22217             Matthias Klose  <doko@debian.org>
22218             Doug Kwan  <dougkwan@google.com>
22219             H.J. Lu  <hongjiu.lu@intel.com>
22221         PR driver/55470
22222         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
22224         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
22226         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
22228         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
22230         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
22232 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
22234         PR target/54461
22235         * doc/install.texi (Cross-Compiler-Specific Options): Document
22236         --with-avrlibc.
22238 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
22240         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
22241         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
22242         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
22243         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
22244         vqmovun_high_s64): Fix source operand number and update copyright.
22246 2013-01-07  Richard Biener  <rguenther@suse.de>
22248         PR middle-end/55890
22249         * gimple.h (gimple_call_builtin_p): New overload.
22250         * gimple.c (validate_call): New function.
22251         (gimple_call_builtin_p): Likewise.
22252         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
22253         Use gimple_call_builtin_p.
22254         (find_func_clobbers): Likewise.
22255         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
22256         (strlen_optimize_stmt): Likewise.
22258 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
22260         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
22261         (vld1q_dup_*): Likewise.
22262         (vld1_*): Likewise.
22263         (vld1q_*): Likewise.
22264         (vld1_lane_*): Likewise.
22265         (vld1q_lane_*): Likewise.
22267 2013-01-07  Richard Biener  <rguenther@suse.de>
22269         * lto-streamer.h (LTO_minor_version): Bump to 2.
22271 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
22273         * config/aarch64/aarch64-protos.h
22274         (aarch64_const_double_zero_rtx_p): Rename to...
22275         (aarch64_float_const_zero_rtx_p): ...this.
22276         (aarch64_float_const_representable_p): New.
22277         (aarch64_output_simd_mov_immediate): Likewise.
22278         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
22279         move immediate case.
22280         * config/aarch64/aarch64.c
22281         (aarch64_const_double_zero_rtx_p): Rename to...
22282         (aarch64_float_const_zero_rtx_p): ...this.
22283         (aarch64_print_operand): Allow printing of new constants.
22284         (aarch64_valid_floating_const): New.
22285         (aarch64_legitimate_constant_p): Check for valid floating-point
22286         constants.
22287         (aarch64_simd_valid_immediate): Likewise.
22288         (aarch64_vect_float_const_representable_p): New.
22289         (aarch64_float_const_representable_p): Likewise.
22290         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
22291         (aarch64_output_simd_mov_immediate): New.
22292         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
22293         (*movdf_aarch64): Likewise.
22294         * config/aarch64/constraints.md (Ufc): New.
22295         (Y): call aarch64_float_const_zero_rtx.
22296         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
22298 2013-01-07  Richard Biener  <rguenther@suse.de>
22300         PR tree-optimization/55888
22301         PR tree-optimization/55862
22302         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
22303         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
22304         not if it is contained therein.
22306 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
22308         * config/avr/t-avr: Typo.
22310 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
22312         PR55243
22313         * config/avr/t-avr: Don't automatically rebuild
22314         $(srcdir)/config/avr/t-multilib
22315         $(srcdir)/config/avr/avr-tables.opt
22316         $(srcdir)/doc/avr-mmcu.texi
22317         (avr-mcus): New phony target to build them on request.
22318         (s-avr-mlib, s-avr-mmcu-texi): Remove.
22319         * avr/avr-mcus.def: Adjust comments.
22321 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
22323         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
22325 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
22327         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
22329 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
22331         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
22333 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
22335         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
22336         to generate profiling.
22337         * config/rs6000/aix64.h (LIB_SPEC): Same.
22339 2013-01-04  Andrew Pinski  <apinski@cavium.com>
22341         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
22342         New function.
22343         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
22345 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
22347         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
22348         unconditionally.
22349         (ix86_expand_move): Ditto.
22350         (ix86_zero_extend_to_Pmode): Ditto.
22351         (ix86_expand_call): Ditto.
22352         (ix86_expand_special_args_builtin): Ditto.
22353         (ix86_expand_builtin): Ditto.
22355 2013-01-04  Richard Biener  <rguenther@suse.de>
22357         PR tree-optimization/55862
22358         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
22359         translating them through PHI nodes.
22361 2013-01-04  Martin Jambor  <mjambor@suse.cz>
22363         PR tree-optimization/55755
22364         * tree-sra.c (sra_modify_assign): Do not check that an access has no
22365         children when trying to avoid producing a VIEW_CONVERT_EXPR.
22367 2013-01-04  Marek Polacek  <polacek@redhat.com>
22369         PR middle-end/55859
22370         * opts.c (default_options_optimization): Clarify error message.
22372 2013-01-04  Richard Biener  <rguenther@suse.de>
22374         PR middle-end/55863
22375         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
22376         reassociation.
22378 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
22380         PR target/53789
22381         * config/pa/pa.md (movsi): Revert previous change.
22382         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
22383         references.
22385 2013-01-03  Richard Henderson  <rth@redhat.com>
22387         * config/i386/i386.c (ix86_expand_move): Always assign to op1
22388         after eliminating TLS symbols.
22390 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
22392         PR bootstrap/50167
22393         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
22394         * graphite-poly.c (debug_gmp_value): Likewise.
22396 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
22398         PR target/55712
22399         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
22400         selected code model, define __code_mode_small__, __code_model_medium__,
22401         __code_model_large__, __code_model_32__ or __code_model_kernel__.
22402         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
22403         xchg temporary register with %k.  Declare temporary register as
22404         early clobbered.
22405         [__x86_64__]: For medium and large code models, preserve %rbx register.
22407 2013-01-03  Richard Biener  <rguenther@suse.de>
22409         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
22410         (dump_subscript): Adjust.
22411         (finalize_ddr_dependent): Do not dump redundant info.
22412         (analyze_siv_subscript): Adjust.
22413         (subscript_dependence_tester): Likewise.
22414         (compute_affine_dependence): Likewise.
22416 2013-01-03  Richard Biener  <rguenther@suse.de>
22418         Revert
22419         2013-01-03  Richard Biener  <rguenther@suse.de>
22421         PR tree-optimization/55857
22422         * tree-vect-stmts.c (vectorizable_load): Do not setup
22423         re-alignment for invariant loads.
22425         2013-01-02  Richard Biener  <rguenther@suse.de>
22427         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
22428         invariant load do not generate a vector load from the scalar location.
22430 2013-01-03  Richard Biener  <rguenther@suse.de>
22432         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
22433         for not vectorizing.
22434         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
22435         not build INDIRECT_REFs, call get_name once only.
22436         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
22437         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
22439 2013-01-03  Richard Biener  <rguenther@suse.de>
22441         PR tree-optimization/55857
22442         * tree-vect-stmts.c (vectorizable_load): Do not setup
22443         re-alignment for invariant loads.
22445 2013-01-03  Richard Biener  <rguenther@suse.de>
22447         PR lto/55848
22448         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
22449         prefer a built-in decl.
22451 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
22453         * gcc.c (process_command): Update copyright notice dates.
22454         * gcov.c (print_version): Likewise.
22455         * gcov-dump.c (print_version): Likewise.
22457         PR rtl-optimization/55838
22458         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
22459         iv0.step, iv1.step and step.
22461 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
22462             Marc Glisse  <marc.glisse@inria.fr>
22464         PR tree-optimization/55832
22465         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
22466         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
22467         integer_{one,zero}_node.
22469 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
22471         PR debug/54402
22472         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
22473         * var-tracking.c (reverse_op): Don't add reverse ops to
22474         VALUEs that have already
22475         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
22477 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
22479         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
22481 2013-01-02  Teresa Johnson  <tejohnson@google.com>
22483         * dumpfile.c (dump_loc): Print filename with location.
22484         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
22485         new location_t parameter to emit complete unroll message with
22486         new dump framework.
22487         (canonicalize_loop_induction_variables): Compute loops location
22488         and pass to try_unroll_loop_completely.
22489         * loop-unroll.c (report_unroll_peel): New function.
22490         (peel_loops_completely): Use new dump format with location
22491         for main dumpfile message, and invoke report_unroll_peel on success.
22492         (decide_unrolling_and_peeling): Ditto.
22493         (decide_peel_once_rolling): Remove old dumpfile message subsumed
22494         by report_unroll_peel.
22495         (decide_peel_completely): Ditto.
22496         (decide_unroll_constant_iterations): Ditto.
22497         (decide_unroll_runtime_iterations): Ditto.
22498         (decide_peel_simple): Ditto.
22499         (decide_unroll_stupid): Ditto.
22500         * cfgloop.c (get_loop_location): New function.
22501         * cfgloop.h (get_loop_location): Declare.
22503 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
22505         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
22506         NULL.
22508 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
22510         PR middle-end/55198
22511         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
22512         BLKmode objects when EXPAND_MEMORY is specified.
22514 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
22516         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
22517         in loop predicate.
22518         (fold_builtin_cpu): Do not share cpu model decls across statements.
22520 2013-01-02  Jason Merrill  <jason@redhat.com>
22522         PR c++/55804
22523         * tree.c (build_array_type_1): Revert earlier change.
22525 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
22527         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
22528         "cortex-a57".
22529         * config/aarch64/aarch64-tune.md: Re-generate.
22531 2013-01-02  Richard Biener  <rguenther@suse.de>
22533         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
22534         invariant load do not generate a vector load from the scalar location.
22536 2013-01-02  Richard Biener  <rguenther@suse.de>
22538         PR bootstrap/55784
22539         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
22540         * configure: Regenerate.
22542 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
22544         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
22545         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
22546         (expand_builtin_int_roundingfn_2): Keep the original target around
22547         for the fallback case.
22549 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
22551         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
22552         to be clear for sign changes.
22554 2013-01-01  Jan Hubicka  <jh@suse.cz>
22556         * ipa-inline-analysis.c: Fix formatting.
22558 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
22560         PR tree-optimization/55831
22561         * tree-vect-loop.c (get_initial_def_for_induction): Use
22562         gsi_after_labels instead of gsi_start_bb.
22564 Copyright (C) 2013 Free Software Foundation, Inc.
22566 Copying and distribution of this file, with or without modification,
22567 are permitted in any medium without royalty provided the copyright
22568 notice and this notice are preserved.